id
stringlengths 7
7
| images
sequencelengths 8
24
| text prompt
stringlengths 0
947
| video link
stringlengths 89
89
| visual quality
int64 1
4
| temporal consistency
int64 1
4
| dynamic degree
int64 1
4
| text-to-video alignment
int64 0
4
| factual consistency
int64 1
4
| conversations
listlengths 2
2
|
---|---|---|---|---|---|---|---|---|---|
6000476 | [
"6000476_00.jpg",
"6000476_01.jpg",
"6000476_02.jpg",
"6000476_03.jpg",
"6000476_04.jpg",
"6000476_05.jpg",
"6000476_06.jpg",
"6000476_07.jpg",
"6000476_08.jpg",
"6000476_09.jpg",
"6000476_10.jpg",
"6000476_11.jpg",
"6000476_12.jpg",
"6000476_13.jpg",
"6000476_14.jpg",
"6000476_15.jpg"
] | An eagle sitting on top of a lush cactus devouring a snake, all framed by a backdrop of mountains. is a companion hybrid of the Valley of Mexico | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/6/6000476.mp4 | 1 | 1 | 3 | 3 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"An eagle sitting on top of a lush cactus devouring a snake, all framed by a backdrop of mountains. is a companion hybrid of the Valley of Mexico\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 1\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 1\n"
}
] |
1002546 | [
"1002546_00.jpg",
"1002546_01.jpg",
"1002546_02.jpg",
"1002546_03.jpg",
"1002546_04.jpg",
"1002546_05.jpg",
"1002546_06.jpg",
"1002546_07.jpg",
"1002546_08.jpg",
"1002546_09.jpg",
"1002546_10.jpg",
"1002546_11.jpg",
"1002546_12.jpg",
"1002546_13.jpg",
"1002546_14.jpg",
"1002546_15.jpg"
] | a carpet of flowers of various colors under the morning sun | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/1/1002546.mp4 | 1 | 1 | 3 | 1 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"a carpet of flowers of various colors under the morning sun\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 1\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 1\n"
}
] |
2000505 | [
"2000505_00.jpg",
"2000505_01.jpg",
"2000505_02.jpg",
"2000505_03.jpg",
"2000505_04.jpg",
"2000505_05.jpg",
"2000505_06.jpg",
"2000505_07.jpg",
"2000505_08.jpg",
"2000505_09.jpg",
"2000505_10.jpg",
"2000505_11.jpg",
"2000505_12.jpg",
"2000505_13.jpg",
"2000505_14.jpg",
"2000505_15.jpg"
] | rayman eating ice cream, globox stretching | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/2/2000505.mp4 | 2 | 2 | 2 | 3 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"rayman eating ice cream, globox stretching\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 2\n temporal consistency: 2\n dynamic degree: 2\n text-to-video alignment: 3\n factual consistency: 1\n"
}
] |
6000767 | [
"6000767_00.jpg",
"6000767_01.jpg",
"6000767_02.jpg",
"6000767_03.jpg",
"6000767_04.jpg",
"6000767_05.jpg",
"6000767_06.jpg",
"6000767_07.jpg",
"6000767_08.jpg",
"6000767_09.jpg",
"6000767_10.jpg",
"6000767_11.jpg",
"6000767_12.jpg",
"6000767_13.jpg",
"6000767_14.jpg",
"6000767_15.jpg"
] | a woman sleeping under a large tree, far shot, cinematic angle, beautiful shot, sunset | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/6/6000767.mp4 | 1 | 3 | 3 | 3 | 2 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"a woman sleeping under a large tree, far shot, cinematic angle, beautiful shot, sunset\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 1\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 2\n"
}
] |
0000966 | [
"0000966_00.jpg",
"0000966_01.jpg",
"0000966_02.jpg",
"0000966_03.jpg",
"0000966_04.jpg",
"0000966_05.jpg",
"0000966_06.jpg",
"0000966_07.jpg",
"0000966_08.jpg",
"0000966_09.jpg",
"0000966_10.jpg",
"0000966_11.jpg",
"0000966_12.jpg",
"0000966_13.jpg",
"0000966_14.jpg",
"0000966_15.jpg",
"0000966_16.jpg",
"0000966_17.jpg",
"0000966_18.jpg",
"0000966_19.jpg",
"0000966_20.jpg",
"0000966_21.jpg",
"0000966_22.jpg",
"0000966_23.jpg"
] | Schematic of fault movement during an earthquake. This video is in 4k quality, like a movie lens with light. | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/0/0000966.mp4 | 3 | 1 | 1 | 3 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Schematic of fault movement during an earthquake. This video is in 4k quality, like a movie lens with light.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 3\n temporal consistency: 1\n dynamic degree: 1\n text-to-video alignment: 3\n factual consistency: 3\n"
}
] |
1002647 | [
"1002647_00.jpg",
"1002647_01.jpg",
"1002647_02.jpg",
"1002647_03.jpg",
"1002647_04.jpg",
"1002647_05.jpg",
"1002647_06.jpg",
"1002647_07.jpg",
"1002647_08.jpg",
"1002647_09.jpg",
"1002647_10.jpg",
"1002647_11.jpg",
"1002647_12.jpg",
"1002647_13.jpg",
"1002647_14.jpg",
"1002647_15.jpg"
] | a Knight still from a retrofuturistic film, Lofi steampunk, cinematic focus, realistic, highly detailed, masterpiece, by albert robida | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/1/1002647.mp4 | 1 | 1 | 3 | 2 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"a Knight still from a retrofuturistic film, Lofi steampunk, cinematic focus, realistic, highly detailed, masterpiece, by albert robida\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 1\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 2\n factual consistency: 1\n"
}
] |
7001235 | [
"7001235_00.jpg",
"7001235_01.jpg",
"7001235_02.jpg",
"7001235_03.jpg",
"7001235_04.jpg",
"7001235_05.jpg",
"7001235_06.jpg",
"7001235_07.jpg"
] | Meera with her family symbolising ties | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/7/7001235.mp4 | 2 | 3 | 3 | 3 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Meera with her family symbolising ties\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 3\n"
}
] |
1001295 | [
"1001295_00.jpg",
"1001295_01.jpg",
"1001295_02.jpg",
"1001295_03.jpg",
"1001295_04.jpg",
"1001295_05.jpg",
"1001295_06.jpg",
"1001295_07.jpg",
"1001295_08.jpg",
"1001295_09.jpg",
"1001295_10.jpg",
"1001295_11.jpg",
"1001295_12.jpg",
"1001295_13.jpg",
"1001295_14.jpg",
"1001295_15.jpg"
] | man playing banjo on porch in summer | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/1/1001295.mp4 | 1 | 1 | 3 | 2 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"man playing banjo on porch in summer\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 1\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 2\n factual consistency: 1\n"
}
] |
6001441 | [
"6001441_00.jpg",
"6001441_01.jpg",
"6001441_02.jpg",
"6001441_03.jpg",
"6001441_04.jpg",
"6001441_05.jpg",
"6001441_06.jpg",
"6001441_07.jpg",
"6001441_08.jpg",
"6001441_09.jpg",
"6001441_10.jpg",
"6001441_11.jpg",
"6001441_12.jpg",
"6001441_13.jpg",
"6001441_14.jpg",
"6001441_15.jpg"
] | Different colored lines flow around resistors | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/6/6001441.mp4 | 1 | 2 | 3 | 3 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Different colored lines flow around resistors\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 1\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 1\n"
}
] |
2000661 | [
"2000661_00.jpg",
"2000661_01.jpg",
"2000661_02.jpg",
"2000661_03.jpg",
"2000661_04.jpg",
"2000661_05.jpg",
"2000661_06.jpg",
"2000661_07.jpg",
"2000661_08.jpg",
"2000661_09.jpg",
"2000661_10.jpg",
"2000661_11.jpg",
"2000661_12.jpg",
"2000661_13.jpg",
"2000661_14.jpg",
"2000661_15.jpg"
] | chirldren book and educational videos, youtube channel logo | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/2/2000661.mp4 | 3 | 3 | 3 | 3 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"chirldren book and educational videos, youtube channel logo\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 3\n"
}
] |
7001546 | [
"7001546_00.jpg",
"7001546_01.jpg",
"7001546_02.jpg",
"7001546_03.jpg",
"7001546_04.jpg",
"7001546_05.jpg",
"7001546_06.jpg",
"7001546_07.jpg"
] | prince say I love to princess | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/7/7001546.mp4 | 3 | 2 | 3 | 2 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"prince say I love to princess\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 3\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 2\n factual consistency: 3\n"
}
] |
2001375 | [
"2001375_00.jpg",
"2001375_01.jpg",
"2001375_02.jpg",
"2001375_03.jpg",
"2001375_04.jpg",
"2001375_05.jpg",
"2001375_06.jpg",
"2001375_07.jpg",
"2001375_08.jpg",
"2001375_09.jpg",
"2001375_10.jpg",
"2001375_11.jpg",
"2001375_12.jpg",
"2001375_13.jpg",
"2001375_14.jpg",
"2001375_15.jpg"
] | Man holding his coffee mug and laughing out loud, realistic video 32k | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/2/2001375.mp4 | 3 | 3 | 3 | 3 | 2 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Man holding his coffee mug and laughing out loud, realistic video 32k\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 2\n"
}
] |
7000897 | [
"7000897_00.jpg",
"7000897_01.jpg",
"7000897_02.jpg",
"7000897_03.jpg",
"7000897_04.jpg",
"7000897_05.jpg",
"7000897_06.jpg",
"7000897_07.jpg"
] | ha scenic forest images with high resolution | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/7/7000897.mp4 | 3 | 3 | 3 | 3 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"ha scenic forest images with high resolution\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 3\n"
}
] |
0000945 | [
"0000945_00.jpg",
"0000945_01.jpg",
"0000945_02.jpg",
"0000945_03.jpg",
"0000945_04.jpg",
"0000945_05.jpg",
"0000945_06.jpg",
"0000945_07.jpg",
"0000945_08.jpg",
"0000945_09.jpg",
"0000945_10.jpg",
"0000945_11.jpg",
"0000945_12.jpg",
"0000945_13.jpg",
"0000945_14.jpg",
"0000945_15.jpg",
"0000945_16.jpg",
"0000945_17.jpg",
"0000945_18.jpg",
"0000945_19.jpg",
"0000945_20.jpg",
"0000945_21.jpg",
"0000945_22.jpg",
"0000945_23.jpg"
] | 3D animation of a soldier shooting at a policeman | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/0/0000945.mp4 | 3 | 2 | 2 | 3 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"3D animation of a soldier shooting at a policeman\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 3\n temporal consistency: 2\n dynamic degree: 2\n text-to-video alignment: 3\n factual consistency: 3\n"
}
] |
6001127 | [
"6001127_00.jpg",
"6001127_01.jpg",
"6001127_02.jpg",
"6001127_03.jpg",
"6001127_04.jpg",
"6001127_05.jpg",
"6001127_06.jpg",
"6001127_07.jpg",
"6001127_08.jpg",
"6001127_09.jpg",
"6001127_10.jpg",
"6001127_11.jpg",
"6001127_12.jpg",
"6001127_13.jpg",
"6001127_14.jpg",
"6001127_15.jpg"
] | tiles of a music band playing over zoom call, cinematic lighting and colors, square tiles for each band member | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/6/6001127.mp4 | 1 | 2 | 3 | 3 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"tiles of a music band playing over zoom call, cinematic lighting and colors, square tiles for each band member\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 1\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 1\n"
}
] |
0000315 | [
"0000315_00.jpg",
"0000315_01.jpg",
"0000315_02.jpg",
"0000315_03.jpg",
"0000315_04.jpg",
"0000315_05.jpg",
"0000315_06.jpg",
"0000315_07.jpg",
"0000315_08.jpg",
"0000315_09.jpg",
"0000315_10.jpg",
"0000315_11.jpg",
"0000315_12.jpg",
"0000315_13.jpg",
"0000315_14.jpg",
"0000315_15.jpg",
"0000315_16.jpg",
"0000315_17.jpg",
"0000315_18.jpg",
"0000315_19.jpg",
"0000315_20.jpg",
"0000315_21.jpg",
"0000315_22.jpg",
"0000315_23.jpg"
] | Two male individuals hugging with their backs turned, looking at the Faculty of Architecture, Design, and Urbanism of Buenos Aires. | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/0/0000315.mp4 | 3 | 3 | 2 | 3 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Two male individuals hugging with their backs turned, looking at the Faculty of Architecture, Design, and Urbanism of Buenos Aires.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 2\n text-to-video alignment: 3\n factual consistency: 3\n"
}
] |
0002617 | [
"0002617_00.jpg",
"0002617_01.jpg",
"0002617_02.jpg",
"0002617_03.jpg",
"0002617_04.jpg",
"0002617_05.jpg",
"0002617_06.jpg",
"0002617_07.jpg",
"0002617_08.jpg",
"0002617_09.jpg",
"0002617_10.jpg",
"0002617_11.jpg",
"0002617_12.jpg",
"0002617_13.jpg",
"0002617_14.jpg",
"0002617_15.jpg",
"0002617_16.jpg",
"0002617_17.jpg",
"0002617_18.jpg",
"0002617_19.jpg",
"0002617_20.jpg",
"0002617_21.jpg",
"0002617_22.jpg",
"0002617_23.jpg"
] | a full scene showing a cloud of smoke come together warping into the face of Lilith, in the artistic style of H.R. Giger | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/0/0002617.mp4 | 3 | 3 | 2 | 3 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"a full scene showing a cloud of smoke come together warping into the face of Lilith, in the artistic style of H.R. Giger\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 2\n text-to-video alignment: 3\n factual consistency: 3\n"
}
] |
7001408 | [
"7001408_00.jpg",
"7001408_01.jpg",
"7001408_02.jpg",
"7001408_03.jpg",
"7001408_04.jpg",
"7001408_05.jpg",
"7001408_06.jpg",
"7001408_07.jpg"
] | electromagnetic blast around a building, top view, lightning, blue electric explosion | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/7/7001408.mp4 | 2 | 3 | 3 | 3 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"electromagnetic blast around a building, top view, lightning, blue electric explosion\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 3\n"
}
] |
6000575 | [
"6000575_00.jpg",
"6000575_01.jpg",
"6000575_02.jpg",
"6000575_03.jpg",
"6000575_04.jpg",
"6000575_05.jpg",
"6000575_06.jpg",
"6000575_07.jpg",
"6000575_08.jpg",
"6000575_09.jpg",
"6000575_10.jpg",
"6000575_11.jpg",
"6000575_12.jpg",
"6000575_13.jpg",
"6000575_14.jpg",
"6000575_15.jpg"
] | noire style. detective in the streets under the rain. looking at another detective walking away. | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/6/6000575.mp4 | 1 | 2 | 3 | 2 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"noire style. detective in the streets under the rain. looking at another detective walking away.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 1\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 2\n factual consistency: 1\n"
}
] |
3002947 | [
"3002947_00.jpg",
"3002947_01.jpg",
"3002947_02.jpg",
"3002947_03.jpg",
"3002947_04.jpg",
"3002947_05.jpg",
"3002947_06.jpg",
"3002947_07.jpg",
"3002947_08.jpg",
"3002947_09.jpg",
"3002947_10.jpg",
"3002947_11.jpg",
"3002947_12.jpg",
"3002947_13.jpg",
"3002947_14.jpg",
"3002947_15.jpg"
] | A man in a dynamic pose is running on an asphalt road under bright sunlight. The world around him is rendered in the pika style with bright and saturated colors. On the horizon, outlines of a city or natural landscape can be seen | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/3/3002947.mp4 | 2 | 1 | 3 | 1 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"A man in a dynamic pose is running on an asphalt road under bright sunlight. The world around him is rendered in the pika style with bright and saturated colors. On the horizon, outlines of a city or natural landscape can be seen\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 2\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 1\n"
}
] |
1001117 | [
"1001117_00.jpg",
"1001117_01.jpg",
"1001117_02.jpg",
"1001117_03.jpg",
"1001117_04.jpg",
"1001117_05.jpg",
"1001117_06.jpg",
"1001117_07.jpg",
"1001117_08.jpg",
"1001117_09.jpg",
"1001117_10.jpg",
"1001117_11.jpg",
"1001117_12.jpg",
"1001117_13.jpg",
"1001117_14.jpg",
"1001117_15.jpg"
] | forest of illusion, based color is purple, use tetradic color harmony, foliages are blown by the wind | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/1/1001117.mp4 | 1 | 2 | 3 | 2 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"forest of illusion, based color is purple, use tetradic color harmony, foliages are blown by the wind\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 1\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 2\n factual consistency: 1\n"
}
] |
6001382 | [
"6001382_00.jpg",
"6001382_01.jpg",
"6001382_02.jpg",
"6001382_03.jpg",
"6001382_04.jpg",
"6001382_05.jpg",
"6001382_06.jpg",
"6001382_07.jpg",
"6001382_08.jpg",
"6001382_09.jpg",
"6001382_10.jpg",
"6001382_11.jpg",
"6001382_12.jpg",
"6001382_13.jpg",
"6001382_14.jpg",
"6001382_15.jpg"
] | Sterile white room with tall life sized mirror reflecting a majestic red mountain | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/6/6001382.mp4 | 1 | 3 | 3 | 3 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Sterile white room with tall life sized mirror reflecting a majestic red mountain\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 1\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 1\n"
}
] |
6000608 | [
"6000608_00.jpg",
"6000608_01.jpg",
"6000608_02.jpg",
"6000608_03.jpg",
"6000608_04.jpg",
"6000608_05.jpg",
"6000608_06.jpg",
"6000608_07.jpg",
"6000608_08.jpg",
"6000608_09.jpg",
"6000608_10.jpg",
"6000608_11.jpg",
"6000608_12.jpg",
"6000608_13.jpg",
"6000608_14.jpg",
"6000608_15.jpg"
] | Rum is poured into a glass | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/6/6000608.mp4 | 3 | 2 | 3 | 3 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Rum is poured into a glass\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 3\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 1\n"
}
] |
0001614 | [
"0001614_00.jpg",
"0001614_01.jpg",
"0001614_02.jpg",
"0001614_03.jpg",
"0001614_04.jpg",
"0001614_05.jpg",
"0001614_06.jpg",
"0001614_07.jpg",
"0001614_08.jpg",
"0001614_09.jpg",
"0001614_10.jpg",
"0001614_11.jpg",
"0001614_12.jpg",
"0001614_13.jpg",
"0001614_14.jpg",
"0001614_15.jpg",
"0001614_16.jpg",
"0001614_17.jpg",
"0001614_18.jpg",
"0001614_19.jpg",
"0001614_20.jpg",
"0001614_21.jpg",
"0001614_22.jpg",
"0001614_23.jpg"
] | man walking to his house in a neighborhood with white houses | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/0/0001614.mp4 | 2 | 2 | 3 | 3 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"man walking to his house in a neighborhood with white houses\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 2\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 3\n"
}
] |
7001821 | [
"7001821_00.jpg",
"7001821_01.jpg",
"7001821_02.jpg",
"7001821_03.jpg",
"7001821_04.jpg",
"7001821_05.jpg",
"7001821_06.jpg",
"7001821_07.jpg"
] | the futuristic city duration video 0:30 | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/7/7001821.mp4 | 2 | 2 | 3 | 3 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"the futuristic city duration video 0:30\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 2\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 3\n"
}
] |
4001166 | [
"4001166_00.jpg",
"4001166_01.jpg",
"4001166_02.jpg",
"4001166_03.jpg",
"4001166_04.jpg",
"4001166_05.jpg",
"4001166_06.jpg",
"4001166_07.jpg",
"4001166_08.jpg",
"4001166_09.jpg",
"4001166_10.jpg",
"4001166_11.jpg",
"4001166_12.jpg",
"4001166_13.jpg",
"4001166_14.jpg",
"4001166_15.jpg"
] | A vast, grass and wildflower covered himalayan landscape stretches out before us. The camera pans down to reveal a young Tibetan boy, Gesar, standing with his mother in the midst of the wilderness, his small nomadic tent from the 11th century behind him. | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/4/4001166.mp4 | 2 | 2 | 1 | 3 | 2 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"A vast, grass and wildflower covered himalayan landscape stretches out before us. The camera pans down to reveal a young Tibetan boy, Gesar, standing with his mother in the midst of the wilderness, his small nomadic tent from the 11th century behind him.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 2\n temporal consistency: 2\n dynamic degree: 1\n text-to-video alignment: 3\n factual consistency: 2\n"
}
] |
2002109 | [
"2002109_00.jpg",
"2002109_01.jpg",
"2002109_02.jpg",
"2002109_03.jpg",
"2002109_04.jpg",
"2002109_05.jpg",
"2002109_06.jpg",
"2002109_07.jpg",
"2002109_08.jpg",
"2002109_09.jpg",
"2002109_10.jpg",
"2002109_11.jpg",
"2002109_12.jpg",
"2002109_13.jpg",
"2002109_14.jpg",
"2002109_15.jpg"
] | rollercoaster is built on the beach | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/2/2002109.mp4 | 2 | 3 | 2 | 3 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"rollercoaster is built on the beach\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 2\n text-to-video alignment: 3\n factual consistency: 3\n"
}
] |
6001129 | [
"6001129_00.jpg",
"6001129_01.jpg",
"6001129_02.jpg",
"6001129_03.jpg",
"6001129_04.jpg",
"6001129_05.jpg",
"6001129_06.jpg",
"6001129_07.jpg",
"6001129_08.jpg",
"6001129_09.jpg",
"6001129_10.jpg",
"6001129_11.jpg",
"6001129_12.jpg",
"6001129_13.jpg",
"6001129_14.jpg",
"6001129_15.jpg"
] | android women in the space station | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/6/6001129.mp4 | 1 | 2 | 3 | 2 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"android women in the space station\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 1\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 2\n factual consistency: 1\n"
}
] |
5000952 | [
"5000952_00.jpg",
"5000952_01.jpg",
"5000952_02.jpg",
"5000952_03.jpg",
"5000952_04.jpg",
"5000952_05.jpg",
"5000952_06.jpg",
"5000952_07.jpg",
"5000952_08.jpg",
"5000952_09.jpg",
"5000952_10.jpg",
"5000952_11.jpg",
"5000952_12.jpg",
"5000952_13.jpg",
"5000952_14.jpg",
"5000952_15.jpg"
] | 3d model of cute fairy dancing in beautiful background | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/5/5000952.mp4 | 3 | 3 | 3 | 3 | 2 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"3d model of cute fairy dancing in beautiful background\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 2\n"
}
] |
1001533 | [
"1001533_00.jpg",
"1001533_01.jpg",
"1001533_02.jpg",
"1001533_03.jpg",
"1001533_04.jpg",
"1001533_05.jpg",
"1001533_06.jpg",
"1001533_07.jpg",
"1001533_08.jpg",
"1001533_09.jpg",
"1001533_10.jpg",
"1001533_11.jpg",
"1001533_12.jpg",
"1001533_13.jpg",
"1001533_14.jpg",
"1001533_15.jpg"
] | wind and star in the sky, | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/1/1001533.mp4 | 1 | 2 | 3 | 2 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"wind and star in the sky,\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 1\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 2\n factual consistency: 1\n"
}
] |
0000633 | [
"0000633_00.jpg",
"0000633_01.jpg",
"0000633_02.jpg",
"0000633_03.jpg",
"0000633_04.jpg",
"0000633_05.jpg",
"0000633_06.jpg",
"0000633_07.jpg",
"0000633_08.jpg",
"0000633_09.jpg",
"0000633_10.jpg",
"0000633_11.jpg",
"0000633_12.jpg",
"0000633_13.jpg",
"0000633_14.jpg",
"0000633_15.jpg",
"0000633_16.jpg",
"0000633_17.jpg",
"0000633_18.jpg",
"0000633_19.jpg",
"0000633_20.jpg",
"0000633_21.jpg",
"0000633_22.jpg",
"0000633_23.jpg"
] | The room was filled with ancient, rotting furniture covered in tattered sheets. A sense of impending doom hung heavy in the air. | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/0/0000633.mp4 | 2 | 3 | 1 | 3 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"The room was filled with ancient, rotting furniture covered in tattered sheets. A sense of impending doom hung heavy in the air.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 1\n text-to-video alignment: 3\n factual consistency: 3\n"
}
] |
2001724 | [
"2001724_00.jpg",
"2001724_01.jpg",
"2001724_02.jpg",
"2001724_03.jpg",
"2001724_04.jpg",
"2001724_05.jpg",
"2001724_06.jpg",
"2001724_07.jpg",
"2001724_08.jpg",
"2001724_09.jpg",
"2001724_10.jpg",
"2001724_11.jpg",
"2001724_12.jpg",
"2001724_13.jpg",
"2001724_14.jpg",
"2001724_15.jpg"
] | a photorealistic movie of a rainy cyberpunk dystopian city at night from street level | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/2/2001724.mp4 | 2 | 3 | 3 | 3 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"a photorealistic movie of a rainy cyberpunk dystopian city at night from street level\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 3\n"
}
] |
5000619 | [
"5000619_00.jpg",
"5000619_01.jpg",
"5000619_02.jpg",
"5000619_03.jpg",
"5000619_04.jpg",
"5000619_05.jpg",
"5000619_06.jpg",
"5000619_07.jpg",
"5000619_08.jpg",
"5000619_09.jpg",
"5000619_10.jpg",
"5000619_11.jpg",
"5000619_12.jpg",
"5000619_13.jpg",
"5000619_14.jpg",
"5000619_15.jpg"
] | The superstar is greeted by his fans and he greets them. | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/5/5000619.mp4 | 3 | 2 | 3 | 3 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"The superstar is greeted by his fans and he greets them.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 3\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 3\n"
}
] |
8000518 | [
"8000518_00.jpg",
"8000518_01.jpg",
"8000518_02.jpg",
"8000518_03.jpg",
"8000518_04.jpg",
"8000518_05.jpg",
"8000518_06.jpg",
"8000518_07.jpg",
"8000518_08.jpg",
"8000518_09.jpg",
"8000518_10.jpg",
"8000518_11.jpg",
"8000518_12.jpg",
"8000518_13.jpg",
"8000518_14.jpg",
"8000518_15.jpg",
"8000518_16.jpg",
"8000518_17.jpg",
"8000518_18.jpg",
"8000518_19.jpg",
"8000518_20.jpg",
"8000518_21.jpg",
"8000518_22.jpg",
"8000518_23.jpg"
] | Show a Viking longship sailing through stormy waves, with thunder and lightning illuminating the sky. | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/8/8000518.mp4 | 1 | 1 | 3 | 2 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Show a Viking longship sailing through stormy waves, with thunder and lightning illuminating the sky.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 1\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 2\n factual consistency: 1\n"
}
] |
2002870 | [
"2002870_00.jpg",
"2002870_01.jpg",
"2002870_02.jpg",
"2002870_03.jpg",
"2002870_04.jpg",
"2002870_05.jpg",
"2002870_06.jpg",
"2002870_07.jpg",
"2002870_08.jpg",
"2002870_09.jpg",
"2002870_10.jpg",
"2002870_11.jpg",
"2002870_12.jpg",
"2002870_13.jpg",
"2002870_14.jpg",
"2002870_15.jpg"
] | It is about realizing that the divine is within you, that you are a part of the infinite cosmos. | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/2/2002870.mp4 | 3 | 3 | 3 | 3 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"It is about realizing that the divine is within you, that you are a part of the infinite cosmos.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 3\n"
}
] |
3001884 | [
"3001884_00.jpg",
"3001884_01.jpg",
"3001884_02.jpg",
"3001884_03.jpg",
"3001884_04.jpg",
"3001884_05.jpg",
"3001884_06.jpg",
"3001884_07.jpg",
"3001884_08.jpg",
"3001884_09.jpg",
"3001884_10.jpg",
"3001884_11.jpg",
"3001884_12.jpg",
"3001884_13.jpg",
"3001884_14.jpg",
"3001884_15.jpg"
] | video of worker nurses in hospitality | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/3/3001884.mp4 | 2 | 3 | 1 | 3 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"video of worker nurses in hospitality\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 1\n text-to-video alignment: 3\n factual consistency: 3\n"
}
] |
3001351 | [
"3001351_00.jpg",
"3001351_01.jpg",
"3001351_02.jpg",
"3001351_03.jpg",
"3001351_04.jpg",
"3001351_05.jpg",
"3001351_06.jpg",
"3001351_07.jpg",
"3001351_08.jpg",
"3001351_09.jpg",
"3001351_10.jpg",
"3001351_11.jpg",
"3001351_12.jpg",
"3001351_13.jpg",
"3001351_14.jpg",
"3001351_15.jpg"
] | a black rainbow in a dark landscape | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/3/3001351.mp4 | 3 | 3 | 3 | 3 | 2 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"a black rainbow in a dark landscape\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 2\n"
}
] |
4000689 | [
"4000689_00.jpg",
"4000689_01.jpg",
"4000689_02.jpg",
"4000689_03.jpg",
"4000689_04.jpg",
"4000689_05.jpg",
"4000689_06.jpg",
"4000689_07.jpg",
"4000689_08.jpg",
"4000689_09.jpg",
"4000689_10.jpg",
"4000689_11.jpg",
"4000689_12.jpg",
"4000689_13.jpg",
"4000689_14.jpg",
"4000689_15.jpg"
] | a 4k canvas of jiraya sensei | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/4/4000689.mp4 | 3 | 3 | 3 | 1 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"a 4k canvas of jiraya sensei\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 3\n"
}
] |
1001516 | [
"1001516_00.jpg",
"1001516_01.jpg",
"1001516_02.jpg",
"1001516_03.jpg",
"1001516_04.jpg",
"1001516_05.jpg",
"1001516_06.jpg",
"1001516_07.jpg",
"1001516_08.jpg",
"1001516_09.jpg",
"1001516_10.jpg",
"1001516_11.jpg",
"1001516_12.jpg",
"1001516_13.jpg",
"1001516_14.jpg",
"1001516_15.jpg"
] | Banker wearing a top hat opening a Bank Vault door to reveal shelves of gold and silver bars | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/1/1001516.mp4 | 1 | 2 | 3 | 2 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Banker wearing a top hat opening a Bank Vault door to reveal shelves of gold and silver bars\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 1\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 2\n factual consistency: 1\n"
}
] |
5000132 | [
"5000132_00.jpg",
"5000132_01.jpg",
"5000132_02.jpg",
"5000132_03.jpg",
"5000132_04.jpg",
"5000132_05.jpg",
"5000132_06.jpg",
"5000132_07.jpg",
"5000132_08.jpg",
"5000132_09.jpg",
"5000132_10.jpg",
"5000132_11.jpg",
"5000132_12.jpg",
"5000132_13.jpg",
"5000132_14.jpg",
"5000132_15.jpg"
] | rain forest Message: SOBUTILNIK NEVER DIE (Font: MODERN) | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/5/5000132.mp4 | 1 | 1 | 3 | 1 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"rain forest Message: SOBUTILNIK NEVER DIE (Font: MODERN)\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 1\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 1\n"
}
] |
7000098 | [
"7000098_00.jpg",
"7000098_01.jpg",
"7000098_02.jpg",
"7000098_03.jpg",
"7000098_04.jpg",
"7000098_05.jpg",
"7000098_06.jpg",
"7000098_07.jpg"
] | Highlight Elephent wisdom and leadership among the herbivores.cartoon | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/7/7000098.mp4 | 3 | 3 | 3 | 2 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Highlight Elephent wisdom and leadership among the herbivores.cartoon\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 2\n factual consistency: 3\n"
}
] |
3001952 | [
"3001952_00.jpg",
"3001952_01.jpg",
"3001952_02.jpg",
"3001952_03.jpg",
"3001952_04.jpg",
"3001952_05.jpg",
"3001952_06.jpg",
"3001952_07.jpg",
"3001952_08.jpg",
"3001952_09.jpg",
"3001952_10.jpg",
"3001952_11.jpg",
"3001952_12.jpg",
"3001952_13.jpg",
"3001952_14.jpg",
"3001952_15.jpg"
] | surprised soldiers in an apocalyptic world seeing something terrible | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/3/3001952.mp4 | 2 | 1 | 3 | 1 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"surprised soldiers in an apocalyptic world seeing something terrible\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 2\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 1\n"
}
] |
6000875 | [
"6000875_00.jpg",
"6000875_01.jpg",
"6000875_02.jpg",
"6000875_03.jpg",
"6000875_04.jpg",
"6000875_05.jpg",
"6000875_06.jpg",
"6000875_07.jpg",
"6000875_08.jpg",
"6000875_09.jpg",
"6000875_10.jpg",
"6000875_11.jpg",
"6000875_12.jpg",
"6000875_13.jpg",
"6000875_14.jpg",
"6000875_15.jpg"
] | Terracotta Army VS the Statue of Liberty | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/6/6000875.mp4 | 1 | 1 | 3 | 3 | 2 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Terracotta Army VS the Statue of Liberty\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 1\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 2\n"
}
] |
8000191 | [
"8000191_00.jpg",
"8000191_01.jpg",
"8000191_02.jpg",
"8000191_03.jpg",
"8000191_04.jpg",
"8000191_05.jpg",
"8000191_06.jpg",
"8000191_07.jpg",
"8000191_08.jpg",
"8000191_09.jpg",
"8000191_10.jpg",
"8000191_11.jpg",
"8000191_12.jpg",
"8000191_13.jpg",
"8000191_14.jpg",
"8000191_15.jpg",
"8000191_16.jpg",
"8000191_17.jpg",
"8000191_18.jpg",
"8000191_19.jpg",
"8000191_20.jpg",
"8000191_21.jpg",
"8000191_22.jpg",
"8000191_23.jpg"
] | cooking restaurant zoom out Message: Hogar Gallego (Font: MODERN) | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/8/8000191.mp4 | 1 | 2 | 3 | 1 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"cooking restaurant zoom out Message: Hogar Gallego (Font: MODERN)\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 1\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 1\n"
}
] |
2001568 | [
"2001568_00.jpg",
"2001568_01.jpg",
"2001568_02.jpg",
"2001568_03.jpg",
"2001568_04.jpg",
"2001568_05.jpg",
"2001568_06.jpg",
"2001568_07.jpg",
"2001568_08.jpg",
"2001568_09.jpg",
"2001568_10.jpg",
"2001568_11.jpg",
"2001568_12.jpg",
"2001568_13.jpg",
"2001568_14.jpg",
"2001568_15.jpg"
] | a caregiver getting vital signs of a patient, realistic | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/2/2001568.mp4 | 3 | 2 | 3 | 2 | 2 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"a caregiver getting vital signs of a patient, realistic\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 3\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 2\n factual consistency: 2\n"
}
] |
4001114 | [
"4001114_00.jpg",
"4001114_01.jpg",
"4001114_02.jpg",
"4001114_03.jpg",
"4001114_04.jpg",
"4001114_05.jpg",
"4001114_06.jpg",
"4001114_07.jpg",
"4001114_08.jpg",
"4001114_09.jpg",
"4001114_10.jpg",
"4001114_11.jpg",
"4001114_12.jpg",
"4001114_13.jpg",
"4001114_14.jpg",
"4001114_15.jpg"
] | a tangerine is approaching from a fairytale winter forest | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/4/4001114.mp4 | 3 | 3 | 3 | 3 | 2 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"a tangerine is approaching from a fairytale winter forest\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 2\n"
}
] |
6001547 | [
"6001547_00.jpg",
"6001547_01.jpg",
"6001547_02.jpg",
"6001547_03.jpg",
"6001547_04.jpg",
"6001547_05.jpg",
"6001547_06.jpg",
"6001547_07.jpg",
"6001547_08.jpg",
"6001547_09.jpg",
"6001547_10.jpg",
"6001547_11.jpg",
"6001547_12.jpg",
"6001547_13.jpg",
"6001547_14.jpg",
"6001547_15.jpg"
] | animated image Introduce Luna, a tiny firefly with magical abilities. | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/6/6001547.mp4 | 1 | 2 | 2 | 1 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"animated image Introduce Luna, a tiny firefly with magical abilities.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 1\n temporal consistency: 2\n dynamic degree: 2\n text-to-video alignment: 1\n factual consistency: 1\n"
}
] |
3001707 | [
"3001707_00.jpg",
"3001707_01.jpg",
"3001707_02.jpg",
"3001707_03.jpg",
"3001707_04.jpg",
"3001707_05.jpg",
"3001707_06.jpg",
"3001707_07.jpg",
"3001707_08.jpg",
"3001707_09.jpg",
"3001707_10.jpg",
"3001707_11.jpg",
"3001707_12.jpg",
"3001707_13.jpg",
"3001707_14.jpg",
"3001707_15.jpg"
] | Clean and transparent sky, clear and bottomless sea, dreamy white clouds, sunset, imaginative space, rich content,tree | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/3/3001707.mp4 | 2 | 3 | 1 | 3 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Clean and transparent sky, clear and bottomless sea, dreamy white clouds, sunset, imaginative space, rich content,tree\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 1\n text-to-video alignment: 3\n factual consistency: 3\n"
}
] |
3001915 | [
"3001915_00.jpg",
"3001915_01.jpg",
"3001915_02.jpg",
"3001915_03.jpg",
"3001915_04.jpg",
"3001915_05.jpg",
"3001915_06.jpg",
"3001915_07.jpg",
"3001915_08.jpg",
"3001915_09.jpg",
"3001915_10.jpg",
"3001915_11.jpg",
"3001915_12.jpg",
"3001915_13.jpg",
"3001915_14.jpg",
"3001915_15.jpg"
] | A french teenager going to dominican republic for vacations | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/3/3001915.mp4 | 2 | 3 | 2 | 3 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"A french teenager going to dominican republic for vacations\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 2\n text-to-video alignment: 3\n factual consistency: 3\n"
}
] |
8000868 | [
"8000868_00.jpg",
"8000868_01.jpg",
"8000868_02.jpg",
"8000868_03.jpg",
"8000868_04.jpg",
"8000868_05.jpg",
"8000868_06.jpg",
"8000868_07.jpg",
"8000868_08.jpg",
"8000868_09.jpg",
"8000868_10.jpg",
"8000868_11.jpg",
"8000868_12.jpg",
"8000868_13.jpg",
"8000868_14.jpg",
"8000868_15.jpg",
"8000868_16.jpg",
"8000868_17.jpg",
"8000868_18.jpg",
"8000868_19.jpg",
"8000868_20.jpg",
"8000868_21.jpg",
"8000868_22.jpg",
"8000868_23.jpg"
] | modern couple walking on old street of village | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/8/8000868.mp4 | 1 | 2 | 3 | 1 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"modern couple walking on old street of village\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 1\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 1\n"
}
] |
5000702 | [
"5000702_00.jpg",
"5000702_01.jpg",
"5000702_02.jpg",
"5000702_03.jpg",
"5000702_04.jpg",
"5000702_05.jpg",
"5000702_06.jpg",
"5000702_07.jpg",
"5000702_08.jpg",
"5000702_09.jpg",
"5000702_10.jpg",
"5000702_11.jpg",
"5000702_12.jpg",
"5000702_13.jpg",
"5000702_14.jpg",
"5000702_15.jpg"
] | A girl eating a hamburger in the 80s, retro style | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/5/5000702.mp4 | 3 | 2 | 3 | 3 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"A girl eating a hamburger in the 80s, retro style\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 3\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 3\n"
}
] |
8000961 | [
"8000961_00.jpg",
"8000961_01.jpg",
"8000961_02.jpg",
"8000961_03.jpg",
"8000961_04.jpg",
"8000961_05.jpg",
"8000961_06.jpg",
"8000961_07.jpg",
"8000961_08.jpg",
"8000961_09.jpg",
"8000961_10.jpg",
"8000961_11.jpg",
"8000961_12.jpg",
"8000961_13.jpg",
"8000961_14.jpg",
"8000961_15.jpg",
"8000961_16.jpg",
"8000961_17.jpg",
"8000961_18.jpg",
"8000961_19.jpg",
"8000961_20.jpg",
"8000961_21.jpg",
"8000961_22.jpg",
"8000961_23.jpg"
] | Mahatma Gandhi: Mahatma Gandhi emerged as a prominent leader against the act. He advocated for nonviolent resistance and civil disobedience, organizing strikes and protests as a form of opposition. | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/8/8000961.mp4 | 1 | 1 | 3 | 2 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Mahatma Gandhi: Mahatma Gandhi emerged as a prominent leader against the act. He advocated for nonviolent resistance and civil disobedience, organizing strikes and protests as a form of opposition.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 1\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 2\n factual consistency: 1\n"
}
] |
2001191 | [
"2001191_00.jpg",
"2001191_01.jpg",
"2001191_02.jpg",
"2001191_03.jpg",
"2001191_04.jpg",
"2001191_05.jpg",
"2001191_06.jpg",
"2001191_07.jpg",
"2001191_08.jpg",
"2001191_09.jpg",
"2001191_10.jpg",
"2001191_11.jpg",
"2001191_12.jpg",
"2001191_13.jpg",
"2001191_14.jpg",
"2001191_15.jpg"
] | Pharaoh Hound: Believed to have been bred by the Pharaohs of ancient Egypt, used for hunting. | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/2/2001191.mp4 | 3 | 3 | 3 | 3 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Pharaoh Hound: Believed to have been bred by the Pharaohs of ancient Egypt, used for hunting.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 3\n"
}
] |
1001903 | [
"1001903_00.jpg",
"1001903_01.jpg",
"1001903_02.jpg",
"1001903_03.jpg",
"1001903_04.jpg",
"1001903_05.jpg",
"1001903_06.jpg",
"1001903_07.jpg",
"1001903_08.jpg",
"1001903_09.jpg",
"1001903_10.jpg",
"1001903_11.jpg",
"1001903_12.jpg",
"1001903_13.jpg",
"1001903_14.jpg",
"1001903_15.jpg"
] | a rockband is playing in a futurist club | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/1/1001903.mp4 | 1 | 1 | 3 | 2 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"a rockband is playing in a futurist club\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 1\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 2\n factual consistency: 1\n"
}
] |
1002849 | [
"1002849_00.jpg",
"1002849_01.jpg",
"1002849_02.jpg",
"1002849_03.jpg",
"1002849_04.jpg",
"1002849_05.jpg",
"1002849_06.jpg",
"1002849_07.jpg",
"1002849_08.jpg",
"1002849_09.jpg",
"1002849_10.jpg",
"1002849_11.jpg",
"1002849_12.jpg",
"1002849_13.jpg",
"1002849_14.jpg",
"1002849_15.jpg"
] | future warrior with armour and machine gun captured by robots and handcuffed | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/1/1002849.mp4 | 1 | 2 | 3 | 2 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"future warrior with armour and machine gun captured by robots and handcuffed\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 1\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 2\n factual consistency: 1\n"
}
] |
4001877 | [
"4001877_00.jpg",
"4001877_01.jpg",
"4001877_02.jpg",
"4001877_03.jpg",
"4001877_04.jpg",
"4001877_05.jpg",
"4001877_06.jpg",
"4001877_07.jpg",
"4001877_08.jpg",
"4001877_09.jpg",
"4001877_10.jpg",
"4001877_11.jpg",
"4001877_12.jpg",
"4001877_13.jpg",
"4001877_14.jpg",
"4001877_15.jpg"
] | a cartoon fox carry school bag | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/4/4001877.mp4 | 3 | 1 | 3 | 2 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"a cartoon fox carry school bag\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 3\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 2\n factual consistency: 3\n"
}
] |
0001165 | [
"0001165_00.jpg",
"0001165_01.jpg",
"0001165_02.jpg",
"0001165_03.jpg",
"0001165_04.jpg",
"0001165_05.jpg",
"0001165_06.jpg",
"0001165_07.jpg",
"0001165_08.jpg",
"0001165_09.jpg",
"0001165_10.jpg",
"0001165_11.jpg",
"0001165_12.jpg",
"0001165_13.jpg",
"0001165_14.jpg",
"0001165_15.jpg",
"0001165_16.jpg",
"0001165_17.jpg",
"0001165_18.jpg",
"0001165_19.jpg",
"0001165_20.jpg",
"0001165_21.jpg",
"0001165_22.jpg",
"0001165_23.jpg"
] | sequoia tree view from a person | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/0/0001165.mp4 | 3 | 2 | 1 | 3 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"sequoia tree view from a person\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 3\n temporal consistency: 2\n dynamic degree: 1\n text-to-video alignment: 3\n factual consistency: 3\n"
}
] |
4001381 | [
"4001381_00.jpg",
"4001381_01.jpg",
"4001381_02.jpg",
"4001381_03.jpg",
"4001381_04.jpg",
"4001381_05.jpg",
"4001381_06.jpg",
"4001381_07.jpg",
"4001381_08.jpg",
"4001381_09.jpg",
"4001381_10.jpg",
"4001381_11.jpg",
"4001381_12.jpg",
"4001381_13.jpg",
"4001381_14.jpg",
"4001381_15.jpg"
] | In Skibide, the tale of Tippy the adventurous toilet and the duo of Carol and Max, the cartoonists, became a legendary story. It reminded everyone that even the most unexpected adventures could bring a community together, and sometimes, a little humor and creativity could make the ordinary extraordinary. | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/4/4001381.mp4 | 2 | 1 | 3 | 2 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"In Skibide, the tale of Tippy the adventurous toilet and the duo of Carol and Max, the cartoonists, became a legendary story. It reminded everyone that even the most unexpected adventures could bring a community together, and sometimes, a little humor and creativity could make the ordinary extraordinary.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 2\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 2\n factual consistency: 3\n"
}
] |
2001341 | [
"2001341_00.jpg",
"2001341_01.jpg",
"2001341_02.jpg",
"2001341_03.jpg",
"2001341_04.jpg",
"2001341_05.jpg",
"2001341_06.jpg",
"2001341_07.jpg",
"2001341_08.jpg",
"2001341_09.jpg",
"2001341_10.jpg",
"2001341_11.jpg",
"2001341_12.jpg",
"2001341_13.jpg",
"2001341_14.jpg",
"2001341_15.jpg"
] | the anime character Rahul is wakes up | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/2/2001341.mp4 | 2 | 3 | 1 | 2 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"the anime character Rahul is wakes up\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 1\n text-to-video alignment: 2\n factual consistency: 3\n"
}
] |
7001120 | [
"7001120_00.jpg",
"7001120_01.jpg",
"7001120_02.jpg",
"7001120_03.jpg",
"7001120_04.jpg",
"7001120_05.jpg",
"7001120_06.jpg",
"7001120_07.jpg"
] | A woman looking into the camera (cinematic, cyberpunk, steam, neon) | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/7/7001120.mp4 | 2 | 3 | 3 | 2 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"A woman looking into the camera (cinematic, cyberpunk, steam, neon)\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 2\n factual consistency: 3\n"
}
] |
7001059 | [
"7001059_00.jpg",
"7001059_01.jpg",
"7001059_02.jpg",
"7001059_03.jpg",
"7001059_04.jpg",
"7001059_05.jpg",
"7001059_06.jpg",
"7001059_07.jpg"
] | insects and mosquitoes splatter on the dashboard of the car that runs fast on the highway, daytime, cinematic style | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/7/7001059.mp4 | 2 | 1 | 3 | 2 | 2 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"insects and mosquitoes splatter on the dashboard of the car that runs fast on the highway, daytime, cinematic style\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 2\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 2\n factual consistency: 2\n"
}
] |
4001433 | [
"4001433_00.jpg",
"4001433_01.jpg",
"4001433_02.jpg",
"4001433_03.jpg",
"4001433_04.jpg",
"4001433_05.jpg",
"4001433_06.jpg",
"4001433_07.jpg",
"4001433_08.jpg",
"4001433_09.jpg",
"4001433_10.jpg",
"4001433_11.jpg",
"4001433_12.jpg",
"4001433_13.jpg",
"4001433_14.jpg",
"4001433_15.jpg"
] | a man and woman warmly embrace | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/4/4001433.mp4 | 2 | 3 | 3 | 3 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"a man and woman warmly embrace\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 3\n"
}
] |
8000851 | [
"8000851_00.jpg",
"8000851_01.jpg",
"8000851_02.jpg",
"8000851_03.jpg",
"8000851_04.jpg",
"8000851_05.jpg",
"8000851_06.jpg",
"8000851_07.jpg",
"8000851_08.jpg",
"8000851_09.jpg",
"8000851_10.jpg",
"8000851_11.jpg",
"8000851_12.jpg",
"8000851_13.jpg",
"8000851_14.jpg",
"8000851_15.jpg",
"8000851_16.jpg",
"8000851_17.jpg",
"8000851_18.jpg",
"8000851_19.jpg",
"8000851_20.jpg",
"8000851_21.jpg",
"8000851_22.jpg",
"8000851_23.jpg"
] | mermaid with blue and green faded hair swimming in the ocean. Shiny blue water. Pirate ship in the background with fog. | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/8/8000851.mp4 | 1 | 3 | 1 | 2 | 2 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"mermaid with blue and green faded hair swimming in the ocean. Shiny blue water. Pirate ship in the background with fog.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 1\n temporal consistency: 3\n dynamic degree: 1\n text-to-video alignment: 2\n factual consistency: 2\n"
}
] |
3002479 | [
"3002479_00.jpg",
"3002479_01.jpg",
"3002479_02.jpg",
"3002479_03.jpg",
"3002479_04.jpg",
"3002479_05.jpg",
"3002479_06.jpg",
"3002479_07.jpg",
"3002479_08.jpg",
"3002479_09.jpg",
"3002479_10.jpg",
"3002479_11.jpg",
"3002479_12.jpg",
"3002479_13.jpg",
"3002479_14.jpg",
"3002479_15.jpg"
] | the greek goddesses on cloud 9 | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/3/3002479.mp4 | 2 | 1 | 3 | 1 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"the greek goddesses on cloud 9\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 2\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 1\n"
}
] |
6001136 | [
"6001136_00.jpg",
"6001136_01.jpg",
"6001136_02.jpg",
"6001136_03.jpg",
"6001136_04.jpg",
"6001136_05.jpg",
"6001136_06.jpg",
"6001136_07.jpg",
"6001136_08.jpg",
"6001136_09.jpg",
"6001136_10.jpg",
"6001136_11.jpg",
"6001136_12.jpg",
"6001136_13.jpg",
"6001136_14.jpg",
"6001136_15.jpg"
] | The scene of celebrating the New Year.New Year‘s Eve reunion dinner, Bright light,photography of high quality, 8K Ultra HD , expressive, Color photography, Cinematic lighting, Cloudy, rainy, sunset, dramatic light, dynamic angle, mesmerizing and dynamic composition, asymmetric shot, photo taken with OLYMPUS OM10, 35mm lens, ISO 800, extreme detail and texture skin,morning glow, volumetric light, shot on Agfa Vista 200, Bokeh, realistic photography, ultra detailed | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/6/6001136.mp4 | 1 | 3 | 3 | 3 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"The scene of celebrating the New Year.New Year‘s Eve reunion dinner, Bright light,photography of high quality, 8K Ultra HD , expressive, Color photography, Cinematic lighting, Cloudy, rainy, sunset, dramatic light, dynamic angle, mesmerizing and dynamic composition, asymmetric shot, photo taken with OLYMPUS OM10, 35mm lens, ISO 800, extreme detail and texture skin,morning glow, volumetric light, shot on Agfa Vista 200, Bokeh, realistic photography, ultra detailed\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 1\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 1\n"
}
] |
2002704 | [
"2002704_00.jpg",
"2002704_01.jpg",
"2002704_02.jpg",
"2002704_03.jpg",
"2002704_04.jpg",
"2002704_05.jpg",
"2002704_06.jpg",
"2002704_07.jpg",
"2002704_08.jpg",
"2002704_09.jpg",
"2002704_10.jpg",
"2002704_11.jpg",
"2002704_12.jpg",
"2002704_13.jpg",
"2002704_14.jpg",
"2002704_15.jpg"
] | She shared tales of her past, of being ensnared by the very fruit that now dictated her existence. | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/2/2002704.mp4 | 3 | 3 | 3 | 3 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"She shared tales of her past, of being ensnared by the very fruit that now dictated her existence.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 3\n"
}
] |
8000084 | [
"8000084_00.jpg",
"8000084_01.jpg",
"8000084_02.jpg",
"8000084_03.jpg",
"8000084_04.jpg",
"8000084_05.jpg",
"8000084_06.jpg",
"8000084_07.jpg",
"8000084_08.jpg",
"8000084_09.jpg",
"8000084_10.jpg",
"8000084_11.jpg",
"8000084_12.jpg",
"8000084_13.jpg",
"8000084_14.jpg",
"8000084_15.jpg",
"8000084_16.jpg",
"8000084_17.jpg",
"8000084_18.jpg",
"8000084_19.jpg",
"8000084_20.jpg",
"8000084_21.jpg",
"8000084_22.jpg",
"8000084_23.jpg"
] | A lovely and gentle snake, painted in a simple but highly textured style | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/8/8000084.mp4 | 1 | 2 | 3 | 2 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"A lovely and gentle snake, painted in a simple but highly textured style\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 1\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 2\n factual consistency: 1\n"
}
] |
5000805 | [
"5000805_00.jpg",
"5000805_01.jpg",
"5000805_02.jpg",
"5000805_03.jpg",
"5000805_04.jpg",
"5000805_05.jpg",
"5000805_06.jpg",
"5000805_07.jpg",
"5000805_08.jpg",
"5000805_09.jpg",
"5000805_10.jpg",
"5000805_11.jpg",
"5000805_12.jpg",
"5000805_13.jpg",
"5000805_14.jpg",
"5000805_15.jpg"
] | convicts go to the stockade in winter, 8K | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/5/5000805.mp4 | 2 | 2 | 3 | 2 | 2 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"convicts go to the stockade in winter, 8K\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 2\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 2\n factual consistency: 2\n"
}
] |
0000850 | [
"0000850_00.jpg",
"0000850_01.jpg",
"0000850_02.jpg",
"0000850_03.jpg",
"0000850_04.jpg",
"0000850_05.jpg",
"0000850_06.jpg",
"0000850_07.jpg",
"0000850_08.jpg",
"0000850_09.jpg",
"0000850_10.jpg",
"0000850_11.jpg",
"0000850_12.jpg",
"0000850_13.jpg",
"0000850_14.jpg",
"0000850_15.jpg",
"0000850_16.jpg",
"0000850_17.jpg",
"0000850_18.jpg",
"0000850_19.jpg",
"0000850_20.jpg",
"0000850_21.jpg",
"0000850_22.jpg",
"0000850_23.jpg"
] | Taj Mahal covered in flooded water, Delhi floods, cinematic picture movie shot 70 mm anamorphic lens | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/0/0000850.mp4 | 3 | 1 | 1 | 3 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Taj Mahal covered in flooded water, Delhi floods, cinematic picture movie shot 70 mm anamorphic lens\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 3\n temporal consistency: 1\n dynamic degree: 1\n text-to-video alignment: 3\n factual consistency: 3\n"
}
] |
1002303 | [
"1002303_00.jpg",
"1002303_01.jpg",
"1002303_02.jpg",
"1002303_03.jpg",
"1002303_04.jpg",
"1002303_05.jpg",
"1002303_06.jpg",
"1002303_07.jpg",
"1002303_08.jpg",
"1002303_09.jpg",
"1002303_10.jpg",
"1002303_11.jpg",
"1002303_12.jpg",
"1002303_13.jpg",
"1002303_14.jpg",
"1002303_15.jpg"
] | Bmw m5 Exterior tyre with cinematic style | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/1/1002303.mp4 | 1 | 2 | 3 | 2 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Bmw m5 Exterior tyre with cinematic style\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 1\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 2\n factual consistency: 1\n"
}
] |
3001282 | [
"3001282_00.jpg",
"3001282_01.jpg",
"3001282_02.jpg",
"3001282_03.jpg",
"3001282_04.jpg",
"3001282_05.jpg",
"3001282_06.jpg",
"3001282_07.jpg",
"3001282_08.jpg",
"3001282_09.jpg",
"3001282_10.jpg",
"3001282_11.jpg",
"3001282_12.jpg",
"3001282_13.jpg",
"3001282_14.jpg",
"3001282_15.jpg"
] | a woman looking down from a window | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/3/3001282.mp4 | 3 | 2 | 3 | 2 | 2 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"a woman looking down from a window\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 3\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 2\n factual consistency: 2\n"
}
] |
2001509 | [
"2001509_00.jpg",
"2001509_01.jpg",
"2001509_02.jpg",
"2001509_03.jpg",
"2001509_04.jpg",
"2001509_05.jpg",
"2001509_06.jpg",
"2001509_07.jpg",
"2001509_08.jpg",
"2001509_09.jpg",
"2001509_10.jpg",
"2001509_11.jpg",
"2001509_12.jpg",
"2001509_13.jpg",
"2001509_14.jpg",
"2001509_15.jpg"
] | animate the boy and the girl | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/2/2001509.mp4 | 3 | 3 | 1 | 3 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"animate the boy and the girl\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 1\n text-to-video alignment: 3\n factual consistency: 1\n"
}
] |
6001936 | [
"6001936_00.jpg",
"6001936_01.jpg",
"6001936_02.jpg",
"6001936_03.jpg",
"6001936_04.jpg",
"6001936_05.jpg",
"6001936_06.jpg",
"6001936_07.jpg",
"6001936_08.jpg",
"6001936_09.jpg",
"6001936_10.jpg",
"6001936_11.jpg",
"6001936_12.jpg",
"6001936_13.jpg",
"6001936_14.jpg",
"6001936_15.jpg"
] | richman driving with smile in the street | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/6/6001936.mp4 | 1 | 2 | 3 | 3 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"richman driving with smile in the street\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 1\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 1\n"
}
] |
0000159 | [
"0000159_00.jpg",
"0000159_01.jpg",
"0000159_02.jpg",
"0000159_03.jpg",
"0000159_04.jpg",
"0000159_05.jpg",
"0000159_06.jpg",
"0000159_07.jpg",
"0000159_08.jpg",
"0000159_09.jpg",
"0000159_10.jpg",
"0000159_11.jpg",
"0000159_12.jpg",
"0000159_13.jpg",
"0000159_14.jpg",
"0000159_15.jpg",
"0000159_16.jpg",
"0000159_17.jpg",
"0000159_18.jpg",
"0000159_19.jpg",
"0000159_20.jpg",
"0000159_21.jpg",
"0000159_22.jpg",
"0000159_23.jpg"
] | A car made out of a lemon driving down the road | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/0/0000159.mp4 | 3 | 2 | 3 | 3 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"A car made out of a lemon driving down the road\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 3\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 3\n"
}
] |
3002234 | [
"3002234_00.jpg",
"3002234_01.jpg",
"3002234_02.jpg",
"3002234_03.jpg",
"3002234_04.jpg",
"3002234_05.jpg",
"3002234_06.jpg",
"3002234_07.jpg",
"3002234_08.jpg",
"3002234_09.jpg",
"3002234_10.jpg",
"3002234_11.jpg",
"3002234_12.jpg",
"3002234_13.jpg",
"3002234_14.jpg",
"3002234_15.jpg"
] | fractal sacred geometric artifact, shimmering and spinning in the light, volumetric lighting, caustic lighting refracting from jewel, ultra realistic, ultimate design, ultra detailed, 4k | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/3/3002234.mp4 | 3 | 3 | 3 | 3 | 2 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"fractal sacred geometric artifact, shimmering and spinning in the light, volumetric lighting, caustic lighting refracting from jewel, ultra realistic, ultimate design, ultra detailed, 4k\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 2\n"
}
] |
4001510 | [
"4001510_00.jpg",
"4001510_01.jpg",
"4001510_02.jpg",
"4001510_03.jpg",
"4001510_04.jpg",
"4001510_05.jpg",
"4001510_06.jpg",
"4001510_07.jpg",
"4001510_08.jpg",
"4001510_09.jpg",
"4001510_10.jpg",
"4001510_11.jpg",
"4001510_12.jpg",
"4001510_13.jpg",
"4001510_14.jpg",
"4001510_15.jpg"
] | The wallpaper peeled away in long, gnarled strips, revealing the rotting wood beneath. | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/4/4001510.mp4 | 2 | 3 | 2 | 1 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"The wallpaper peeled away in long, gnarled strips, revealing the rotting wood beneath.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 2\n text-to-video alignment: 1\n factual consistency: 3\n"
}
] |
6000501 | [
"6000501_00.jpg",
"6000501_01.jpg",
"6000501_02.jpg",
"6000501_03.jpg",
"6000501_04.jpg",
"6000501_05.jpg",
"6000501_06.jpg",
"6000501_07.jpg",
"6000501_08.jpg",
"6000501_09.jpg",
"6000501_10.jpg",
"6000501_11.jpg",
"6000501_12.jpg",
"6000501_13.jpg",
"6000501_14.jpg",
"6000501_15.jpg"
] | spartan shield warrior defending outnumbered valiant hero | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/6/6000501.mp4 | 2 | 2 | 3 | 3 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"spartan shield warrior defending outnumbered valiant hero\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 2\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 1\n"
}
] |
2000341 | [
"2000341_00.jpg",
"2000341_01.jpg",
"2000341_02.jpg",
"2000341_03.jpg",
"2000341_04.jpg",
"2000341_05.jpg",
"2000341_06.jpg",
"2000341_07.jpg",
"2000341_08.jpg",
"2000341_09.jpg",
"2000341_10.jpg",
"2000341_11.jpg",
"2000341_12.jpg",
"2000341_13.jpg",
"2000341_14.jpg",
"2000341_15.jpg"
] | A shoes salesman in a shoestore | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/2/2000341.mp4 | 2 | 2 | 3 | 3 | 2 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"A shoes salesman in a shoestore\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 2\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 2\n"
}
] |
6001846 | [
"6001846_00.jpg",
"6001846_01.jpg",
"6001846_02.jpg",
"6001846_03.jpg",
"6001846_04.jpg",
"6001846_05.jpg",
"6001846_06.jpg",
"6001846_07.jpg",
"6001846_08.jpg",
"6001846_09.jpg",
"6001846_10.jpg",
"6001846_11.jpg",
"6001846_12.jpg",
"6001846_13.jpg",
"6001846_14.jpg",
"6001846_15.jpg"
] | The girl slowly opens the can of cola | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/6/6001846.mp4 | 2 | 1 | 3 | 1 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"The girl slowly opens the can of cola\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 2\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 1\n"
}
] |
4000249 | [
"4000249_00.jpg",
"4000249_01.jpg",
"4000249_02.jpg",
"4000249_03.jpg",
"4000249_04.jpg",
"4000249_05.jpg",
"4000249_06.jpg",
"4000249_07.jpg",
"4000249_08.jpg",
"4000249_09.jpg",
"4000249_10.jpg",
"4000249_11.jpg",
"4000249_12.jpg",
"4000249_13.jpg",
"4000249_14.jpg",
"4000249_15.jpg"
] | a flash bolt from sky is flashing through the screen, cinematic lightining, sunset lightning, shot on a canon d5, | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/4/4000249.mp4 | 2 | 2 | 3 | 3 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"a flash bolt from sky is flashing through the screen, cinematic lightining, sunset lightning, shot on a canon d5,\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 2\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 3\n"
}
] |
8000572 | [
"8000572_00.jpg",
"8000572_01.jpg",
"8000572_02.jpg",
"8000572_03.jpg",
"8000572_04.jpg",
"8000572_05.jpg",
"8000572_06.jpg",
"8000572_07.jpg",
"8000572_08.jpg",
"8000572_09.jpg",
"8000572_10.jpg",
"8000572_11.jpg",
"8000572_12.jpg",
"8000572_13.jpg",
"8000572_14.jpg",
"8000572_15.jpg",
"8000572_16.jpg",
"8000572_17.jpg",
"8000572_18.jpg",
"8000572_19.jpg",
"8000572_20.jpg",
"8000572_21.jpg",
"8000572_22.jpg",
"8000572_23.jpg"
] | a formula 1 car running fast ,3d render, cgi, symetrical, octane render, 35mm, bokeh, 9:16, (intricate details:1. 12), hdr, (intricate details, hyperdetailed:1. 15), (natural skin texture, hyperrealism, soft light, sharp:1 | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/8/8000572.mp4 | 1 | 1 | 3 | 2 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"a formula 1 car running fast ,3d render, cgi, symetrical, octane render, 35mm, bokeh, 9:16, (intricate details:1. 12), hdr, (intricate details, hyperdetailed:1. 15), (natural skin texture, hyperrealism, soft light, sharp:1\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 1\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 2\n factual consistency: 1\n"
}
] |
2002098 | [
"2002098_00.jpg",
"2002098_01.jpg",
"2002098_02.jpg",
"2002098_03.jpg",
"2002098_04.jpg",
"2002098_05.jpg",
"2002098_06.jpg",
"2002098_07.jpg",
"2002098_08.jpg",
"2002098_09.jpg",
"2002098_10.jpg",
"2002098_11.jpg",
"2002098_12.jpg",
"2002098_13.jpg",
"2002098_14.jpg",
"2002098_15.jpg"
] | humsn lung seen in a whole big picture | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/2/2002098.mp4 | 3 | 3 | 2 | 1 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"humsn lung seen in a whole big picture\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 2\n text-to-video alignment: 1\n factual consistency: 1\n"
}
] |
6001618 | [
"6001618_00.jpg",
"6001618_01.jpg",
"6001618_02.jpg",
"6001618_03.jpg",
"6001618_04.jpg",
"6001618_05.jpg",
"6001618_06.jpg",
"6001618_07.jpg",
"6001618_08.jpg",
"6001618_09.jpg",
"6001618_10.jpg",
"6001618_11.jpg",
"6001618_12.jpg",
"6001618_13.jpg",
"6001618_14.jpg",
"6001618_15.jpg"
] | a man anxiously awaiting his date . Standing on a front porch of a suburban home on a bright and sunny day | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/6/6001618.mp4 | 1 | 1 | 3 | 3 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"a man anxiously awaiting his date . Standing on a front porch of a suburban home on a bright and sunny day\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 1\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 1\n"
}
] |
3001460 | [
"3001460_00.jpg",
"3001460_01.jpg",
"3001460_02.jpg",
"3001460_03.jpg",
"3001460_04.jpg",
"3001460_05.jpg",
"3001460_06.jpg",
"3001460_07.jpg",
"3001460_08.jpg",
"3001460_09.jpg",
"3001460_10.jpg",
"3001460_11.jpg",
"3001460_12.jpg",
"3001460_13.jpg",
"3001460_14.jpg",
"3001460_15.jpg"
] | A labrador wearing cowboy attire, riding into the sunset, 14:3. | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/3/3001460.mp4 | 2 | 2 | 3 | 2 | 2 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"A labrador wearing cowboy attire, riding into the sunset, 14:3.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 2\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 2\n factual consistency: 2\n"
}
] |
2002939 | [
"2002939_00.jpg",
"2002939_01.jpg",
"2002939_02.jpg",
"2002939_03.jpg",
"2002939_04.jpg",
"2002939_05.jpg",
"2002939_06.jpg",
"2002939_07.jpg",
"2002939_08.jpg",
"2002939_09.jpg",
"2002939_10.jpg",
"2002939_11.jpg",
"2002939_12.jpg",
"2002939_13.jpg",
"2002939_14.jpg",
"2002939_15.jpg"
] | stacked rolls of denim in a warehouse on a shelf Message: OVRSTITCH (Font: MODERN) | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/2/2002939.mp4 | 3 | 3 | 3 | 3 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"stacked rolls of denim in a warehouse on a shelf Message: OVRSTITCH (Font: MODERN)\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 3\n"
}
] |
2001525 | [
"2001525_00.jpg",
"2001525_01.jpg",
"2001525_02.jpg",
"2001525_03.jpg",
"2001525_04.jpg",
"2001525_05.jpg",
"2001525_06.jpg",
"2001525_07.jpg",
"2001525_08.jpg",
"2001525_09.jpg",
"2001525_10.jpg",
"2001525_11.jpg",
"2001525_12.jpg",
"2001525_13.jpg",
"2001525_14.jpg",
"2001525_15.jpg"
] | A tennogen military mecha suit tapestry of emotions, woven by notes | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/2/2001525.mp4 | 3 | 3 | 1 | 3 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"A tennogen military mecha suit tapestry of emotions, woven by notes\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 1\n text-to-video alignment: 3\n factual consistency: 3\n"
}
] |
6001950 | [
"6001950_00.jpg",
"6001950_01.jpg",
"6001950_02.jpg",
"6001950_03.jpg",
"6001950_04.jpg",
"6001950_05.jpg",
"6001950_06.jpg",
"6001950_07.jpg",
"6001950_08.jpg",
"6001950_09.jpg",
"6001950_10.jpg",
"6001950_11.jpg",
"6001950_12.jpg",
"6001950_13.jpg",
"6001950_14.jpg",
"6001950_15.jpg"
] | sink in the boat of love | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/6/6001950.mp4 | 1 | 2 | 3 | 1 | 1 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"sink in the boat of love\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 1\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 1\n"
}
] |
2001955 | [
"2001955_00.jpg",
"2001955_01.jpg",
"2001955_02.jpg",
"2001955_03.jpg",
"2001955_04.jpg",
"2001955_05.jpg",
"2001955_06.jpg",
"2001955_07.jpg",
"2001955_08.jpg",
"2001955_09.jpg",
"2001955_10.jpg",
"2001955_11.jpg",
"2001955_12.jpg",
"2001955_13.jpg",
"2001955_14.jpg",
"2001955_15.jpg"
] | An old woman seeking medical treatment everywhere | https://huggingface.co/datasets/hexuan21/VideoFeedback-videos-mp4/blob/main/2/2001955.mp4 | 2 | 3 | 3 | 3 | 3 | [
{
"from": "human",
"value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"An old woman seeking medical treatment everywhere\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> "
},
{
"from": "gpt",
"value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 3\n factual consistency: 3\n"
}
] |
End of preview. Expand
in Dataset Viewer.
README.md exists but content is empty.
- Downloads last month
- 47