Datasets:
text stringclasses 20
values |
|---|
User: Generate a video
Prompt: A ballet dancer performing Swan Lake on stage, graceful movements, perfect pirouettes, dramatic lighting, 4K, 60fps, cinematic, slow motion, ultra-detailed costume, emotional performance
Negative prompt: blurry, amateur, low quality, cartoon
Parameters:
- Resolution: 3840x2160
- FPS: 6... |
User: Generate a video
Prompt: A ballet dancer performing Swan Lake on stage, graceful movements, perfect pirouettes, dramatic lighting, 4K, 60fps, cinematic, slow motion, ultra-detailed costume, emotional performance
Negative prompt: blurry, amateur, low quality, cartoon
Parameters:
- Resolution: 3840x2160
- FPS: 6... |
User: Generate an image
Prompt: A majestic snow leopard perched on a rocky cliff at golden hour, ultra-realistic, 8K, sharp focus, shallow depth of field, bokeh background, National Geographic style, photorealistic, hyper-detailed fur, piercing blue eyes, dramatic lighting
Negative prompt: blurry, low quality, cartoo... |
User: Generate an image
Prompt: A delicate cherry blossom tree in full bloom, soft pink petals falling, traditional Japanese garden, koi pond, wooden bridge, spring morning, soft sunlight, serene atmosphere, watercolor painting style, ultra-detailed, 8K, pastel colors
Negative prompt: winter, snow, dark, night, urban... |
User: Generate a video
Prompt: A hummingbird hovering in slow motion, wings beating at 50fps, iridescent feathers, flower background, macro shot, 4K, 120fps, ultra-detailed, National Geographic documentary style, smooth camera movement
Negative prompt: blurry, low fps, shaky camera, cartoon
Parameters:
- Resolution:... |
User: Generate an image
Prompt: A futuristic cyberpunk city at night, neon lights reflecting on wet streets, flying cars, towering skyscrapers, holographic advertisements, rain, cinematic composition, Blade Runner style, 8K, ultra-detailed, atmospheric fog, vibrant colors
Negative prompt: daytime, sunny, low resoluti... |
User: Generate an image
Prompt: A futuristic cyberpunk city at night, neon lights reflecting on wet streets, flying cars, towering skyscrapers, holographic advertisements, rain, cinematic composition, Blade Runner style, 8K, ultra-detailed, atmospheric fog, vibrant colors
Negative prompt: daytime, sunny, low resoluti... |
User: Generate an image
Prompt: A futuristic cyberpunk city at night, neon lights reflecting on wet streets, flying cars, towering skyscrapers, holographic advertisements, rain, cinematic composition, Blade Runner style, 8K, ultra-detailed, atmospheric fog, vibrant colors
Negative prompt: daytime, sunny, low resoluti... |
User: How to set steps for image?
Values: 20, 30, 40, 50, 60, 80, 100
Recommendation: 50 steps for balance, 80+ for max quality
Trade-off: More steps = slower + diminishing returns after 80 |
User: What is CLIP Score?
Description: Measures how well the image matches the text prompt
Good value: > 0.25
Excellent value: > 0.35
Why: Higher is better, measures prompt adherence |
User: Generate an image
Prompt: Portrait of an elderly wise man with a long white beard, deep wrinkles, kind eyes, wearing traditional clothing, sitting in a library with ancient books, warm candlelight, Rembrandt lighting, oil painting style, masterful brushstrokes, 8K, ultra-detailed
Negative prompt: young, smooth ... |
User: Generate an image
Prompt: Portrait of an elderly wise man with a long white beard, deep wrinkles, kind eyes, wearing traditional clothing, sitting in a library with ancient books, warm candlelight, Rembrandt lighting, oil painting style, masterful brushstrokes, 8K, ultra-detailed
Negative prompt: young, smooth ... |
User: How to set duration for video?
Values: 5s, 10s, 15s, 30s, 60s
Recommendation: 5-10s for Sora, longer for specialized models
Trade-off: Longer = harder to maintain coherence |
User: Generate an image
Prompt: A futuristic cyberpunk city at night, neon lights reflecting on wet streets, flying cars, towering skyscrapers, holographic advertisements, rain, cinematic composition, Blade Runner style, 8K, ultra-detailed, atmospheric fog, vibrant colors
Negative prompt: daytime, sunny, low resoluti... |
User: Generate an image
Prompt: A cozy cabin in the woods during autumn, golden leaves falling, smoke rising from chimney, warm light in windows, misty morning, mountain background, photorealistic, 8K, National Geographic, hyper-detailed, cinematic
Negative prompt: summer, green, sunny, urban, modern
Parameters:
- R... |
User: Generate an image
Prompt: A futuristic cyberpunk city at night, neon lights reflecting on wet streets, flying cars, towering skyscrapers, holographic advertisements, rain, cinematic composition, Blade Runner style, 8K, ultra-detailed, atmospheric fog, vibrant colors
Negative prompt: daytime, sunny, low resoluti... |
User: How to set cfg_scale for image?
Values: 1, 3, 5, 7, 9, 12, 15
Recommendation: 7-8 for balance, 3-5 for creativity, 12+ for precision
Trade-off: Higher CFG = more prompt adherence but less creativity |
User: What is IS (Inception Score)?
Description: Measures quality and diversity of generated images
Good value: > 10
Excellent value: > 30
Why: Higher is better, measures variety + quality |
User: Generate an image
Prompt: A majestic snow leopard perched on a rocky cliff at golden hour, ultra-realistic, 8K, sharp focus, shallow depth of field, bokeh background, National Geographic style, photorealistic, hyper-detailed fur, piercing blue eyes, dramatic lighting
Negative prompt: blurry, low quality, cartoo... |
User: Generate an image
Prompt: A cozy cabin in the woods during autumn, golden leaves falling, smoke rising from chimney, warm light in windows, misty morning, mountain background, photorealistic, 8K, National Geographic, hyper-detailed, cinematic
Negative prompt: summer, green, sunny, urban, modern
Parameters:
- R... |
User: What is LPIPS?
Description: Learned Perceptual Image Patch Similarity
Good value: < 0.3
Excellent value: < 0.1
Why: Lower is better, measures perceptual similarity |
User: How to set steps for image?
Values: 20, 30, 40, 50, 60, 80, 100
Recommendation: 50 steps for balance, 80+ for max quality
Trade-off: More steps = slower + diminishing returns after 80 |
User: Generate a video
Prompt: A spaceship traveling through a nebula, stars passing by, colorful gas clouds, cosmic dust, 4K, 60fps, cinematic, Interstellar style, smooth dolly zoom, ultra-detailed
Negative prompt: low resolution, static, cartoon, anime
Parameters:
- Resolution: 3840x2160
- FPS: 60
- Duration: 15s
... |
User: How to set steps for image?
Values: 20, 30, 40, 50, 60, 80, 100
Recommendation: 50 steps for balance, 80+ for max quality
Trade-off: More steps = slower + diminishing returns after 80 |
User: Generate an image
Prompt: A futuristic cyberpunk city at night, neon lights reflecting on wet streets, flying cars, towering skyscrapers, holographic advertisements, rain, cinematic composition, Blade Runner style, 8K, ultra-detailed, atmospheric fog, vibrant colors
Negative prompt: daytime, sunny, low resoluti... |
User: Generate an image
Prompt: A majestic snow leopard perched on a rocky cliff at golden hour, ultra-realistic, 8K, sharp focus, shallow depth of field, bokeh background, National Geographic style, photorealistic, hyper-detailed fur, piercing blue eyes, dramatic lighting
Negative prompt: blurry, low quality, cartoo... |
User: What is CLIP Score?
Description: Measures how well the image matches the text prompt
Good value: > 0.25
Excellent value: > 0.35
Why: Higher is better, measures prompt adherence |
User: Generate an image
Prompt: A cozy cabin in the woods during autumn, golden leaves falling, smoke rising from chimney, warm light in windows, misty morning, mountain background, photorealistic, 8K, National Geographic, hyper-detailed, cinematic
Negative prompt: summer, green, sunny, urban, modern
Parameters:
- R... |
User: Generate an image
Prompt: A delicate cherry blossom tree in full bloom, soft pink petals falling, traditional Japanese garden, koi pond, wooden bridge, spring morning, soft sunlight, serene atmosphere, watercolor painting style, ultra-detailed, 8K, pastel colors
Negative prompt: winter, snow, dark, night, urban... |
User: How to set fps for video?
Values: 24, 30, 60, 120, 240
Recommendation: 30 for normal, 60 for smooth, 120+ for slow motion
Trade-off: Higher fps = smoother but larger file size |
User: Generate an image
Prompt: A majestic snow leopard perched on a rocky cliff at golden hour, ultra-realistic, 8K, sharp focus, shallow depth of field, bokeh background, National Geographic style, photorealistic, hyper-detailed fur, piercing blue eyes, dramatic lighting
Negative prompt: blurry, low quality, cartoo... |
User: Generate an image
Prompt: A delicate cherry blossom tree in full bloom, soft pink petals falling, traditional Japanese garden, koi pond, wooden bridge, spring morning, soft sunlight, serene atmosphere, watercolor painting style, ultra-detailed, 8K, pastel colors
Negative prompt: winter, snow, dark, night, urban... |
User: Generate a video
Prompt: A spaceship traveling through a nebula, stars passing by, colorful gas clouds, cosmic dust, 4K, 60fps, cinematic, Interstellar style, smooth dolly zoom, ultra-detailed
Negative prompt: low resolution, static, cartoon, anime
Parameters:
- Resolution: 3840x2160
- FPS: 60
- Duration: 15s
... |
User: How to set duration for video?
Values: 5s, 10s, 15s, 30s, 60s
Recommendation: 5-10s for Sora, longer for specialized models
Trade-off: Longer = harder to maintain coherence |
User: Generate an image
Prompt: A majestic snow leopard perched on a rocky cliff at golden hour, ultra-realistic, 8K, sharp focus, shallow depth of field, bokeh background, National Geographic style, photorealistic, hyper-detailed fur, piercing blue eyes, dramatic lighting
Negative prompt: blurry, low quality, cartoo... |
User: What is CLIP Score?
Description: Measures how well the image matches the text prompt
Good value: > 0.25
Excellent value: > 0.35
Why: Higher is better, measures prompt adherence |
User: What is FID (Fréchet Inception Distance)?
Description: Measures how similar generated images are to real images
Good value: < 10
Excellent value: < 5
Why: Lower is better, measures distribution similarity |
User: Generate a video
Prompt: A spaceship traveling through a nebula, stars passing by, colorful gas clouds, cosmic dust, 4K, 60fps, cinematic, Interstellar style, smooth dolly zoom, ultra-detailed
Negative prompt: low resolution, static, cartoon, anime
Parameters:
- Resolution: 3840x2160
- FPS: 60
- Duration: 15s
... |
User: How to set duration for video?
Values: 5s, 10s, 15s, 30s, 60s
Recommendation: 5-10s for Sora, longer for specialized models
Trade-off: Longer = harder to maintain coherence |
User: How to set steps for image?
Values: 20, 30, 40, 50, 60, 80, 100
Recommendation: 50 steps for balance, 80+ for max quality
Trade-off: More steps = slower + diminishing returns after 80 |
User: How to set resolution for image?
Values: 512x512, 1024x1024, 2048x2048, 4096x4096, 8192x8192
Recommendation: Use 1024x1024 for SDXL, 2048+ for high quality
Trade-off: Higher resolution = slower generation + more VRAM |
User: Generate a video
Prompt: A hummingbird hovering in slow motion, wings beating at 50fps, iridescent feathers, flower background, macro shot, 4K, 120fps, ultra-detailed, National Geographic documentary style, smooth camera movement
Negative prompt: blurry, low fps, shaky camera, cartoon
Parameters:
- Resolution:... |
User: Generate a video
Prompt: A chef preparing sushi in a traditional Japanese restaurant, precise knife skills, fresh fish, rice, wasabi, close-up shots, 4K, 60fps, food documentary, steam rising, warm lighting
Negative prompt: blurry, fast cuts, cartoon, low quality
Parameters:
- Resolution: 3840x2160
- FPS: 60
-... |
User: How to set duration for video?
Values: 5s, 10s, 15s, 30s, 60s
Recommendation: 5-10s for Sora, longer for specialized models
Trade-off: Longer = harder to maintain coherence |
User: Generate an image
Prompt: A cozy cabin in the woods during autumn, golden leaves falling, smoke rising from chimney, warm light in windows, misty morning, mountain background, photorealistic, 8K, National Geographic, hyper-detailed, cinematic
Negative prompt: summer, green, sunny, urban, modern
Parameters:
- R... |
User: Generate an image
Prompt: A futuristic cyberpunk city at night, neon lights reflecting on wet streets, flying cars, towering skyscrapers, holographic advertisements, rain, cinematic composition, Blade Runner style, 8K, ultra-detailed, atmospheric fog, vibrant colors
Negative prompt: daytime, sunny, low resoluti... |
User: What is CLIP Score?
Description: Measures how well the image matches the text prompt
Good value: > 0.25
Excellent value: > 0.35
Why: Higher is better, measures prompt adherence |
User: What is LPIPS?
Description: Learned Perceptual Image Patch Similarity
Good value: < 0.3
Excellent value: < 0.1
Why: Lower is better, measures perceptual similarity |
User: How to set fps for video?
Values: 24, 30, 60, 120, 240
Recommendation: 30 for normal, 60 for smooth, 120+ for slow motion
Trade-off: Higher fps = smoother but larger file size |
User: What is IS (Inception Score)?
Description: Measures quality and diversity of generated images
Good value: > 10
Excellent value: > 30
Why: Higher is better, measures variety + quality |
User: Generate a video
Prompt: A chef preparing sushi in a traditional Japanese restaurant, precise knife skills, fresh fish, rice, wasabi, close-up shots, 4K, 60fps, food documentary, steam rising, warm lighting
Negative prompt: blurry, fast cuts, cartoon, low quality
Parameters:
- Resolution: 3840x2160
- FPS: 60
-... |
User: Generate a video
Prompt: A ballet dancer performing Swan Lake on stage, graceful movements, perfect pirouettes, dramatic lighting, 4K, 60fps, cinematic, slow motion, ultra-detailed costume, emotional performance
Negative prompt: blurry, amateur, low quality, cartoon
Parameters:
- Resolution: 3840x2160
- FPS: 6... |
User: How to set duration for video?
Values: 5s, 10s, 15s, 30s, 60s
Recommendation: 5-10s for Sora, longer for specialized models
Trade-off: Longer = harder to maintain coherence |
User: Generate a video
Prompt: A spaceship traveling through a nebula, stars passing by, colorful gas clouds, cosmic dust, 4K, 60fps, cinematic, Interstellar style, smooth dolly zoom, ultra-detailed
Negative prompt: low resolution, static, cartoon, anime
Parameters:
- Resolution: 3840x2160
- FPS: 60
- Duration: 15s
... |
User: Generate an image
Prompt: A delicate cherry blossom tree in full bloom, soft pink petals falling, traditional Japanese garden, koi pond, wooden bridge, spring morning, soft sunlight, serene atmosphere, watercolor painting style, ultra-detailed, 8K, pastel colors
Negative prompt: winter, snow, dark, night, urban... |
User: Generate a video
Prompt: A hummingbird hovering in slow motion, wings beating at 50fps, iridescent feathers, flower background, macro shot, 4K, 120fps, ultra-detailed, National Geographic documentary style, smooth camera movement
Negative prompt: blurry, low fps, shaky camera, cartoon
Parameters:
- Resolution:... |
User: How to set steps for image?
Values: 20, 30, 40, 50, 60, 80, 100
Recommendation: 50 steps for balance, 80+ for max quality
Trade-off: More steps = slower + diminishing returns after 80 |
User: How to set resolution for image?
Values: 512x512, 1024x1024, 2048x2048, 4096x4096, 8192x8192
Recommendation: Use 1024x1024 for SDXL, 2048+ for high quality
Trade-off: Higher resolution = slower generation + more VRAM |
User: What is IS (Inception Score)?
Description: Measures quality and diversity of generated images
Good value: > 10
Excellent value: > 30
Why: Higher is better, measures variety + quality |
User: Generate an image
Prompt: Portrait of an elderly wise man with a long white beard, deep wrinkles, kind eyes, wearing traditional clothing, sitting in a library with ancient books, warm candlelight, Rembrandt lighting, oil painting style, masterful brushstrokes, 8K, ultra-detailed
Negative prompt: young, smooth ... |
User: What is IS (Inception Score)?
Description: Measures quality and diversity of generated images
Good value: > 10
Excellent value: > 30
Why: Higher is better, measures variety + quality |
User: Generate a video
Prompt: A hummingbird hovering in slow motion, wings beating at 50fps, iridescent feathers, flower background, macro shot, 4K, 120fps, ultra-detailed, National Geographic documentary style, smooth camera movement
Negative prompt: blurry, low fps, shaky camera, cartoon
Parameters:
- Resolution:... |
User: Generate an image
Prompt: A delicate cherry blossom tree in full bloom, soft pink petals falling, traditional Japanese garden, koi pond, wooden bridge, spring morning, soft sunlight, serene atmosphere, watercolor painting style, ultra-detailed, 8K, pastel colors
Negative prompt: winter, snow, dark, night, urban... |
User: How to set duration for video?
Values: 5s, 10s, 15s, 30s, 60s
Recommendation: 5-10s for Sora, longer for specialized models
Trade-off: Longer = harder to maintain coherence |
User: Generate an image
Prompt: A cozy cabin in the woods during autumn, golden leaves falling, smoke rising from chimney, warm light in windows, misty morning, mountain background, photorealistic, 8K, National Geographic, hyper-detailed, cinematic
Negative prompt: summer, green, sunny, urban, modern
Parameters:
- R... |
User: Generate a video
Prompt: A ballet dancer performing Swan Lake on stage, graceful movements, perfect pirouettes, dramatic lighting, 4K, 60fps, cinematic, slow motion, ultra-detailed costume, emotional performance
Negative prompt: blurry, amateur, low quality, cartoon
Parameters:
- Resolution: 3840x2160
- FPS: 6... |
User: What is FVD (Fréchet Video Distance)?
Description: Video quality metric
Good value: < 500
Excellent value: < 200
Why: Lower is better, measures video quality |
User: What is FVD (Fréchet Video Distance)?
Description: Video quality metric
Good value: < 500
Excellent value: < 200
Why: Lower is better, measures video quality |
User: Generate a video
Prompt: A chef preparing sushi in a traditional Japanese restaurant, precise knife skills, fresh fish, rice, wasabi, close-up shots, 4K, 60fps, food documentary, steam rising, warm lighting
Negative prompt: blurry, fast cuts, cartoon, low quality
Parameters:
- Resolution: 3840x2160
- FPS: 60
-... |
User: How to set cfg_scale for image?
Values: 1, 3, 5, 7, 9, 12, 15
Recommendation: 7-8 for balance, 3-5 for creativity, 12+ for precision
Trade-off: Higher CFG = more prompt adherence but less creativity |
User: Generate a video
Prompt: A ballet dancer performing Swan Lake on stage, graceful movements, perfect pirouettes, dramatic lighting, 4K, 60fps, cinematic, slow motion, ultra-detailed costume, emotional performance
Negative prompt: blurry, amateur, low quality, cartoon
Parameters:
- Resolution: 3840x2160
- FPS: 6... |
User: What is CLIP Score?
Description: Measures how well the image matches the text prompt
Good value: > 0.25
Excellent value: > 0.35
Why: Higher is better, measures prompt adherence |
User: How to set fps for video?
Values: 24, 30, 60, 120, 240
Recommendation: 30 for normal, 60 for smooth, 120+ for slow motion
Trade-off: Higher fps = smoother but larger file size |
User: Generate a video
Prompt: A hummingbird hovering in slow motion, wings beating at 50fps, iridescent feathers, flower background, macro shot, 4K, 120fps, ultra-detailed, National Geographic documentary style, smooth camera movement
Negative prompt: blurry, low fps, shaky camera, cartoon
Parameters:
- Resolution:... |
User: Generate a video
Prompt: A chef preparing sushi in a traditional Japanese restaurant, precise knife skills, fresh fish, rice, wasabi, close-up shots, 4K, 60fps, food documentary, steam rising, warm lighting
Negative prompt: blurry, fast cuts, cartoon, low quality
Parameters:
- Resolution: 3840x2160
- FPS: 60
-... |
User: Generate a video
Prompt: A chef preparing sushi in a traditional Japanese restaurant, precise knife skills, fresh fish, rice, wasabi, close-up shots, 4K, 60fps, food documentary, steam rising, warm lighting
Negative prompt: blurry, fast cuts, cartoon, low quality
Parameters:
- Resolution: 3840x2160
- FPS: 60
-... |
User: What is LPIPS?
Description: Learned Perceptual Image Patch Similarity
Good value: < 0.3
Excellent value: < 0.1
Why: Lower is better, measures perceptual similarity |
User: What is IS (Inception Score)?
Description: Measures quality and diversity of generated images
Good value: > 10
Excellent value: > 30
Why: Higher is better, measures variety + quality |
User: What is CLIP Score?
Description: Measures how well the image matches the text prompt
Good value: > 0.25
Excellent value: > 0.35
Why: Higher is better, measures prompt adherence |
User: How to set resolution for image?
Values: 512x512, 1024x1024, 2048x2048, 4096x4096, 8192x8192
Recommendation: Use 1024x1024 for SDXL, 2048+ for high quality
Trade-off: Higher resolution = slower generation + more VRAM |
User: What is IS (Inception Score)?
Description: Measures quality and diversity of generated images
Good value: > 10
Excellent value: > 30
Why: Higher is better, measures variety + quality |
User: What is LPIPS?
Description: Learned Perceptual Image Patch Similarity
Good value: < 0.3
Excellent value: < 0.1
Why: Lower is better, measures perceptual similarity |
User: Generate a video
Prompt: A drone flying over a tropical island, turquoise water, white sand beach, palm trees, waves crashing, 4K, 60fps, aerial cinematography, smooth gimbal movement, golden hour
Negative prompt: blurry, shaky, low resolution, night
Parameters:
- Resolution: 3840x2160
- FPS: 60
- Duration: 25... |
User: Generate a video
Prompt: A drone flying over a tropical island, turquoise water, white sand beach, palm trees, waves crashing, 4K, 60fps, aerial cinematography, smooth gimbal movement, golden hour
Negative prompt: blurry, shaky, low resolution, night
Parameters:
- Resolution: 3840x2160
- FPS: 60
- Duration: 25... |
User: What is LPIPS?
Description: Learned Perceptual Image Patch Similarity
Good value: < 0.3
Excellent value: < 0.1
Why: Lower is better, measures perceptual similarity |
User: How to set duration for video?
Values: 5s, 10s, 15s, 30s, 60s
Recommendation: 5-10s for Sora, longer for specialized models
Trade-off: Longer = harder to maintain coherence |
User: Generate a video
Prompt: A chef preparing sushi in a traditional Japanese restaurant, precise knife skills, fresh fish, rice, wasabi, close-up shots, 4K, 60fps, food documentary, steam rising, warm lighting
Negative prompt: blurry, fast cuts, cartoon, low quality
Parameters:
- Resolution: 3840x2160
- FPS: 60
-... |
User: How to set duration for video?
Values: 5s, 10s, 15s, 30s, 60s
Recommendation: 5-10s for Sora, longer for specialized models
Trade-off: Longer = harder to maintain coherence |
User: Generate an image
Prompt: Portrait of an elderly wise man with a long white beard, deep wrinkles, kind eyes, wearing traditional clothing, sitting in a library with ancient books, warm candlelight, Rembrandt lighting, oil painting style, masterful brushstrokes, 8K, ultra-detailed
Negative prompt: young, smooth ... |
User: How to set resolution for image?
Values: 512x512, 1024x1024, 2048x2048, 4096x4096, 8192x8192
Recommendation: Use 1024x1024 for SDXL, 2048+ for high quality
Trade-off: Higher resolution = slower generation + more VRAM |
User: Generate a video
Prompt: A ballet dancer performing Swan Lake on stage, graceful movements, perfect pirouettes, dramatic lighting, 4K, 60fps, cinematic, slow motion, ultra-detailed costume, emotional performance
Negative prompt: blurry, amateur, low quality, cartoon
Parameters:
- Resolution: 3840x2160
- FPS: 6... |
User: Generate a video
Prompt: A drone flying over a tropical island, turquoise water, white sand beach, palm trees, waves crashing, 4K, 60fps, aerial cinematography, smooth gimbal movement, golden hour
Negative prompt: blurry, shaky, low resolution, night
Parameters:
- Resolution: 3840x2160
- FPS: 60
- Duration: 25... |
User: What is CLIP Score?
Description: Measures how well the image matches the text prompt
Good value: > 0.25
Excellent value: > 0.35
Why: Higher is better, measures prompt adherence |
User: How to set duration for video?
Values: 5s, 10s, 15s, 30s, 60s
Recommendation: 5-10s for Sora, longer for specialized models
Trade-off: Longer = harder to maintain coherence |
User: Generate an image
Prompt: A futuristic cyberpunk city at night, neon lights reflecting on wet streets, flying cars, towering skyscrapers, holographic advertisements, rain, cinematic composition, Blade Runner style, 8K, ultra-detailed, atmospheric fog, vibrant colors
Negative prompt: daytime, sunny, low resoluti... |
User: Generate a video
Prompt: A drone flying over a tropical island, turquoise water, white sand beach, palm trees, waves crashing, 4K, 60fps, aerial cinematography, smooth gimbal movement, golden hour
Negative prompt: blurry, shaky, low resolution, night
Parameters:
- Resolution: 3840x2160
- FPS: 60
- Duration: 25... |
User: How to set cfg_scale for image?
Values: 1, 3, 5, 7, 9, 12, 15
Recommendation: 7-8 for balance, 3-5 for creativity, 12+ for precision
Trade-off: Higher CFG = more prompt adherence but less creativity |
User: What is FVD (Fréchet Video Distance)?
Description: Video quality metric
Good value: < 500
Excellent value: < 200
Why: Lower is better, measures video quality |
User: Generate a video
Prompt: A chef preparing sushi in a traditional Japanese restaurant, precise knife skills, fresh fish, rice, wasabi, close-up shots, 4K, 60fps, food documentary, steam rising, warm lighting
Negative prompt: blurry, fast cuts, cartoon, low quality
Parameters:
- Resolution: 3840x2160
- FPS: 60
-... |
User: How to set duration for video?
Values: 5s, 10s, 15s, 30s, 60s
Recommendation: 5-10s for Sora, longer for specialized models
Trade-off: Longer = harder to maintain coherence |
End of preview. Expand in Data Studio
Mega Precision Image & Video Dataset
Description
A high-quality dataset for training image and video generation models with ultra-detailed prompts, technical parameters, and quality metrics.
Content
- 10,000,000 examples
- 4 categories: image prompts, video prompts, technical parameters, quality metrics
- English language
Categories
| Category | Description | % |
|---|---|---|
| Image Prompts | Ultra-detailed prompts for image generation | 30% |
| Video Prompts | Ultra-detailed prompts for video generation | 30% |
| Technical Parameters | Resolution, steps, CFG, FPS settings | 20% |
| Quality Metrics | FID, CLIP, IS, LPIPS, FVD | 20% |
Usage
from datasets import load_dataset
dataset = load_dataset("Lelonthecodeur/mega-precision-image-video", data_files="dataset.json")
Created by
Léon (13 years old) - Motif AI Labs
- Downloads last month
- 68