File size: 6,079 Bytes
eb65e9e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
import gradio as gr
import numpy as np
import torch

from video_diffusion.stable_diffusion_video.stable_diffusion_pipeline import StableDiffusionWalkPipeline
from video_diffusion.utils.model_list import stable_model_list


class StableDiffusionText2VideoGenerator:
    def __init__(self):
        self.pipe = None

    def load_model(
        self,
        model_path,
    ):
        if self.pipe is None:
            self.pipe = StableDiffusionWalkPipeline.from_pretrained(
                model_path,
                torch_dtype=torch.float16,
                revision="fp16",
            )

            self.pipe.to("cuda")
            self.pipe.enable_xformers_memory_efficient_attention()
            self.pipe.enable_attention_slicing()
            
        return self.pipe

    def generate_video(
        self,
        model_path: str,
        first_prompts: str,
        second_prompts: str,
        negative_prompt: str,
        num_interpolation_steps: int,
        guidance_scale: int,
        num_inference_step: int,
        height: int,
        width: int,
        upsample: bool,
        fps=int,
    ):
        first_seed = np.random.randint(0, 100000)
        second_seed = np.random.randint(0, 100000)
        seeds = [first_seed, second_seed]
        prompts = [first_prompts, second_prompts]
        pipe = self.load_model(model_path=model_path)

        output_video = pipe.walk(
            prompts=prompts,
            num_interpolation_steps=int(num_interpolation_steps),
            height=height,
            width=width,
            guidance_scale=guidance_scale,
            num_inference_steps=num_inference_step,
            negative_prompt=negative_prompt,
            seeds=seeds,
            upsample=upsample,
            fps=fps,
        )

        return output_video

    def app():
        with gr.Blocks():
            with gr.Row():
                with gr.Column():
                    stable_text2video_first_prompt = gr.Textbox(
                        lines=1,
                        placeholder="First Prompt",
                        show_label=False,
                    )
                    stable_text2video_second_prompt = gr.Textbox(
                        lines=1,
                        placeholder="Second Prompt",
                        show_label=False,
                    )
                    stable_text2video_negative_prompt = gr.Textbox(
                        lines=1,
                        placeholder="Negative Prompt ",
                        show_label=False,
                    )
                    with gr.Row():
                        with gr.Column():
                            stable_text2video_model_path = gr.Dropdown(
                                choices=stable_model_list,
                                label="Stable Model List",
                                value=stable_model_list[0],
                            )
                            stable_text2video_guidance_scale = gr.Slider(
                                minimum=0,
                                maximum=15,
                                step=1,
                                value=8.5,
                                label="Guidance Scale",
                            )
                            stable_text2video_num_inference_steps = gr.Slider(
                                minimum=1,
                                maximum=100,
                                step=1,
                                value=30,
                                label="Number of Inference Steps",
                            )
                            stable_text2video_fps = gr.Slider(
                                minimum=1,
                                maximum=60,
                                step=1,
                                value=10,
                                label="Fps",
                            )
                        with gr.Row():
                            with gr.Column():
                                stable_text2video_num_interpolation_steps = gr.Number(
                                    value=10,
                                    label="Number of Interpolation Steps",
                                )
                                stable_text2video_height = gr.Slider(
                                    minimum=1,
                                    maximum=1000,
                                    step=1,
                                    value=512,
                                    label="Height",
                                )
                                stable_text2video_width = gr.Slider(
                                    minimum=1,
                                    maximum=1000,
                                    step=1,
                                    value=512,
                                    label="Width",
                                )
                                stable_text2video_upsample = gr.Checkbox(
                                    label="Upsample",
                                    default=False,
                                )

                    text2video_generate = gr.Button(value="Generator")

                with gr.Column():
                    text2video_output = gr.Video(label="Output")

            text2video_generate.click(
                fn=StableDiffusionText2VideoGenerator().generate_video,
                inputs=[
                    stable_text2video_model_path,
                    stable_text2video_first_prompt,
                    stable_text2video_second_prompt,
                    stable_text2video_negative_prompt,
                    stable_text2video_num_interpolation_steps,
                    stable_text2video_guidance_scale,
                    stable_text2video_num_inference_steps,
                    stable_text2video_height,
                    stable_text2video_width,
                    stable_text2video_upsample,
                    stable_text2video_fps,
                ],
                outputs=text2video_output,
            )