File size: 3,941 Bytes
56be023
 
37615de
 
4171748
20d0c1e
37615de
 
 
56be023
37615de
1789e44
 
37615de
 
fbb30c1
 
56be023
 
 
 
2d7f410
 
 
 
 
 
 
 
1b52f82
56be023
1b52f82
2d7f410
fbb30c1
6f27020
a024324
f9d0c71
 
56be023
 
 
b7a0604
 
56be023
b7a0604
56be023
 
 
b7a0604
56be023
 
 
 
 
 
b7a0604
6f27020
 
 
 
 
 
 
 
 
 
 
b7a0604
 
 
 
56be023
b7a0604
 
 
 
 
56be023
58c6bee
 
 
 
56be023
 
 
 
58c6bee
56be023
58c6bee
56be023
 
 
 
 
 
 
 
 
 
 
 
58c6bee
e3f6fa0
56be023
e3f6fa0
56be023
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6f27020
56be023
 
6f27020
56be023
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
37615de
 
56be023
 
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
import os

import gradio as gr
import spaces
import torch
from diffusers import AutoPipelineForImage2Image, StableDiffusionInstructPix2PixPipeline
from loguru import logger
from PIL import Image

SUPPORTED_MODELS = [
    "stabilityai/sdxl-turbo",
    "stabilityai/stable-diffusion-3-medium-diffusers",
    "stabilityai/stable-diffusion-xl-refiner-1.0",
    "timbrooks/instruct-pix2pix",
]
DEFAULT_MODEL = "stabilityai/stable-diffusion-xl-refiner-1.0"


model = os.environ.get("MODEL_ID", DEFAULT_MODEL)
gpu_duration = int(os.environ.get("GPU_DURATION", 60))


def load_pipeline(model):
    pipeline_type = (
        StableDiffusionInstructPix2PixPipeline
        if model == "timbrooks/instruct-pix2pix"
        else AutoPipelineForImage2Image
    )

    return pipeline_type.from_pretrained(
        model, torch_dtype=torch.float16, use_safetensors=True, variant="fp16"
    )


logger.debug(f"Loading pipeline: {dict(model=model)}")
pipe = load_pipeline(model).to("cuda")


@logger.catch(reraise=True)
@spaces.GPU(duration=gpu_duration)
def infer(
    prompt: str,
    init_image: Image.Image,
    negative_prompt: str,
    strength: float,
    num_inference_steps: int,
    guidance_scale: float,
    progress=gr.Progress(track_tqdm=True),
):
    logger.info(
        f"Starting image generation: {dict(model=model, prompt=prompt, image=init_image)}"
    )

    # Downscale the image
    init_image.thumbnail((1024, 1024))

    additional_args = {
        k: v
        for k, v in dict(
            strength=strength,
            num_inference_steps=num_inference_steps,
            guidance_scale=guidance_scale,
        ).items()
        if v
    }

    logger.debug(f"Generating image: {dict(prompt=prompt, **additional_args)}")

    images = pipe(
        prompt=prompt,
        image=init_image,
        negative_prompt=negative_prompt,
        **additional_args,
    ).images
    return images[0]


css = """
@media (max-width: 1280px) {
  #images-container {
    flex-direction: column;
  }
}
"""

with gr.Blocks(css=css) as demo:
    with gr.Column():
        gr.Markdown("# Image-to-Image")
        gr.Markdown(f"## Model: `{model}`")

        with gr.Row():
            prompt = gr.Text(
                label="Prompt",
                show_label=False,
                max_lines=1,
                placeholder="Enter your prompt",
                container=False,
            )

            run_button = gr.Button("Run", scale=0, variant="primary")

        with gr.Row(elem_id="images-container"):
            init_image = gr.Image(label="Initial image", type="pil")

            result = gr.Image(label="Result")

        with gr.Accordion("Advanced Settings", open=False):
            negative_prompt = gr.Text(
                label="Negative prompt",
                max_lines=1,
                placeholder="Enter a negative prompt",
            )

            with gr.Row():
                strength = gr.Slider(
                    label="Strength",
                    minimum=0.0,
                    maximum=1.0,
                    step=0.01,
                    value=0.0,
                )

                num_inference_steps = gr.Slider(
                    label="Number of inference steps",
                    minimum=0,
                    maximum=100,
                    step=1,
                    value=0,
                )

                guidance_scale = gr.Slider(
                    label="Guidance scale",
                    minimum=0.0,
                    maximum=100.0,
                    step=0.1,
                    value=0.0,
                )
    gr.on(
        triggers=[run_button.click, prompt.submit],
        fn=infer,
        inputs=[
            prompt,
            init_image,
            negative_prompt,
            strength,
            num_inference_steps,
            guidance_scale,
        ],
        outputs=[result],
    )

if __name__ == "__main__":
    demo.launch()