File size: 11,789 Bytes
453ed2e
 
49ad6a5
453ed2e
1a833ba
453ed2e
a29e3ba
00f6a78
9ad92f4
453ed2e
9ad92f4
4984c7e
 
d58d62b
 
453ed2e
b31f6c0
96e351a
c000f9c
96e351a
453ed2e
00f6a78
 
a29e3ba
4984c7e
00f6a78
4984c7e
9ad92f4
00f6a78
453ed2e
00f6a78
453ed2e
4984c7e
453ed2e
7391723
4984c7e
 
 
00f6a78
7391723
 
4984c7e
00f6a78
 
9ad92f4
453ed2e
a29e3ba
453ed2e
 
 
 
 
9ad92f4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4984c7e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
86d5e88
 
 
b31f6c0
 
49ad6a5
 
 
 
 
 
 
 
 
a29e3ba
453ed2e
811e3ea
453ed2e
 
01e1199
 
4984c7e
 
 
453ed2e
 
c000f9c
 
453ed2e
b31f6c0
 
 
 
453ed2e
a29e3ba
9ad92f4
a29e3ba
 
4984c7e
7391723
 
a29e3ba
1a833ba
7391723
1a833ba
a29e3ba
453ed2e
 
4984c7e
9ad92f4
 
453ed2e
4984c7e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
b31f6c0
 
 
 
c000f9c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
49ad6a5
4984c7e
c000f9c
453ed2e
 
4984c7e
 
8963f5c
4cdfd9c
94cc393
8963f5c
792633a
4984c7e
453ed2e
 
49ad6a5
 
453ed2e
 
96e351a
4984c7e
 
1d029c0
 
453ed2e
 
9ad92f4
4984c7e
 
 
1a833ba
 
453ed2e
 
4984c7e
96e351a
 
 
 
dc7aed1
fc70300
86d5e88
 
b31f6c0
49ad6a5
 
 
 
 
181da96
ad4d288
1a833ba
49ad6a5
 
 
 
 
 
 
 
1a833ba
453ed2e
86d5e88
 
b31f6c0
49ad6a5
 
 
 
 
 
ad4d288
453ed2e
49ad6a5
 
 
 
 
 
 
 
453ed2e
96e351a
c000f9c
 
 
 
 
 
 
3960c94
453ed2e
 
3960c94
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
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
import torch
import gradio as gr
from gradio import processing_utils, utils
from PIL import Image
import random
from diffusers import (
    DiffusionPipeline,
    AutoencoderKL,
    StableDiffusionControlNetPipeline,
    ControlNetModel,
    StableDiffusionLatentUpscalePipeline,
    StableDiffusionImg2ImgPipeline,
    StableDiffusionControlNetImg2ImgPipeline,
    DPMSolverMultistepScheduler,  # <-- Added import
    EulerDiscreteScheduler  # <-- Added import
)
import time
from share_btn import community_icon_html, loading_icon_html, share_js
import user_history
from illusion_style import css

BASE_MODEL = "SG161222/Realistic_Vision_V5.1_noVAE"

# Initialize both pipelines
vae = AutoencoderKL.from_pretrained("stabilityai/sd-vae-ft-mse", torch_dtype=torch.float16)
#init_pipe = DiffusionPipeline.from_pretrained("SG161222/Realistic_Vision_V5.1_noVAE", torch_dtype=torch.float16)
controlnet = ControlNetModel.from_pretrained("monster-labs/control_v1p_sd15_qrcode_monster", torch_dtype=torch.float16)#, torch_dtype=torch.float16)
main_pipe = StableDiffusionControlNetPipeline.from_pretrained(
    BASE_MODEL,
    controlnet=controlnet,
    vae=vae,
    safety_checker=None,
    torch_dtype=torch.float16,
).to("cuda")

#main_pipe.unet = torch.compile(main_pipe.unet, mode="reduce-overhead", fullgraph=True)
#main_pipe.unet.to(memory_format=torch.channels_last)
#main_pipe.unet = torch.compile(main_pipe.unet, mode="reduce-overhead", fullgraph=True)
#model_id = "stabilityai/sd-x2-latent-upscaler"
image_pipe = StableDiffusionControlNetImg2ImgPipeline(**main_pipe.components)

#image_pipe.unet = torch.compile(image_pipe.unet, mode="reduce-overhead", fullgraph=True)
#upscaler = StableDiffusionLatentUpscalePipeline.from_pretrained(model_id, torch_dtype=torch.float16)
#upscaler.to("cuda")


# Sampler map
SAMPLER_MAP = {
    "DPM++ Karras SDE": lambda config: DPMSolverMultistepScheduler.from_config(config, use_karras=True, algorithm_type="sde-dpmsolver++"),
    "Euler": lambda config: EulerDiscreteScheduler.from_config(config),
}

def center_crop_resize(img, output_size=(512, 512)):
    width, height = img.size

    # Calculate dimensions to crop to the center
    new_dimension = min(width, height)
    left = (width - new_dimension)/2
    top = (height - new_dimension)/2
    right = (width + new_dimension)/2
    bottom = (height + new_dimension)/2

    # Crop and resize
    img = img.crop((left, top, right, bottom))
    img = img.resize(output_size)

    return img

def common_upscale(samples, width, height, upscale_method, crop=False):
        if crop == "center":
            old_width = samples.shape[3]
            old_height = samples.shape[2]
            old_aspect = old_width / old_height
            new_aspect = width / height
            x = 0
            y = 0
            if old_aspect > new_aspect:
                x = round((old_width - old_width * (new_aspect / old_aspect)) / 2)
            elif old_aspect < new_aspect:
                y = round((old_height - old_height * (old_aspect / new_aspect)) / 2)
            s = samples[:,:,y:old_height-y,x:old_width-x]
        else:
            s = samples

        return torch.nn.functional.interpolate(s, size=(height, width), mode=upscale_method)

def upscale(samples, upscale_method, scale_by):
        #s = samples.copy()
        width = round(samples["images"].shape[3] * scale_by)
        height = round(samples["images"].shape[2] * scale_by)
        s = common_upscale(samples["images"], width, height, upscale_method, "disabled")
        return (s)

def check_inputs(prompt: str, control_image: Image.Image):
    if control_image is None:
        raise gr.Error("Please select or upload an Input Illusion")
    if prompt is None or prompt == "":
        raise gr.Error("Prompt is required")

def convert_to_pil(base64_image):
    pil_image = processing_utils.decode_base64_to_image(base64_image)
    return pil_image

def convert_to_base64(pil_image):
    base64_image = processing_utils.encode_pil_to_base64(pil_image)
    return base64_image

# Inference function
def inference(
    control_image: Image.Image,
    prompt: str,
    negative_prompt: str,
    guidance_scale: float = 8.0,
    controlnet_conditioning_scale: float = 1,
    control_guidance_start: float = 1,    
    control_guidance_end: float = 1,
    upscaler_strength: float = 0.5,
    seed: int = -1,
    sampler = "DPM++ Karras SDE",
    progress = gr.Progress(track_tqdm=True),
    profile: gr.OAuthProfile | None = None,
):
    start_time = time.time()
    start_time_struct = time.localtime(start_time)
    start_time_formatted = time.strftime("%H:%M:%S", start_time_struct)
    print(f"Inference started at {start_time_formatted}")
    
    # Generate the initial image
    #init_image = init_pipe(prompt).images[0]

    # Rest of your existing code
    control_image_small = center_crop_resize(control_image)
    control_image_large = center_crop_resize(control_image, (1024, 1024))

    main_pipe.scheduler = SAMPLER_MAP[sampler](main_pipe.scheduler.config)
    my_seed = random.randint(0, 2**32 - 1) if seed == -1 else seed
    generator = torch.Generator(device="cuda").manual_seed(my_seed)
    
    out = main_pipe(
        prompt=prompt,
        negative_prompt=negative_prompt,
        image=control_image_small,
        guidance_scale=float(guidance_scale),
        controlnet_conditioning_scale=float(controlnet_conditioning_scale),
        generator=generator,
        control_guidance_start=float(control_guidance_start),
        control_guidance_end=float(control_guidance_end),
        num_inference_steps=15,
        output_type="latent"
    )
    upscaled_latents = upscale(out, "nearest-exact", 2)
    out_image = image_pipe(
        prompt=prompt,
        negative_prompt=negative_prompt,
        control_image=control_image_large,        
        image=upscaled_latents,
        guidance_scale=float(guidance_scale),
        generator=generator,
        num_inference_steps=20,
        strength=upscaler_strength,
        control_guidance_start=float(control_guidance_start),
        control_guidance_end=float(control_guidance_end),
        controlnet_conditioning_scale=float(controlnet_conditioning_scale)
    )
    end_time = time.time()
    end_time_struct = time.localtime(end_time)
    end_time_formatted = time.strftime("%H:%M:%S", end_time_struct)
    print(f"Inference ended at {end_time_formatted}, taking {end_time-start_time}s")

    # Save image + metadata
    user_history.save_image(
        label=prompt,
        image=out_image["images"][0],
        profile=profile,
        metadata={
            "prompt": prompt,
            "negative_prompt": negative_prompt,
            "guidance_scale": guidance_scale,
            "controlnet_conditioning_scale": controlnet_conditioning_scale,
            "control_guidance_start": control_guidance_start,
            "control_guidance_end": control_guidance_end,
            "upscaler_strength": upscaler_strength,
            "seed": seed,
            "sampler": sampler,
        },
    )

    return out_image["images"][0], gr.update(visible=True), gr.update(visible=True), my_seed
        
with gr.Blocks() as app:
    gr.Markdown(
        '''
        <center><h1>Illusion Diffusion HQ 🌀</h1></span>  
        <span font-size:16px;">Generate stunning high quality illusion artwork with Stable Diffusion</span>  
        </center>
 
        A space by AP [Follow me on Twitter](https://twitter.com/angrypenguinPNG) with big contributions from [multimodalart](https://twitter.com/multimodalart)

        This project works by using [Monster Labs QR Control Net](https://huggingface.co/monster-labs/control_v1p_sd15_qrcode_monster).
        Given a prompt and your pattern, we use a QR code conditioned controlnet to create a stunning illusion! Credit to: [MrUgleh](https://twitter.com/MrUgleh) for discovering the workflow :)
        '''
    )
    state_img_input = gr.State()
    state_img_output = gr.State()
    with gr.Row():
        with gr.Column():
            control_image = gr.Image(label="Input Illusion", type="pil", elem_id="control_image")
            controlnet_conditioning_scale = gr.Slider(minimum=0.0, maximum=5.0, step=0.01, value=0.8, label="Illusion strength", elem_id="illusion_strength", info="ControlNet conditioning scale")
            gr.Examples(examples=["checkers.png", "checkers_mid.jpg", "pattern.png", "ultra_checkers.png", "spiral.jpeg", "funky.jpeg" ], inputs=control_image)
            prompt = gr.Textbox(label="Prompt", elem_id="prompt", info="Type what you want to generate", placeholder="Medieval village scene with busy streets and castle in the distance")
            negative_prompt = gr.Textbox(label="Negative Prompt", info="Type what you don't want to see", value="low quality", elem_id="negative_prompt")
            with gr.Accordion(label="Advanced Options", open=False):
                guidance_scale = gr.Slider(minimum=0.0, maximum=50.0, step=0.25, value=7.5, label="Guidance Scale")
                sampler = gr.Dropdown(choices=list(SAMPLER_MAP.keys()), value="Euler")
                control_start = gr.Slider(minimum=0.0, maximum=1.0, step=0.1, value=0, label="Start of ControlNet")
                control_end = gr.Slider(minimum=0.0, maximum=1.0, step=0.1, value=1, label="End of ControlNet")
                strength = gr.Slider(minimum=0.0, maximum=1.0, step=0.1, value=1, label="Strength of the upscaler")
                seed = gr.Slider(minimum=-1, maximum=9999999999, step=1, value=-1, label="Seed", info="-1 means random seed")
                used_seed = gr.Number(label="Last seed used",interactive=False)
            run_btn = gr.Button("Run")
        with gr.Column():
            result_image = gr.Image(label="Illusion Diffusion Output", interactive=False, elem_id="output")
            with gr.Group(elem_id="share-btn-container", visible=False) as share_group:
                community_icon = gr.HTML(community_icon_html)
                loading_icon = gr.HTML(loading_icon_html)
                share_button = gr.Button("Share to community", elem_id="share-btn")

    prompt.submit(
        check_inputs,
        inputs=[prompt, control_image],
        queue=False
    ).success(
        convert_to_pil,
        inputs=[control_image],
        outputs=[state_img_input],
        queue=False,
        preprocess=False,
    ).success(
        inference,
        inputs=[state_img_input, prompt, negative_prompt, guidance_scale, controlnet_conditioning_scale, control_start, control_end, strength, seed, sampler],
        outputs=[state_img_output, result_image, share_group, used_seed]
    ).success(
        convert_to_base64,
        inputs=[state_img_output],
        outputs=[result_image],
        queue=False,
        postprocess=False
    )
    run_btn.click(
        check_inputs,
        inputs=[prompt, control_image],
        queue=False
    ).success(
        convert_to_pil,
        inputs=[control_image],
        outputs=[state_img_input],
        queue=False,
        preprocess=False,
    ).success(
        inference,
        inputs=[state_img_input, prompt, negative_prompt, guidance_scale, controlnet_conditioning_scale, control_start, control_end, strength, seed, sampler],
        outputs=[state_img_output, result_image, share_group, used_seed]
    ).success(
        convert_to_base64,
        inputs=[state_img_output],
        outputs=[result_image],
        queue=False,
        postprocess=False
    )
    share_button.click(None, [], [], _js=share_js)

with gr.Blocks(css=css) as app_with_history:
    with gr.Tab("Demo"):
        app.render()
    with gr.Tab("Past generations"):
        user_history.render()

app_with_history.queue(max_size=20,api_open=False )

if __name__ == "__main__":
    app_with_history.launch(max_threads=400)