Spaces:
Running
on
Zero
Running
on
Zero
code optimization
Browse filesfree up cuda chache + gc + load models on cpu only when needed and unload after used + saving up on zerogpu duration
app.py
CHANGED
@@ -3,8 +3,9 @@ from diffusers import StableDiffusion3Pipeline, StableDiffusionPipeline, StableD
|
|
3 |
import gradio as gr
|
4 |
import os
|
5 |
import random
|
6 |
-
import numpy
|
7 |
from PIL import Image
|
|
|
8 |
import spaces
|
9 |
|
10 |
HF_TOKEN = os.getenv("HF_TOKEN") # login with hf read token to access sd gated models
|
@@ -17,61 +18,65 @@ else:
|
|
17 |
print("Using CPU")
|
18 |
|
19 |
|
20 |
-
MAX_SEED =
|
21 |
|
22 |
-
#
|
|
|
23 |
|
24 |
-
# sd3 medium
|
25 |
-
sd3_medium_pipe = StableDiffusion3Pipeline.from_pretrained(
|
26 |
-
"stabilityai/stable-diffusion-3-medium-diffusers", torch_dtype=torch.float16
|
27 |
-
)
|
28 |
-
sd3_medium_pipe.enable_model_cpu_offload()
|
29 |
-
|
30 |
-
# sd 2.1
|
31 |
-
sd2_1_pipe = StableDiffusionPipeline.from_pretrained(
|
32 |
-
"stabilityai/stable-diffusion-2-1", torch_dtype=torch.float16
|
33 |
-
)
|
34 |
-
sd2_1_pipe.enable_model_cpu_offload()
|
35 |
-
|
36 |
-
# sdxl
|
37 |
-
sdxl_pipe = StableDiffusionXLPipeline.from_pretrained(
|
38 |
-
"stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16
|
39 |
-
)
|
40 |
-
sdxl_pipe.enable_model_cpu_offload()
|
41 |
|
42 |
-
|
43 |
-
|
44 |
-
|
45 |
-
|
46 |
-
|
47 |
-
|
48 |
-
|
49 |
-
|
50 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
51 |
|
52 |
-
|
53 |
-
stable_cascade_prior_pipe = StableCascadePriorPipeline.from_pretrained(
|
54 |
-
"stabilityai/stable-cascade-prior", variant="bf16", torch_dtype=torch.bfloat16
|
55 |
-
)
|
56 |
-
stable_cascade_prior_pipe.enable_model_cpu_offload()
|
57 |
-
stable_cascade_decoder_pipe = StableCascadeDecoderPipeline.from_pretrained(
|
58 |
-
"stabilityai/stable-cascade", variant="bf16", torch_dtype=torch.float16
|
59 |
-
)
|
60 |
-
stable_cascade_decoder_pipe.enable_model_cpu_offload()
|
61 |
|
62 |
-
# sd 1.5
|
63 |
-
sd1_5_pipe = StableDiffusionPipeline.from_pretrained(
|
64 |
-
"runwayml/stable-diffusion-v1-5", torch_dtype=torch.float16
|
65 |
-
)
|
66 |
-
sd1_5_pipe.enable_model_cpu_offload()
|
67 |
|
68 |
-
|
69 |
-
|
|
|
|
|
|
|
|
|
|
|
70 |
|
71 |
-
# Helper function to generate images for a single model
|
72 |
@spaces.GPU(duration=80)
|
73 |
-
def
|
74 |
prompt,
|
|
|
75 |
negative_prompt,
|
76 |
num_inference_steps,
|
77 |
guidance_scale,
|
@@ -79,71 +84,114 @@ def generate_single_image(
|
|
79 |
width,
|
80 |
seed,
|
81 |
num_images_per_prompt,
|
82 |
-
model_choice,
|
83 |
-
generator,
|
84 |
prior_num_inference_steps=None,
|
85 |
prior_guidance_scale=None,
|
86 |
decoder_num_inference_steps=None,
|
87 |
decoder_guidance_scale=None,
|
88 |
):
|
89 |
-
|
90 |
-
|
91 |
-
|
92 |
-
|
93 |
-
pipe
|
94 |
-
|
95 |
-
|
96 |
-
|
97 |
-
|
98 |
-
elif model_choice == "stable cascade":
|
99 |
-
pipe = stable_cascade_prior_pipe
|
100 |
-
elif model_choice == "sd1.5":
|
101 |
-
pipe = sd1_5_pipe
|
102 |
-
else:
|
103 |
-
raise ValueError(f"Invalid model choice: {model_choice}")
|
104 |
|
105 |
-
#
|
106 |
-
if model_choice
|
107 |
-
|
108 |
-
|
109 |
-
|
110 |
-
|
111 |
-
|
112 |
-
|
113 |
-
width=width,
|
114 |
-
generator=generator,
|
115 |
-
num_images_per_prompt=num_images_per_prompt,
|
116 |
)
|
117 |
|
118 |
-
|
119 |
-
|
120 |
-
|
121 |
-
|
122 |
-
num_inference_steps=decoder_num_inference_steps,
|
123 |
-
guidance_scale=decoder_guidance_scale,
|
124 |
-
).images
|
125 |
|
126 |
-
# the rest of the models have similar pipeline
|
127 |
-
else:
|
128 |
-
output = pipe(
|
129 |
-
prompt=prompt,
|
130 |
-
negative_prompt=negative_prompt,
|
131 |
-
num_inference_steps=num_inference_steps,
|
132 |
-
guidance_scale=guidance_scale,
|
133 |
-
height=height,
|
134 |
-
width=width,
|
135 |
-
generator=generator,
|
136 |
-
num_images_per_prompt=num_images_per_prompt,
|
137 |
-
).images
|
138 |
|
139 |
-
|
140 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
141 |
|
142 |
return output
|
143 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
144 |
|
145 |
# Define the image generation function for the Arena tab
|
146 |
-
@spaces.GPU(duration=240)
|
147 |
def generate_arena_images(
|
148 |
prompt,
|
149 |
negative_prompt,
|
@@ -188,15 +236,12 @@ def generate_arena_images(
|
|
188 |
decoder_guidance_scale_d,
|
189 |
progress=gr.Progress(track_tqdm=True),
|
190 |
):
|
191 |
-
if seed == 0:
|
192 |
-
seed = random.randint(1, MAX_SEED)
|
193 |
-
|
194 |
-
generator = torch.Generator().manual_seed(seed)
|
195 |
|
196 |
# Generate images for selected models
|
197 |
if num_models_to_compare >= 2:
|
198 |
images_a = generate_single_image(
|
199 |
prompt,
|
|
|
200 |
negative_prompt,
|
201 |
num_inference_steps_a,
|
202 |
guidance_scale_a,
|
@@ -204,8 +249,6 @@ def generate_arena_images(
|
|
204 |
width_a,
|
205 |
seed,
|
206 |
num_images_per_prompt,
|
207 |
-
model_choice_a,
|
208 |
-
generator,
|
209 |
prior_num_inference_steps_a,
|
210 |
prior_guidance_scale_a,
|
211 |
decoder_num_inference_steps_a,
|
@@ -213,6 +256,7 @@ def generate_arena_images(
|
|
213 |
)
|
214 |
images_b = generate_single_image(
|
215 |
prompt,
|
|
|
216 |
negative_prompt,
|
217 |
num_inference_steps_b,
|
218 |
guidance_scale_b,
|
@@ -220,8 +264,6 @@ def generate_arena_images(
|
|
220 |
width_b,
|
221 |
seed,
|
222 |
num_images_per_prompt,
|
223 |
-
model_choice_b,
|
224 |
-
generator,
|
225 |
prior_num_inference_steps_b,
|
226 |
prior_guidance_scale_b,
|
227 |
decoder_num_inference_steps_b,
|
@@ -233,6 +275,7 @@ def generate_arena_images(
|
|
233 |
if num_models_to_compare >= 3:
|
234 |
images_c = generate_single_image(
|
235 |
prompt,
|
|
|
236 |
negative_prompt,
|
237 |
num_inference_steps_c,
|
238 |
guidance_scale_c,
|
@@ -240,8 +283,6 @@ def generate_arena_images(
|
|
240 |
width_c,
|
241 |
seed,
|
242 |
num_images_per_prompt,
|
243 |
-
model_choice_c,
|
244 |
-
generator,
|
245 |
prior_num_inference_steps_c,
|
246 |
prior_guidance_scale_c,
|
247 |
decoder_num_inference_steps_c,
|
@@ -253,6 +294,7 @@ def generate_arena_images(
|
|
253 |
if num_models_to_compare >= 4:
|
254 |
images_d = generate_single_image(
|
255 |
prompt,
|
|
|
256 |
negative_prompt,
|
257 |
num_inference_steps_d,
|
258 |
guidance_scale_d,
|
@@ -260,8 +302,6 @@ def generate_arena_images(
|
|
260 |
width_d,
|
261 |
seed,
|
262 |
num_images_per_prompt,
|
263 |
-
model_choice_d,
|
264 |
-
generator,
|
265 |
prior_num_inference_steps_d,
|
266 |
prior_guidance_scale_d,
|
267 |
decoder_num_inference_steps_d,
|
@@ -274,9 +314,9 @@ def generate_arena_images(
|
|
274 |
|
275 |
|
276 |
# Define the image generation function for the Individual tab
|
277 |
-
@spaces.GPU(duration=90)
|
278 |
def generate_individual_image(
|
279 |
prompt,
|
|
|
280 |
negative_prompt,
|
281 |
num_inference_steps,
|
282 |
guidance_scale,
|
@@ -284,20 +324,16 @@ def generate_individual_image(
|
|
284 |
width,
|
285 |
seed,
|
286 |
num_images_per_prompt,
|
287 |
-
model_choice,
|
288 |
prior_num_inference_steps,
|
289 |
prior_guidance_scale,
|
290 |
decoder_num_inference_steps,
|
291 |
decoder_guidance_scale,
|
292 |
progress=gr.Progress(track_tqdm=True),
|
293 |
):
|
294 |
-
if seed == 0:
|
295 |
-
seed = random.randint(1, MAX_SEED)
|
296 |
-
|
297 |
-
generator = torch.Generator().manual_seed(seed)
|
298 |
|
299 |
output = generate_single_image(
|
300 |
prompt,
|
|
|
301 |
negative_prompt,
|
302 |
num_inference_steps,
|
303 |
guidance_scale,
|
@@ -305,8 +341,6 @@ def generate_individual_image(
|
|
305 |
width,
|
306 |
seed,
|
307 |
num_images_per_prompt,
|
308 |
-
model_choice,
|
309 |
-
generator,
|
310 |
prior_num_inference_steps,
|
311 |
prior_guidance_scale,
|
312 |
decoder_num_inference_steps,
|
@@ -630,18 +664,18 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
630 |
width_a = gr.Slider(
|
631 |
label="Width (Model A)",
|
632 |
info="Width of the Image",
|
633 |
-
minimum=
|
634 |
-
maximum=
|
635 |
-
step=32,
|
636 |
value=1024,
|
|
|
637 |
)
|
638 |
height_a = gr.Slider(
|
639 |
label="Height (Model A)",
|
640 |
info="Height of the Image",
|
641 |
-
minimum=
|
642 |
-
maximum=
|
643 |
-
step=32,
|
644 |
value=1024,
|
|
|
645 |
)
|
646 |
with gr.Column():
|
647 |
num_inference_steps_b = gr.Slider(
|
@@ -650,7 +684,7 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
650 |
minimum=1,
|
651 |
maximum=50,
|
652 |
value=25,
|
653 |
-
step=
|
654 |
visible=True,
|
655 |
)
|
656 |
guidance_scale_b = gr.Slider(
|
@@ -701,18 +735,18 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
701 |
width_b = gr.Slider(
|
702 |
label="Width (Model B)",
|
703 |
info="Width of the Image",
|
704 |
-
minimum=
|
705 |
-
maximum=
|
706 |
-
step=32,
|
707 |
value=1024,
|
|
|
708 |
)
|
709 |
height_b = gr.Slider(
|
710 |
label="Height (Model B)",
|
711 |
info="Height of the Image",
|
712 |
-
minimum=
|
713 |
-
maximum=
|
714 |
-
step=32,
|
715 |
value=1024,
|
|
|
716 |
)
|
717 |
with gr.Column(visible=False) as model_c_options:
|
718 |
num_inference_steps_c = gr.Slider(
|
@@ -772,18 +806,18 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
772 |
width_c = gr.Slider(
|
773 |
label="Width (Model C)",
|
774 |
info="Width of the Image",
|
775 |
-
minimum=
|
776 |
-
maximum=
|
777 |
-
step=32,
|
778 |
value=1024,
|
|
|
779 |
)
|
780 |
height_c = gr.Slider(
|
781 |
label="Height (Model C)",
|
782 |
info="Height of the Image",
|
783 |
-
minimum=
|
784 |
-
maximum=
|
785 |
-
step=32,
|
786 |
value=1024,
|
|
|
787 |
)
|
788 |
with gr.Column(visible=False) as model_d_options:
|
789 |
num_inference_steps_d = gr.Slider(
|
@@ -843,18 +877,18 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
843 |
width_d = gr.Slider(
|
844 |
label="Width (Model D)",
|
845 |
info="Width of the Image",
|
846 |
-
minimum=
|
847 |
-
maximum=
|
848 |
-
step=32,
|
849 |
value=1024,
|
|
|
850 |
)
|
851 |
height_d = gr.Slider(
|
852 |
label="Height (Model D)",
|
853 |
info="Height of the Image",
|
854 |
-
minimum=
|
855 |
-
maximum=
|
856 |
-
step=32,
|
857 |
value=1024,
|
|
|
858 |
)
|
859 |
with gr.Row():
|
860 |
seed = gr.Slider(
|
@@ -883,6 +917,8 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
883 |
prior_guidance_scale_a: gr.update(visible=True),
|
884 |
decoder_num_inference_steps_a: gr.update(visible=True),
|
885 |
decoder_guidance_scale_a: gr.update(visible=True),
|
|
|
|
|
886 |
}
|
887 |
elif model_choice_a == "sdxl flash":
|
888 |
return {
|
@@ -892,6 +928,8 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
892 |
prior_guidance_scale_a: gr.update(visible=False),
|
893 |
decoder_num_inference_steps_a: gr.update(visible=False),
|
894 |
decoder_guidance_scale_a: gr.update(visible=False),
|
|
|
|
|
895 |
}
|
896 |
elif model_choice_a == "sd1.5":
|
897 |
return {
|
@@ -900,26 +938,8 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
900 |
prior_guidance_scale_a: gr.update(visible=True),
|
901 |
decoder_num_inference_steps_a: gr.update(visible=True),
|
902 |
decoder_guidance_scale_a: gr.update(visible=True),
|
903 |
-
|
904 |
-
|
905 |
-
return {
|
906 |
-
num_inference_steps_a: gr.update(visible=True, maximum=15, value=8),
|
907 |
-
guidance_scale_a: gr.update(visible=True, maximum=6.0, value=3.5),
|
908 |
-
prior_num_inference_steps_a: gr.update(visible=False),
|
909 |
-
prior_guidance_scale_a: gr.update(visible=False),
|
910 |
-
decoder_num_inference_steps_a: gr.update(visible=False),
|
911 |
-
decoder_guidance_scale_a: gr.update(visible=False),
|
912 |
-
}
|
913 |
-
elif model_choice_a == "sd1.5":
|
914 |
-
return {
|
915 |
-
num_inference_steps_a: gr.update(visible=True, maximum=50, value=25),
|
916 |
-
guidance_scale_a: gr.update(visible=True, maximum=10.0, value=7.5),
|
917 |
-
prior_num_inference_steps_a: gr.update(visible=False),
|
918 |
-
prior_guidance_scale_a: gr.update(visible=False),
|
919 |
-
decoder_num_inference_steps_a: gr.update(visible=False),
|
920 |
-
decoder_guidance_scale_a: gr.update(visible=False),
|
921 |
-
width_a: gr.update(value=512, maximum=768),
|
922 |
-
height_a: gr.update(value=512, maximum=768),
|
923 |
}
|
924 |
elif model_choice_a == "sd2.1":
|
925 |
return {
|
@@ -929,8 +949,8 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
929 |
prior_guidance_scale_a: gr.update(visible=False),
|
930 |
decoder_num_inference_steps_a: gr.update(visible=False),
|
931 |
decoder_guidance_scale_a: gr.update(visible=False),
|
932 |
-
width_a: gr.update(value=768, maximum=1024),
|
933 |
-
height_a: gr.update(value=768, maximum=1024),
|
934 |
}
|
935 |
else:
|
936 |
return {
|
@@ -940,8 +960,8 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
940 |
prior_guidance_scale_a: gr.update(visible=False),
|
941 |
decoder_num_inference_steps_a: gr.update(visible=False),
|
942 |
decoder_guidance_scale_a: gr.update(visible=False),
|
943 |
-
width_a: gr.update(maximum=
|
944 |
-
height_a: gr.update(maximum=
|
945 |
}
|
946 |
|
947 |
def toggle_visibility_arena_b(model_choice_b):
|
@@ -953,6 +973,8 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
953 |
prior_guidance_scale_b: gr.update(visible=True),
|
954 |
decoder_num_inference_steps_b: gr.update(visible=True),
|
955 |
decoder_guidance_scale_b: gr.update(visible=True),
|
|
|
|
|
956 |
}
|
957 |
elif model_choice_b == "sdxl flash":
|
958 |
return {
|
@@ -962,6 +984,8 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
962 |
prior_guidance_scale_b: gr.update(visible=False),
|
963 |
decoder_num_inference_steps_b: gr.update(visible=False),
|
964 |
decoder_guidance_scale_b: gr.update(visible=False),
|
|
|
|
|
965 |
}
|
966 |
elif model_choice_b == "sd1.5":
|
967 |
return {
|
@@ -971,8 +995,8 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
971 |
prior_guidance_scale_b: gr.update(visible=False),
|
972 |
decoder_num_inference_steps_b: gr.update(visible=False),
|
973 |
decoder_guidance_scale_b: gr.update(visible=False),
|
974 |
-
width_b: gr.update(value=512, maximum=768),
|
975 |
-
height_b: gr.update(value=512, maximum=768),
|
976 |
}
|
977 |
elif model_choice_b == "sd2.1":
|
978 |
return {
|
@@ -982,8 +1006,8 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
982 |
prior_guidance_scale_b: gr.update(visible=False),
|
983 |
decoder_num_inference_steps_b: gr.update(visible=False),
|
984 |
decoder_guidance_scale_b: gr.update(visible=False),
|
985 |
-
width_b: gr.update(value=768, maximum=1024),
|
986 |
-
height_b: gr.update(value=768, maximum=1024),
|
987 |
}
|
988 |
else:
|
989 |
return {
|
@@ -993,8 +1017,8 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
993 |
prior_guidance_scale_b: gr.update(visible=False),
|
994 |
decoder_num_inference_steps_b: gr.update(visible=False),
|
995 |
decoder_guidance_scale_b: gr.update(visible=False),
|
996 |
-
width_b: gr.update(maximum=
|
997 |
-
height_b: gr.update(maximum=
|
998 |
}
|
999 |
|
1000 |
def toggle_visibility_arena_c(model_choice_c):
|
@@ -1006,8 +1030,8 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
1006 |
prior_guidance_scale_c: gr.update(visible=True),
|
1007 |
decoder_num_inference_steps_c: gr.update(visible=True),
|
1008 |
decoder_guidance_scale_c: gr.update(visible=True),
|
1009 |
-
width_c: gr.update(value=1024, maximum=
|
1010 |
-
height_c: gr.update(value=1024, maximum=
|
1011 |
}
|
1012 |
elif model_choice_c == "sdxl flash":
|
1013 |
return {
|
@@ -1017,8 +1041,8 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
1017 |
prior_guidance_scale_c: gr.update(visible=False),
|
1018 |
decoder_num_inference_steps_c: gr.update(visible=False),
|
1019 |
decoder_guidance_scale_c: gr.update(visible=False),
|
1020 |
-
width_c: gr.update(value=1024, maximum=
|
1021 |
-
height_c: gr.update(value=1024, maximum=
|
1022 |
}
|
1023 |
elif model_choice_c == "sd1.5":
|
1024 |
return {
|
@@ -1028,8 +1052,8 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
1028 |
prior_guidance_scale_c: gr.update(visible=False),
|
1029 |
decoder_num_inference_steps_c: gr.update(visible=False),
|
1030 |
decoder_guidance_scale_c: gr.update(visible=False),
|
1031 |
-
width_c: gr.update(value=512, maximum=768),
|
1032 |
-
height_c: gr.update(value=512, maximum=768),
|
1033 |
}
|
1034 |
elif model_choice_c == "sd2.1":
|
1035 |
return {
|
@@ -1039,8 +1063,8 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
1039 |
prior_guidance_scale_c: gr.update(visible=False),
|
1040 |
decoder_num_inference_steps_c: gr.update(visible=False),
|
1041 |
decoder_guidance_scale_c: gr.update(visible=False),
|
1042 |
-
width_c: gr.update(value=768, maximum=1024),
|
1043 |
-
height_c: gr.update(value=768, maximum=1024),
|
1044 |
}
|
1045 |
else:
|
1046 |
return {
|
@@ -1050,8 +1074,8 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
1050 |
prior_guidance_scale_c: gr.update(visible=False),
|
1051 |
decoder_num_inference_steps_c: gr.update(visible=False),
|
1052 |
decoder_guidance_scale_c: gr.update(visible=False),
|
1053 |
-
width_c: gr.update(value=1024, maximum=
|
1054 |
-
height_c: gr.update(value=1024, maximum=
|
1055 |
}
|
1056 |
|
1057 |
def toggle_visibility_arena_d(model_choice_d):
|
@@ -1063,8 +1087,8 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
1063 |
prior_guidance_scale_d: gr.update(visible=True),
|
1064 |
decoder_num_inference_steps_d: gr.update(visible=True),
|
1065 |
decoder_guidance_scale_d: gr.update(visible=True),
|
1066 |
-
width_d: gr.update(value=1024, maximum=
|
1067 |
-
height_d: gr.update(value=1024, maximum=
|
1068 |
}
|
1069 |
elif model_choice_d == "sdxl flash":
|
1070 |
return {
|
@@ -1074,8 +1098,8 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
1074 |
prior_guidance_scale_d: gr.update(visible=False),
|
1075 |
decoder_num_inference_steps_d: gr.update(visible=False),
|
1076 |
decoder_guidance_scale_d: gr.update(visible=False),
|
1077 |
-
width_d: gr.update(value=1024, maximum=
|
1078 |
-
height_d: gr.update(value=1024, maximum=
|
1079 |
}
|
1080 |
elif model_choice_d == "sd1.5":
|
1081 |
return {
|
@@ -1085,8 +1109,8 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
1085 |
prior_guidance_scale_d: gr.update(visible=False),
|
1086 |
decoder_num_inference_steps_d: gr.update(visible=False),
|
1087 |
decoder_guidance_scale_d: gr.update(visible=False),
|
1088 |
-
width_d: gr.update(value=512, maximum=768),
|
1089 |
-
height_d: gr.update(value=512, maximum=768),
|
1090 |
}
|
1091 |
elif model_choice_d == "sd2.1":
|
1092 |
return {
|
@@ -1096,8 +1120,8 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
1096 |
prior_guidance_scale_d: gr.update(visible=False),
|
1097 |
decoder_num_inference_steps_d: gr.update(visible=False),
|
1098 |
decoder_guidance_scale_d: gr.update(visible=False),
|
1099 |
-
width_d: gr.update(value=768, maximum=1024),
|
1100 |
-
height_d: gr.update(value=768, maximum=1024),
|
1101 |
}
|
1102 |
else:
|
1103 |
return {
|
@@ -1107,8 +1131,8 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
1107 |
prior_guidance_scale_d: gr.update(visible=False),
|
1108 |
decoder_num_inference_steps_d: gr.update(visible=False),
|
1109 |
decoder_guidance_scale_d: gr.update(visible=False),
|
1110 |
-
width_d: gr.update(value=1024, maximum=
|
1111 |
-
height_d: gr.update(value=1024, maximum=
|
1112 |
}
|
1113 |
|
1114 |
model_choice_a.change(
|
@@ -1402,18 +1426,18 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
1402 |
width = gr.Slider(
|
1403 |
label="Width",
|
1404 |
info="Width of the Image",
|
1405 |
-
minimum=
|
1406 |
-
maximum=
|
1407 |
-
step=32,
|
1408 |
value=1024,
|
|
|
1409 |
)
|
1410 |
height = gr.Slider(
|
1411 |
label="Height",
|
1412 |
info="Height of the Image",
|
1413 |
-
minimum=
|
1414 |
-
maximum=
|
1415 |
-
step=32,
|
1416 |
value=1024,
|
|
|
1417 |
)
|
1418 |
with gr.Row():
|
1419 |
seed = gr.Slider(
|
@@ -1442,8 +1466,8 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
1442 |
prior_guidance_scale: gr.update(visible=True),
|
1443 |
decoder_num_inference_steps: gr.update(visible=True),
|
1444 |
decoder_guidance_scale: gr.update(visible=True),
|
1445 |
-
width: gr.update(value=1024, maximum=
|
1446 |
-
height: gr.update(value=1024, maximum=
|
1447 |
}
|
1448 |
elif model_choice == "sdxl flash":
|
1449 |
return {
|
@@ -1453,8 +1477,8 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
1453 |
prior_guidance_scale: gr.update(visible=False),
|
1454 |
decoder_num_inference_steps: gr.update(visible=False),
|
1455 |
decoder_guidance_scale: gr.update(visible=False),
|
1456 |
-
width: gr.update(value=1024, maximum=
|
1457 |
-
height: gr.update(value=1024, maximum=
|
1458 |
}
|
1459 |
elif model_choice == "sd1.5":
|
1460 |
return {
|
@@ -1464,8 +1488,8 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
1464 |
prior_guidance_scale: gr.update(visible=False),
|
1465 |
decoder_num_inference_steps: gr.update(visible=False),
|
1466 |
decoder_guidance_scale: gr.update(visible=False),
|
1467 |
-
width: gr.update(value=512, maximum=768),
|
1468 |
-
height: gr.update(value=512, maximum=768),
|
1469 |
}
|
1470 |
elif model_choice == "sd2.1":
|
1471 |
return {
|
@@ -1475,8 +1499,8 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
1475 |
prior_guidance_scale: gr.update(visible=False),
|
1476 |
decoder_num_inference_steps: gr.update(visible=False),
|
1477 |
decoder_guidance_scale: gr.update(visible=False),
|
1478 |
-
width: gr.update(value=768, maximum=1024),
|
1479 |
-
height: gr.update(value=768, maximum=1024),
|
1480 |
}
|
1481 |
else:
|
1482 |
return {
|
@@ -1486,8 +1510,8 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
1486 |
prior_guidance_scale: gr.update(visible=False),
|
1487 |
decoder_num_inference_steps: gr.update(visible=False),
|
1488 |
decoder_guidance_scale: gr.update(visible=False),
|
1489 |
-
width: gr.update(value=1024, maximum=
|
1490 |
-
height: gr.update(value=1024, maximum=
|
1491 |
}
|
1492 |
|
1493 |
model_choice.change(
|
@@ -1509,6 +1533,7 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
1509 |
examples=examples_individual,
|
1510 |
inputs=[
|
1511 |
prompt,
|
|
|
1512 |
negative_prompt,
|
1513 |
num_inference_steps,
|
1514 |
guidance_scale,
|
@@ -1516,7 +1541,6 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
1516 |
width,
|
1517 |
seed,
|
1518 |
num_images_per_prompt,
|
1519 |
-
model_choice,
|
1520 |
prior_num_inference_steps,
|
1521 |
prior_guidance_scale,
|
1522 |
decoder_num_inference_steps,
|
@@ -1534,6 +1558,7 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
1534 |
fn=generate_individual_image,
|
1535 |
inputs=[
|
1536 |
prompt,
|
|
|
1537 |
negative_prompt,
|
1538 |
num_inference_steps,
|
1539 |
guidance_scale,
|
@@ -1541,7 +1566,6 @@ with gr.Blocks(theme=theme, css=css) as demo:
|
|
1541 |
width,
|
1542 |
seed,
|
1543 |
num_images_per_prompt,
|
1544 |
-
model_choice,
|
1545 |
prior_num_inference_steps,
|
1546 |
prior_guidance_scale,
|
1547 |
decoder_num_inference_steps,
|
|
|
3 |
import gradio as gr
|
4 |
import os
|
5 |
import random
|
6 |
+
import numpy
|
7 |
from PIL import Image
|
8 |
+
import gc # free up memory
|
9 |
import spaces
|
10 |
|
11 |
HF_TOKEN = os.getenv("HF_TOKEN") # login with hf read token to access sd gated models
|
|
|
18 |
print("Using CPU")
|
19 |
|
20 |
|
21 |
+
MAX_SEED = numpy.iinfo(numpy.int32).max
|
22 |
|
23 |
+
# Global dictionary to store pipelines
|
24 |
+
PIPELINES = {}
|
25 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
26 |
|
27 |
+
def load_pipeline(model_choice):
|
28 |
+
"""Loads the specified pipeline and stores it in the PIPELINES dictionary."""
|
29 |
+
if model_choice not in PIPELINES:
|
30 |
+
if model_choice == "sd3 medium":
|
31 |
+
PIPELINES[model_choice] = StableDiffusion3Pipeline.from_pretrained(
|
32 |
+
"stabilityai/stable-diffusion-3-medium-diffusers", torch_dtype=torch.float16
|
33 |
+
)
|
34 |
+
elif model_choice == "sd2.1":
|
35 |
+
PIPELINES[model_choice] = StableDiffusionPipeline.from_pretrained(
|
36 |
+
"stabilityai/stable-diffusion-2-1", torch_dtype=torch.float16
|
37 |
+
)
|
38 |
+
elif model_choice == "sdxl":
|
39 |
+
PIPELINES[model_choice] = StableDiffusionXLPipeline.from_pretrained(
|
40 |
+
"stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16
|
41 |
+
)
|
42 |
+
elif model_choice == "sdxl flash":
|
43 |
+
PIPELINES[model_choice] = StableDiffusionXLPipeline.from_pretrained(
|
44 |
+
"sd-community/sdxl-flash", torch_dtype=torch.float16
|
45 |
+
)
|
46 |
+
# Store the original scheduler for resetting
|
47 |
+
PIPELINES[model_choice].original_scheduler = PIPELINES[model_choice].scheduler
|
48 |
+
elif model_choice == "stable cascade":
|
49 |
+
# Store both prior and decoder pipelines under 'stable cascade'
|
50 |
+
PIPELINES[model_choice] = {
|
51 |
+
'prior': StableCascadePriorPipeline.from_pretrained(
|
52 |
+
"stabilityai/stable-cascade-prior", variant="bf16", torch_dtype=torch.bfloat16
|
53 |
+
),
|
54 |
+
'decoder': StableCascadeDecoderPipeline.from_pretrained(
|
55 |
+
"stabilityai/stable-cascade", variant="bf16", torch_dtype=torch.float16
|
56 |
+
)
|
57 |
+
}
|
58 |
+
elif model_choice == "sd1.5":
|
59 |
+
PIPELINES[model_choice] = StableDiffusionPipeline.from_pretrained(
|
60 |
+
"runwayml/stable-diffusion-v1-5", torch_dtype=torch.float16
|
61 |
+
)
|
62 |
+
else:
|
63 |
+
raise ValueError(f"Invalid model choice: {model_choice}")
|
64 |
|
65 |
+
return PIPELINES[model_choice]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
66 |
|
|
|
|
|
|
|
|
|
|
|
67 |
|
68 |
+
def unload_pipeline(model_choice):
|
69 |
+
"""Unloads the specified pipeline from the PIPELINES dictionary and frees GPU memory."""
|
70 |
+
if model_choice in PIPELINES:
|
71 |
+
del PIPELINES[model_choice]
|
72 |
+
|
73 |
+
torch.cuda.empty_cache()
|
74 |
+
gc.collect()
|
75 |
|
|
|
76 |
@spaces.GPU(duration=80)
|
77 |
+
def run_inference(
|
78 |
prompt,
|
79 |
+
pipe,
|
80 |
negative_prompt,
|
81 |
num_inference_steps,
|
82 |
guidance_scale,
|
|
|
84 |
width,
|
85 |
seed,
|
86 |
num_images_per_prompt,
|
|
|
|
|
87 |
prior_num_inference_steps=None,
|
88 |
prior_guidance_scale=None,
|
89 |
decoder_num_inference_steps=None,
|
90 |
decoder_guidance_scale=None,
|
91 |
):
|
92 |
+
"""Runs inference with the specified pipeline and parameters."""
|
93 |
+
|
94 |
+
# Enable CPU offloading only if a GPU is available, for saving up RAM
|
95 |
+
if torch.cuda.is_available():
|
96 |
+
if isinstance(pipe, dict): # Special handling for stable cascade
|
97 |
+
pipe['prior'].enable_model_cpu_offload()
|
98 |
+
pipe['decoder'].enable_model_cpu_offload()
|
99 |
+
else:
|
100 |
+
pipe.enable_model_cpu_offload()
|
|
|
|
|
|
|
|
|
|
|
|
|
101 |
|
102 |
+
# Reset the sampler if the model is NOT SDXL Flash
|
103 |
+
if model_choice != "sdxl flash" and "sdxl flash" in PIPELINES:
|
104 |
+
PIPELINES["sdxl flash"].scheduler = PIPELINES["sdxl flash"].original_scheduler
|
105 |
+
|
106 |
+
# Apply SDXL Flash sampler ONLY if model_choice is 'sdxl flash'
|
107 |
+
if model_choice == "sdxl flash":
|
108 |
+
pipe.scheduler = DPMSolverSinglestepScheduler.from_config(
|
109 |
+
pipe.scheduler.config, timestep_spacing="trailing"
|
|
|
|
|
|
|
110 |
)
|
111 |
|
112 |
+
if seed == 0:
|
113 |
+
seed = random.randint(1, MAX_SEED)
|
114 |
+
|
115 |
+
generator = torch.Generator().manual_seed(seed)
|
|
|
|
|
|
|
116 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
117 |
|
118 |
+
if isinstance(pipe, dict): # Stable Cascade
|
119 |
+
with torch.inference_mode():
|
120 |
+
prior_output = pipe['prior'](
|
121 |
+
prompt=prompt,
|
122 |
+
negative_prompt=negative_prompt,
|
123 |
+
num_inference_steps=prior_num_inference_steps,
|
124 |
+
guidance_scale=prior_guidance_scale,
|
125 |
+
height=height,
|
126 |
+
width=width,
|
127 |
+
generator=generator,
|
128 |
+
num_images_per_prompt=num_images_per_prompt,
|
129 |
+
)
|
130 |
+
with torch.inference_mode():
|
131 |
+
output = pipe['decoder'](
|
132 |
+
image_embeddings=prior_output.image_embeddings.to(torch.float16),
|
133 |
+
prompt=prompt,
|
134 |
+
negative_prompt=negative_prompt,
|
135 |
+
num_inference_steps=decoder_num_inference_steps,
|
136 |
+
guidance_scale=decoder_guidance_scale,
|
137 |
+
).images
|
138 |
+
else: # Other pipelines
|
139 |
+
with torch.inference_mode():
|
140 |
+
output = pipe(
|
141 |
+
prompt=prompt,
|
142 |
+
negative_prompt=negative_prompt,
|
143 |
+
num_inference_steps=num_inference_steps,
|
144 |
+
guidance_scale=guidance_scale,
|
145 |
+
height=height,
|
146 |
+
width=width,
|
147 |
+
generator=generator,
|
148 |
+
num_images_per_prompt=num_images_per_prompt,
|
149 |
+
).images
|
150 |
|
151 |
return output
|
152 |
|
153 |
+
# Helper function to generate images for a single model
|
154 |
+
def generate_single_image(
|
155 |
+
prompt,
|
156 |
+
model_choice,
|
157 |
+
negative_prompt,
|
158 |
+
num_inference_steps,
|
159 |
+
guidance_scale,
|
160 |
+
height,
|
161 |
+
width,
|
162 |
+
seed,
|
163 |
+
num_images_per_prompt,
|
164 |
+
prior_num_inference_steps=None,
|
165 |
+
prior_guidance_scale=None,
|
166 |
+
decoder_num_inference_steps=None,
|
167 |
+
decoder_guidance_scale=None,
|
168 |
+
):
|
169 |
+
# Load the pipeline
|
170 |
+
pipe = load_pipeline(model_choice)
|
171 |
+
|
172 |
+
# Run inference
|
173 |
+
output = run_inference(
|
174 |
+
prompt,
|
175 |
+
pipe,
|
176 |
+
negative_prompt,
|
177 |
+
num_inference_steps,
|
178 |
+
guidance_scale,
|
179 |
+
height,
|
180 |
+
width,
|
181 |
+
seed,
|
182 |
+
num_images_per_prompt,
|
183 |
+
prior_num_inference_steps,
|
184 |
+
prior_guidance_scale,
|
185 |
+
decoder_num_inference_steps,
|
186 |
+
decoder_guidance_scale,
|
187 |
+
)
|
188 |
+
|
189 |
+
# Unload the pipeline
|
190 |
+
unload_pipeline(model_choice)
|
191 |
+
|
192 |
+
return output
|
193 |
|
194 |
# Define the image generation function for the Arena tab
|
|
|
195 |
def generate_arena_images(
|
196 |
prompt,
|
197 |
negative_prompt,
|
|
|
236 |
decoder_guidance_scale_d,
|
237 |
progress=gr.Progress(track_tqdm=True),
|
238 |
):
|
|
|
|
|
|
|
|
|
239 |
|
240 |
# Generate images for selected models
|
241 |
if num_models_to_compare >= 2:
|
242 |
images_a = generate_single_image(
|
243 |
prompt,
|
244 |
+
model_choice_a,
|
245 |
negative_prompt,
|
246 |
num_inference_steps_a,
|
247 |
guidance_scale_a,
|
|
|
249 |
width_a,
|
250 |
seed,
|
251 |
num_images_per_prompt,
|
|
|
|
|
252 |
prior_num_inference_steps_a,
|
253 |
prior_guidance_scale_a,
|
254 |
decoder_num_inference_steps_a,
|
|
|
256 |
)
|
257 |
images_b = generate_single_image(
|
258 |
prompt,
|
259 |
+
model_choice_b,
|
260 |
negative_prompt,
|
261 |
num_inference_steps_b,
|
262 |
guidance_scale_b,
|
|
|
264 |
width_b,
|
265 |
seed,
|
266 |
num_images_per_prompt,
|
|
|
|
|
267 |
prior_num_inference_steps_b,
|
268 |
prior_guidance_scale_b,
|
269 |
decoder_num_inference_steps_b,
|
|
|
275 |
if num_models_to_compare >= 3:
|
276 |
images_c = generate_single_image(
|
277 |
prompt,
|
278 |
+
model_choice_c,
|
279 |
negative_prompt,
|
280 |
num_inference_steps_c,
|
281 |
guidance_scale_c,
|
|
|
283 |
width_c,
|
284 |
seed,
|
285 |
num_images_per_prompt,
|
|
|
|
|
286 |
prior_num_inference_steps_c,
|
287 |
prior_guidance_scale_c,
|
288 |
decoder_num_inference_steps_c,
|
|
|
294 |
if num_models_to_compare >= 4:
|
295 |
images_d = generate_single_image(
|
296 |
prompt,
|
297 |
+
model_choice_d,
|
298 |
negative_prompt,
|
299 |
num_inference_steps_d,
|
300 |
guidance_scale_d,
|
|
|
302 |
width_d,
|
303 |
seed,
|
304 |
num_images_per_prompt,
|
|
|
|
|
305 |
prior_num_inference_steps_d,
|
306 |
prior_guidance_scale_d,
|
307 |
decoder_num_inference_steps_d,
|
|
|
314 |
|
315 |
|
316 |
# Define the image generation function for the Individual tab
|
|
|
317 |
def generate_individual_image(
|
318 |
prompt,
|
319 |
+
model_choice,
|
320 |
negative_prompt,
|
321 |
num_inference_steps,
|
322 |
guidance_scale,
|
|
|
324 |
width,
|
325 |
seed,
|
326 |
num_images_per_prompt,
|
|
|
327 |
prior_num_inference_steps,
|
328 |
prior_guidance_scale,
|
329 |
decoder_num_inference_steps,
|
330 |
decoder_guidance_scale,
|
331 |
progress=gr.Progress(track_tqdm=True),
|
332 |
):
|
|
|
|
|
|
|
|
|
333 |
|
334 |
output = generate_single_image(
|
335 |
prompt,
|
336 |
+
model_choice,
|
337 |
negative_prompt,
|
338 |
num_inference_steps,
|
339 |
guidance_scale,
|
|
|
341 |
width,
|
342 |
seed,
|
343 |
num_images_per_prompt,
|
|
|
|
|
344 |
prior_num_inference_steps,
|
345 |
prior_guidance_scale,
|
346 |
decoder_num_inference_steps,
|
|
|
664 |
width_a = gr.Slider(
|
665 |
label="Width (Model A)",
|
666 |
info="Width of the Image",
|
667 |
+
minimum=512,
|
668 |
+
maximum=1280,
|
|
|
669 |
value=1024,
|
670 |
+
step=32,
|
671 |
)
|
672 |
height_a = gr.Slider(
|
673 |
label="Height (Model A)",
|
674 |
info="Height of the Image",
|
675 |
+
minimum=512,
|
676 |
+
maximum=1280,
|
|
|
677 |
value=1024,
|
678 |
+
step=32,
|
679 |
)
|
680 |
with gr.Column():
|
681 |
num_inference_steps_b = gr.Slider(
|
|
|
684 |
minimum=1,
|
685 |
maximum=50,
|
686 |
value=25,
|
687 |
+
step=32,
|
688 |
visible=True,
|
689 |
)
|
690 |
guidance_scale_b = gr.Slider(
|
|
|
735 |
width_b = gr.Slider(
|
736 |
label="Width (Model B)",
|
737 |
info="Width of the Image",
|
738 |
+
minimum=512,
|
739 |
+
maximum=1280,
|
|
|
740 |
value=1024,
|
741 |
+
step=32,
|
742 |
)
|
743 |
height_b = gr.Slider(
|
744 |
label="Height (Model B)",
|
745 |
info="Height of the Image",
|
746 |
+
minimum=512,
|
747 |
+
maximum=1280,
|
|
|
748 |
value=1024,
|
749 |
+
step=32,
|
750 |
)
|
751 |
with gr.Column(visible=False) as model_c_options:
|
752 |
num_inference_steps_c = gr.Slider(
|
|
|
806 |
width_c = gr.Slider(
|
807 |
label="Width (Model C)",
|
808 |
info="Width of the Image",
|
809 |
+
minimum=512,
|
810 |
+
maximum=1280,
|
|
|
811 |
value=1024,
|
812 |
+
step=32,
|
813 |
)
|
814 |
height_c = gr.Slider(
|
815 |
label="Height (Model C)",
|
816 |
info="Height of the Image",
|
817 |
+
minimum=512,
|
818 |
+
maximum=1280,
|
|
|
819 |
value=1024,
|
820 |
+
step=32,
|
821 |
)
|
822 |
with gr.Column(visible=False) as model_d_options:
|
823 |
num_inference_steps_d = gr.Slider(
|
|
|
877 |
width_d = gr.Slider(
|
878 |
label="Width (Model D)",
|
879 |
info="Width of the Image",
|
880 |
+
minimum=512,
|
881 |
+
maximum=1280,
|
|
|
882 |
value=1024,
|
883 |
+
step=32,
|
884 |
)
|
885 |
height_d = gr.Slider(
|
886 |
label="Height (Model D)",
|
887 |
info="Height of the Image",
|
888 |
+
minimum=512,
|
889 |
+
maximum=1280,
|
|
|
890 |
value=1024,
|
891 |
+
step=32,
|
892 |
)
|
893 |
with gr.Row():
|
894 |
seed = gr.Slider(
|
|
|
917 |
prior_guidance_scale_a: gr.update(visible=True),
|
918 |
decoder_num_inference_steps_a: gr.update(visible=True),
|
919 |
decoder_guidance_scale_a: gr.update(visible=True),
|
920 |
+
width_a: gr.update(step=512, value=1024, maximum=1536),
|
921 |
+
height_a: gr.update(step=512, value=1024, maximum=1536),
|
922 |
}
|
923 |
elif model_choice_a == "sdxl flash":
|
924 |
return {
|
|
|
928 |
prior_guidance_scale_a: gr.update(visible=False),
|
929 |
decoder_num_inference_steps_a: gr.update(visible=False),
|
930 |
decoder_guidance_scale_a: gr.update(visible=False),
|
931 |
+
width_a: gr.update(step=32, value=1024, maximum=1536),
|
932 |
+
height_a: gr.update(step=32, value=1024, maximum=1536),
|
933 |
}
|
934 |
elif model_choice_a == "sd1.5":
|
935 |
return {
|
|
|
938 |
prior_guidance_scale_a: gr.update(visible=True),
|
939 |
decoder_num_inference_steps_a: gr.update(visible=True),
|
940 |
decoder_guidance_scale_a: gr.update(visible=True),
|
941 |
+
width_a: gr.update(step=32, value=512, maximum=768),
|
942 |
+
height_a: gr.update(step=32, value=512, maximum=768),
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
943 |
}
|
944 |
elif model_choice_a == "sd2.1":
|
945 |
return {
|
|
|
949 |
prior_guidance_scale_a: gr.update(visible=False),
|
950 |
decoder_num_inference_steps_a: gr.update(visible=False),
|
951 |
decoder_guidance_scale_a: gr.update(visible=False),
|
952 |
+
width_a: gr.update(step=32, value=768, maximum=1024),
|
953 |
+
height_a: gr.update(step=32, value=768, maximum=1024),
|
954 |
}
|
955 |
else:
|
956 |
return {
|
|
|
960 |
prior_guidance_scale_a: gr.update(visible=False),
|
961 |
decoder_num_inference_steps_a: gr.update(visible=False),
|
962 |
decoder_guidance_scale_a: gr.update(visible=False),
|
963 |
+
width_a: gr.update(step=32, value=1024, maximum=1536),
|
964 |
+
height_a: gr.update(step=32, value=1024, maximum=1536),
|
965 |
}
|
966 |
|
967 |
def toggle_visibility_arena_b(model_choice_b):
|
|
|
973 |
prior_guidance_scale_b: gr.update(visible=True),
|
974 |
decoder_num_inference_steps_b: gr.update(visible=True),
|
975 |
decoder_guidance_scale_b: gr.update(visible=True),
|
976 |
+
width_b: gr.update(step=256, value=1024, maximum=1536),
|
977 |
+
height_b: gr.update(step=256, value=1024, maximum=1536),
|
978 |
}
|
979 |
elif model_choice_b == "sdxl flash":
|
980 |
return {
|
|
|
984 |
prior_guidance_scale_b: gr.update(visible=False),
|
985 |
decoder_num_inference_steps_b: gr.update(visible=False),
|
986 |
decoder_guidance_scale_b: gr.update(visible=False),
|
987 |
+
width_a: gr.update(step=32, value=1024, maximum=1536),
|
988 |
+
height_a: gr.update(step=32, value=1024, maximum=1536),
|
989 |
}
|
990 |
elif model_choice_b == "sd1.5":
|
991 |
return {
|
|
|
995 |
prior_guidance_scale_b: gr.update(visible=False),
|
996 |
decoder_num_inference_steps_b: gr.update(visible=False),
|
997 |
decoder_guidance_scale_b: gr.update(visible=False),
|
998 |
+
width_b: gr.update(step=32, value=512, maximum=768),
|
999 |
+
height_b: gr.update(step=32, value=512, maximum=768),
|
1000 |
}
|
1001 |
elif model_choice_b == "sd2.1":
|
1002 |
return {
|
|
|
1006 |
prior_guidance_scale_b: gr.update(visible=False),
|
1007 |
decoder_num_inference_steps_b: gr.update(visible=False),
|
1008 |
decoder_guidance_scale_b: gr.update(visible=False),
|
1009 |
+
width_b: gr.update(step=32, value=768, maximum=1024),
|
1010 |
+
height_b: gr.update(step=32, value=768, maximum=1024),
|
1011 |
}
|
1012 |
else:
|
1013 |
return {
|
|
|
1017 |
prior_guidance_scale_b: gr.update(visible=False),
|
1018 |
decoder_num_inference_steps_b: gr.update(visible=False),
|
1019 |
decoder_guidance_scale_b: gr.update(visible=False),
|
1020 |
+
width_b: gr.update(step=32, value=1024, maximum=1536),
|
1021 |
+
height_b: gr.update(step=32, value=1024, maximum=1536),
|
1022 |
}
|
1023 |
|
1024 |
def toggle_visibility_arena_c(model_choice_c):
|
|
|
1030 |
prior_guidance_scale_c: gr.update(visible=True),
|
1031 |
decoder_num_inference_steps_c: gr.update(visible=True),
|
1032 |
decoder_guidance_scale_c: gr.update(visible=True),
|
1033 |
+
width_c: gr.update(step=256, value=1024, maximum=1536),
|
1034 |
+
height_c: gr.update(step=256, value=1024, maximum=1536),
|
1035 |
}
|
1036 |
elif model_choice_c == "sdxl flash":
|
1037 |
return {
|
|
|
1041 |
prior_guidance_scale_c: gr.update(visible=False),
|
1042 |
decoder_num_inference_steps_c: gr.update(visible=False),
|
1043 |
decoder_guidance_scale_c: gr.update(visible=False),
|
1044 |
+
width_c: gr.update(step=32, value=1024, maximum=1536),
|
1045 |
+
height_c: gr.update(step=32, value=1024, maximum=1536),
|
1046 |
}
|
1047 |
elif model_choice_c == "sd1.5":
|
1048 |
return {
|
|
|
1052 |
prior_guidance_scale_c: gr.update(visible=False),
|
1053 |
decoder_num_inference_steps_c: gr.update(visible=False),
|
1054 |
decoder_guidance_scale_c: gr.update(visible=False),
|
1055 |
+
width_c: gr.update(step=32, value=512, maximum=768),
|
1056 |
+
height_c: gr.update(step=32, value=512, maximum=768),
|
1057 |
}
|
1058 |
elif model_choice_c == "sd2.1":
|
1059 |
return {
|
|
|
1063 |
prior_guidance_scale_c: gr.update(visible=False),
|
1064 |
decoder_num_inference_steps_c: gr.update(visible=False),
|
1065 |
decoder_guidance_scale_c: gr.update(visible=False),
|
1066 |
+
width_c: gr.update(step=32, value=768, maximum=1024),
|
1067 |
+
height_c: gr.update(step=32, value=768, maximum=1024),
|
1068 |
}
|
1069 |
else:
|
1070 |
return {
|
|
|
1074 |
prior_guidance_scale_c: gr.update(visible=False),
|
1075 |
decoder_num_inference_steps_c: gr.update(visible=False),
|
1076 |
decoder_guidance_scale_c: gr.update(visible=False),
|
1077 |
+
width_c: gr.update(step=32, value=1024, maximum=1536),
|
1078 |
+
height_c: gr.update(step=32, value=1024, maximum=1536),
|
1079 |
}
|
1080 |
|
1081 |
def toggle_visibility_arena_d(model_choice_d):
|
|
|
1087 |
prior_guidance_scale_d: gr.update(visible=True),
|
1088 |
decoder_num_inference_steps_d: gr.update(visible=True),
|
1089 |
decoder_guidance_scale_d: gr.update(visible=True),
|
1090 |
+
width_d: gr.update(step=256, value=1024, maximum=1536),
|
1091 |
+
height_d: gr.update(step=256, value=1024, maximum=1536),
|
1092 |
}
|
1093 |
elif model_choice_d == "sdxl flash":
|
1094 |
return {
|
|
|
1098 |
prior_guidance_scale_d: gr.update(visible=False),
|
1099 |
decoder_num_inference_steps_d: gr.update(visible=False),
|
1100 |
decoder_guidance_scale_d: gr.update(visible=False),
|
1101 |
+
width_d: gr.update(step=32, value=1024, maximum=1536),
|
1102 |
+
height_d: gr.update(step=32, value=1024, maximum=1536),
|
1103 |
}
|
1104 |
elif model_choice_d == "sd1.5":
|
1105 |
return {
|
|
|
1109 |
prior_guidance_scale_d: gr.update(visible=False),
|
1110 |
decoder_num_inference_steps_d: gr.update(visible=False),
|
1111 |
decoder_guidance_scale_d: gr.update(visible=False),
|
1112 |
+
width_d: gr.update(step=32, value=512, maximum=768),
|
1113 |
+
height_d: gr.update(step=32, value=512, maximum=768),
|
1114 |
}
|
1115 |
elif model_choice_d == "sd2.1":
|
1116 |
return {
|
|
|
1120 |
prior_guidance_scale_d: gr.update(visible=False),
|
1121 |
decoder_num_inference_steps_d: gr.update(visible=False),
|
1122 |
decoder_guidance_scale_d: gr.update(visible=False),
|
1123 |
+
width_d: gr.update(step=32, value=768, maximum=1024),
|
1124 |
+
height_d: gr.update(step=32, value=768, maximum=1024),
|
1125 |
}
|
1126 |
else:
|
1127 |
return {
|
|
|
1131 |
prior_guidance_scale_d: gr.update(visible=False),
|
1132 |
decoder_num_inference_steps_d: gr.update(visible=False),
|
1133 |
decoder_guidance_scale_d: gr.update(visible=False),
|
1134 |
+
width_d: gr.update(step=32, value=1024, maximum=1536),
|
1135 |
+
height_d: gr.update(step=32, value=1024, maximum=1536),
|
1136 |
}
|
1137 |
|
1138 |
model_choice_a.change(
|
|
|
1426 |
width = gr.Slider(
|
1427 |
label="Width",
|
1428 |
info="Width of the Image",
|
1429 |
+
minimum=512,
|
1430 |
+
maximum=1280,
|
|
|
1431 |
value=1024,
|
1432 |
+
step=32,
|
1433 |
)
|
1434 |
height = gr.Slider(
|
1435 |
label="Height",
|
1436 |
info="Height of the Image",
|
1437 |
+
minimum=512,
|
1438 |
+
maximum=1280,
|
|
|
1439 |
value=1024,
|
1440 |
+
step=32,
|
1441 |
)
|
1442 |
with gr.Row():
|
1443 |
seed = gr.Slider(
|
|
|
1466 |
prior_guidance_scale: gr.update(visible=True),
|
1467 |
decoder_num_inference_steps: gr.update(visible=True),
|
1468 |
decoder_guidance_scale: gr.update(visible=True),
|
1469 |
+
width: gr.update(step=256, value=1024, maximum=1536),
|
1470 |
+
height: gr.update(step=256, value=1024, maximum=1536),
|
1471 |
}
|
1472 |
elif model_choice == "sdxl flash":
|
1473 |
return {
|
|
|
1477 |
prior_guidance_scale: gr.update(visible=False),
|
1478 |
decoder_num_inference_steps: gr.update(visible=False),
|
1479 |
decoder_guidance_scale: gr.update(visible=False),
|
1480 |
+
width: gr.update(step=32, value=1024, maximum=1536),
|
1481 |
+
height: gr.update(step=32, value=1024, maximum=1536),
|
1482 |
}
|
1483 |
elif model_choice == "sd1.5":
|
1484 |
return {
|
|
|
1488 |
prior_guidance_scale: gr.update(visible=False),
|
1489 |
decoder_num_inference_steps: gr.update(visible=False),
|
1490 |
decoder_guidance_scale: gr.update(visible=False),
|
1491 |
+
width: gr.update(step=32, value=512, maximum=768),
|
1492 |
+
height: gr.update(step=32, value=512, maximum=768),
|
1493 |
}
|
1494 |
elif model_choice == "sd2.1":
|
1495 |
return {
|
|
|
1499 |
prior_guidance_scale: gr.update(visible=False),
|
1500 |
decoder_num_inference_steps: gr.update(visible=False),
|
1501 |
decoder_guidance_scale: gr.update(visible=False),
|
1502 |
+
width: gr.update(step=32, value=768, maximum=1024),
|
1503 |
+
height: gr.update(step=32, value=768, maximum=1024),
|
1504 |
}
|
1505 |
else:
|
1506 |
return {
|
|
|
1510 |
prior_guidance_scale: gr.update(visible=False),
|
1511 |
decoder_num_inference_steps: gr.update(visible=False),
|
1512 |
decoder_guidance_scale: gr.update(visible=False),
|
1513 |
+
width: gr.update(step=32, value=1024, maximum=1536),
|
1514 |
+
height: gr.update(step=32, value=1024, maximum=1536),
|
1515 |
}
|
1516 |
|
1517 |
model_choice.change(
|
|
|
1533 |
examples=examples_individual,
|
1534 |
inputs=[
|
1535 |
prompt,
|
1536 |
+
model_choice,
|
1537 |
negative_prompt,
|
1538 |
num_inference_steps,
|
1539 |
guidance_scale,
|
|
|
1541 |
width,
|
1542 |
seed,
|
1543 |
num_images_per_prompt,
|
|
|
1544 |
prior_num_inference_steps,
|
1545 |
prior_guidance_scale,
|
1546 |
decoder_num_inference_steps,
|
|
|
1558 |
fn=generate_individual_image,
|
1559 |
inputs=[
|
1560 |
prompt,
|
1561 |
+
model_choice,
|
1562 |
negative_prompt,
|
1563 |
num_inference_steps,
|
1564 |
guidance_scale,
|
|
|
1566 |
width,
|
1567 |
seed,
|
1568 |
num_images_per_prompt,
|
|
|
1569 |
prior_num_inference_steps,
|
1570 |
prior_guidance_scale,
|
1571 |
decoder_num_inference_steps,
|