Spaces:
Runtime error
Runtime error
use compel for prompt encoding
Browse files- app.py +109 -12
- requirements.txt +2 -1
app.py
CHANGED
@@ -2,6 +2,7 @@ import gradio as gr
|
|
2 |
from gradio_imageslider import ImageSlider
|
3 |
import torch
|
4 |
from diffusers import DiffusionPipeline, AutoencoderKL
|
|
|
5 |
from PIL import Image
|
6 |
from torchvision import transforms
|
7 |
import tempfile
|
@@ -30,7 +31,12 @@ pipe = DiffusionPipeline.from_pretrained(
|
|
30 |
use_safetensors=True,
|
31 |
vae=vae,
|
32 |
)
|
33 |
-
|
|
|
|
|
|
|
|
|
|
|
34 |
pipe = pipe.to(device)
|
35 |
|
36 |
|
@@ -70,6 +76,11 @@ def predict(
|
|
70 |
prompt,
|
71 |
negative_prompt,
|
72 |
seed,
|
|
|
|
|
|
|
|
|
|
|
73 |
scale=2,
|
74 |
progress=gr.Progress(track_tqdm=True),
|
75 |
):
|
@@ -77,11 +88,14 @@ def predict(
|
|
77 |
raise gr.Error("Please upload an image.")
|
78 |
padded_image = pad_image(input_image).resize((1024, 1024)).convert("RGB")
|
79 |
image_lr = load_and_process_image(padded_image).to(device)
|
|
|
80 |
generator = torch.manual_seed(seed)
|
81 |
last_time = time.time()
|
82 |
images = pipe(
|
83 |
-
|
84 |
-
|
|
|
|
|
85 |
image_lr=image_lr,
|
86 |
width=1024 * scale,
|
87 |
height=1024 * scale,
|
@@ -89,11 +103,11 @@ def predict(
|
|
89 |
stride=64,
|
90 |
generator=generator,
|
91 |
num_inference_steps=40,
|
92 |
-
guidance_scale=
|
93 |
-
cosine_scale_1=
|
94 |
-
cosine_scale_2=
|
95 |
-
cosine_scale_3=
|
96 |
-
sigma=
|
97 |
multi_decoder=1024 * scale > 2048,
|
98 |
show_image=False,
|
99 |
lowvram=LOW_MEMORY,
|
@@ -145,13 +159,48 @@ GPU Time Comparison: T4: ~276s - A10G: ~113.6s A100: ~43.5s RTX 4090: ~48.1s
|
|
145 |
label="Negative Prompt",
|
146 |
value="blurry, ugly, duplicate, poorly drawn, deformed, mosaic",
|
147 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
148 |
scale = gr.Slider(
|
149 |
minimum=1,
|
150 |
maximum=5,
|
151 |
value=2,
|
152 |
step=1,
|
153 |
label="x Scale",
|
154 |
-
interactive=
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
155 |
)
|
156 |
seed = gr.Slider(
|
157 |
minimum=0,
|
@@ -165,8 +214,19 @@ GPU Time Comparison: T4: ~276s - A10G: ~113.6s A100: ~43.5s RTX 4090: ~48.1s
|
|
165 |
with gr.Column(scale=2):
|
166 |
image_slider = ImageSlider(position=0.5)
|
167 |
files = gr.Files()
|
168 |
-
|
169 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
170 |
outputs = [image_slider, files]
|
171 |
btn.click(predict, inputs=inputs, outputs=outputs, concurrency_limit=1)
|
172 |
gr.Examples(
|
@@ -177,6 +237,12 @@ GPU Time Comparison: T4: ~276s - A10G: ~113.6s A100: ~43.5s RTX 4090: ~48.1s
|
|
177 |
"photography of lara croft 8k high definition award winning",
|
178 |
"blurry, ugly, duplicate, poorly drawn, deformed, mosaic",
|
179 |
5436236241,
|
|
|
|
|
|
|
|
|
|
|
|
|
180 |
2,
|
181 |
],
|
182 |
[
|
@@ -184,6 +250,12 @@ GPU Time Comparison: T4: ~276s - A10G: ~113.6s A100: ~43.5s RTX 4090: ~48.1s
|
|
184 |
"photo of tesla cybertruck futuristic car 8k high definition on a sand dune in mars, future",
|
185 |
"blurry, ugly, duplicate, poorly drawn, deformed, mosaic",
|
186 |
383472451451,
|
|
|
|
|
|
|
|
|
|
|
|
|
187 |
2,
|
188 |
],
|
189 |
[
|
@@ -191,6 +263,7 @@ GPU Time Comparison: T4: ~276s - A10G: ~113.6s A100: ~43.5s RTX 4090: ~48.1s
|
|
191 |
"a photorealistic painting of Jesus Christ, 4k high definition",
|
192 |
"blurry, ugly, duplicate, poorly drawn, deformed, mosaic",
|
193 |
13317204146129588000,
|
|
|
194 |
2,
|
195 |
],
|
196 |
[
|
@@ -198,6 +271,12 @@ GPU Time Comparison: T4: ~276s - A10G: ~113.6s A100: ~43.5s RTX 4090: ~48.1s
|
|
198 |
"A crowded stadium with enthusiastic fans watching a daytime sporting event, the stands filled with colorful attire and the sun casting a warm glow",
|
199 |
"blurry, ugly, duplicate, poorly drawn, deformed, mosaic",
|
200 |
5623124123512,
|
|
|
|
|
|
|
|
|
|
|
|
|
201 |
2,
|
202 |
],
|
203 |
[
|
@@ -205,12 +284,30 @@ GPU Time Comparison: T4: ~276s - A10G: ~113.6s A100: ~43.5s RTX 4090: ~48.1s
|
|
205 |
"a large red flower on a black background 4k high definition",
|
206 |
"blurry, ugly, duplicate, poorly drawn, deformed, mosaic",
|
207 |
23123412341234,
|
|
|
|
|
|
|
|
|
|
|
|
|
208 |
2,
|
209 |
],
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
210 |
],
|
211 |
inputs=inputs,
|
212 |
outputs=outputs,
|
213 |
-
cache_examples=
|
214 |
)
|
215 |
|
216 |
|
|
|
2 |
from gradio_imageslider import ImageSlider
|
3 |
import torch
|
4 |
from diffusers import DiffusionPipeline, AutoencoderKL
|
5 |
+
from compel import Compel, ReturnedEmbeddingsType
|
6 |
from PIL import Image
|
7 |
from torchvision import transforms
|
8 |
import tempfile
|
|
|
31 |
use_safetensors=True,
|
32 |
vae=vae,
|
33 |
)
|
34 |
+
compel = Compel(
|
35 |
+
tokenizer=[pipe.tokenizer, pipe.tokenizer_2],
|
36 |
+
text_encoder=[pipe.text_encoder, pipe.text_encoder_2],
|
37 |
+
returned_embeddings_type=ReturnedEmbeddingsType.PENULTIMATE_HIDDEN_STATES_NON_NORMALIZED,
|
38 |
+
requires_pooled=[False, True],
|
39 |
+
)
|
40 |
pipe = pipe.to(device)
|
41 |
|
42 |
|
|
|
76 |
prompt,
|
77 |
negative_prompt,
|
78 |
seed,
|
79 |
+
guidance_scale=8.5,
|
80 |
+
cosine_scale_1=3,
|
81 |
+
cosine_scale_2=1,
|
82 |
+
cosine_scale_3=1,
|
83 |
+
sigma=0.8,
|
84 |
scale=2,
|
85 |
progress=gr.Progress(track_tqdm=True),
|
86 |
):
|
|
|
88 |
raise gr.Error("Please upload an image.")
|
89 |
padded_image = pad_image(input_image).resize((1024, 1024)).convert("RGB")
|
90 |
image_lr = load_and_process_image(padded_image).to(device)
|
91 |
+
conditioning, pooled = compel([prompt, negative_prompt])
|
92 |
generator = torch.manual_seed(seed)
|
93 |
last_time = time.time()
|
94 |
images = pipe(
|
95 |
+
prompt_embeds=conditioning[0:1],
|
96 |
+
pooled_prompt_embeds=pooled[0:1],
|
97 |
+
negative_prompt_embeds=conditioning[1:2],
|
98 |
+
negative_pooled_prompt_embeds=pooled[1:2],
|
99 |
image_lr=image_lr,
|
100 |
width=1024 * scale,
|
101 |
height=1024 * scale,
|
|
|
103 |
stride=64,
|
104 |
generator=generator,
|
105 |
num_inference_steps=40,
|
106 |
+
guidance_scale=guidance_scale,
|
107 |
+
cosine_scale_1=cosine_scale_1,
|
108 |
+
cosine_scale_2=cosine_scale_2,
|
109 |
+
cosine_scale_3=cosine_scale_3,
|
110 |
+
sigma=sigma,
|
111 |
multi_decoder=1024 * scale > 2048,
|
112 |
show_image=False,
|
113 |
lowvram=LOW_MEMORY,
|
|
|
159 |
label="Negative Prompt",
|
160 |
value="blurry, ugly, duplicate, poorly drawn, deformed, mosaic",
|
161 |
)
|
162 |
+
guidance_scale = gr.Slider(
|
163 |
+
minimum=0,
|
164 |
+
maximum=50,
|
165 |
+
value=8.5,
|
166 |
+
step=0.001,
|
167 |
+
label="Guidance Scale",
|
168 |
+
)
|
169 |
scale = gr.Slider(
|
170 |
minimum=1,
|
171 |
maximum=5,
|
172 |
value=2,
|
173 |
step=1,
|
174 |
label="x Scale",
|
175 |
+
interactive=True,
|
176 |
+
)
|
177 |
+
cosine_scale_1 = gr.Slider(
|
178 |
+
minimum=0,
|
179 |
+
maximum=5,
|
180 |
+
value=3,
|
181 |
+
step=0.01,
|
182 |
+
label="Cosine Scale 1",
|
183 |
+
)
|
184 |
+
cosine_scale_2 = gr.Slider(
|
185 |
+
minimum=0,
|
186 |
+
maximum=5,
|
187 |
+
value=1,
|
188 |
+
step=0.01,
|
189 |
+
label="Cosine Scale 2",
|
190 |
+
)
|
191 |
+
cosine_scale_3 = gr.Slider(
|
192 |
+
minimum=0,
|
193 |
+
maximum=5,
|
194 |
+
value=1,
|
195 |
+
step=0.01,
|
196 |
+
label="Cosine Scale 3",
|
197 |
+
)
|
198 |
+
sigma = gr.Slider(
|
199 |
+
minimum=0,
|
200 |
+
maximum=1,
|
201 |
+
value=0.8,
|
202 |
+
step=0.01,
|
203 |
+
label="Sigma",
|
204 |
)
|
205 |
seed = gr.Slider(
|
206 |
minimum=0,
|
|
|
214 |
with gr.Column(scale=2):
|
215 |
image_slider = ImageSlider(position=0.5)
|
216 |
files = gr.Files()
|
217 |
+
inputs = [
|
218 |
+
image_input,
|
219 |
+
prompt,
|
220 |
+
negative_prompt,
|
221 |
+
seed,
|
222 |
+
guidance_scale,
|
223 |
+
cosine_scale_1,
|
224 |
+
cosine_scale_2,
|
225 |
+
cosine_scale_3,
|
226 |
+
sigma,
|
227 |
+
scale,
|
228 |
+
]
|
229 |
+
# inputs = [image_input, prompt, negative_prompt, seed]
|
230 |
outputs = [image_slider, files]
|
231 |
btn.click(predict, inputs=inputs, outputs=outputs, concurrency_limit=1)
|
232 |
gr.Examples(
|
|
|
237 |
"photography of lara croft 8k high definition award winning",
|
238 |
"blurry, ugly, duplicate, poorly drawn, deformed, mosaic",
|
239 |
5436236241,
|
240 |
+
8.5,
|
241 |
+
3,
|
242 |
+
1,
|
243 |
+
1,
|
244 |
+
1,
|
245 |
+
0.8,
|
246 |
2,
|
247 |
],
|
248 |
[
|
|
|
250 |
"photo of tesla cybertruck futuristic car 8k high definition on a sand dune in mars, future",
|
251 |
"blurry, ugly, duplicate, poorly drawn, deformed, mosaic",
|
252 |
383472451451,
|
253 |
+
8.5,
|
254 |
+
3,
|
255 |
+
1,
|
256 |
+
1,
|
257 |
+
1,
|
258 |
+
0.8,
|
259 |
2,
|
260 |
],
|
261 |
[
|
|
|
263 |
"a photorealistic painting of Jesus Christ, 4k high definition",
|
264 |
"blurry, ugly, duplicate, poorly drawn, deformed, mosaic",
|
265 |
13317204146129588000,
|
266 |
+
8.5,
|
267 |
2,
|
268 |
],
|
269 |
[
|
|
|
271 |
"A crowded stadium with enthusiastic fans watching a daytime sporting event, the stands filled with colorful attire and the sun casting a warm glow",
|
272 |
"blurry, ugly, duplicate, poorly drawn, deformed, mosaic",
|
273 |
5623124123512,
|
274 |
+
8.5,
|
275 |
+
3,
|
276 |
+
1,
|
277 |
+
1,
|
278 |
+
1,
|
279 |
+
0.8,
|
280 |
2,
|
281 |
],
|
282 |
[
|
|
|
284 |
"a large red flower on a black background 4k high definition",
|
285 |
"blurry, ugly, duplicate, poorly drawn, deformed, mosaic",
|
286 |
23123412341234,
|
287 |
+
8.5,
|
288 |
+
3,
|
289 |
+
1,
|
290 |
+
1,
|
291 |
+
1,
|
292 |
+
0.8,
|
293 |
2,
|
294 |
],
|
295 |
+
[
|
296 |
+
"./examples/huggingface.jpg",
|
297 |
+
"photo realistic huggingface human+++ emoji costume, round, yellow, skin+++ texture+++",
|
298 |
+
"blurry, ugly, duplicate, poorly drawn, deformed, mosaic, emoji cartoon, drawing, pixelated",
|
299 |
+
5532144938416372000,
|
300 |
+
20.0,
|
301 |
+
4.64,
|
302 |
+
1,
|
303 |
+
1,
|
304 |
+
0.49,
|
305 |
+
3,
|
306 |
+
],
|
307 |
],
|
308 |
inputs=inputs,
|
309 |
outputs=outputs,
|
310 |
+
cache_examples=False,
|
311 |
)
|
312 |
|
313 |
|
requirements.txt
CHANGED
@@ -10,4 +10,5 @@ accelerate
|
|
10 |
invisible-watermark
|
11 |
huggingface-hub
|
12 |
hf-transfer
|
13 |
-
gradio_imageslider==0.0.16
|
|
|
|
10 |
invisible-watermark
|
11 |
huggingface-hub
|
12 |
hf-transfer
|
13 |
+
gradio_imageslider==0.0.16
|
14 |
+
compel
|