Spaces:
Running
on
A10G
Running
on
A10G
Linoy Tsaban
commited on
Commit
•
277aca5
1
Parent(s):
4eb55a5
Update app.py
Browse files
app.py
CHANGED
@@ -50,25 +50,22 @@ device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
|
50 |
sd_pipe = StableDiffusionPipeline.from_pretrained(sd_model_id).to(device)
|
51 |
sd_pipe.scheduler = DDIMScheduler.from_config(sd_model_id, subfolder = "scheduler")
|
52 |
sem_pipe = SemanticStableDiffusionPipeline.from_pretrained(sd_model_id).to(device)
|
|
|
53 |
|
54 |
-
|
55 |
-
def edit(input_image,
|
56 |
src_prompt,
|
57 |
tar_prompt,
|
58 |
steps,
|
59 |
-
src_cfg_scale,
|
60 |
skip,
|
61 |
-
tar_cfg_scale
|
62 |
-
edit_concept,
|
63 |
-
sega_edit_guidance,
|
64 |
-
warm_up,
|
65 |
-
neg_guidance):
|
66 |
offsets=(0,0,0,0)
|
67 |
x0 = load_512(input_image, *offsets, device)
|
68 |
|
69 |
|
70 |
# invert
|
71 |
-
wt, zs, wts = invert(x0 =x0 , prompt_src=src_prompt, num_diffusion_steps=steps, cfg_scale_src=src_cfg_scale)
|
|
|
72 |
latnets = wts[skip].expand(1, -1, -1, -1)
|
73 |
|
74 |
eta = 1
|
@@ -76,7 +73,15 @@ def edit(input_image,
|
|
76 |
pure_ddpm_out = sample(wt, zs, wts, prompt_tar=tar_prompt,
|
77 |
cfg_scale_tar=tar_cfg_scale, skip=skip,
|
78 |
eta = eta)
|
79 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
80 |
editing_args = dict(
|
81 |
editing_prompt = [edit_concept],
|
82 |
reverse_editing_direction = [neg_guidance],
|
@@ -90,7 +95,48 @@ def edit(input_image,
|
|
90 |
num_images_per_prompt=1,
|
91 |
num_inference_steps=steps,
|
92 |
use_ddpm=True, wts=wts, zs=zs[skip:], **editing_args)
|
93 |
-
return
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
94 |
|
95 |
|
96 |
####################################
|
@@ -132,7 +178,7 @@ with gr.Blocks() as demo:
|
|
132 |
with gr.Row():
|
133 |
#inversion
|
134 |
steps = gr.Number(value=100, precision=0, label="Steps", interactive=True)
|
135 |
-
src_cfg_scale = gr.Number(value=3.5, label=f"Source CFG", interactive=True)
|
136 |
# reconstruction
|
137 |
skip = gr.Number(value=36, precision=0, label="Skip", interactive=True)
|
138 |
tar_cfg_scale = gr.Number(value=15, label=f"Reconstruction CFG", interactive=True)
|
@@ -146,20 +192,28 @@ with gr.Blocks() as demo:
|
|
146 |
# gr.Markdown(help_text)
|
147 |
|
148 |
generate_button.click(
|
149 |
-
fn=
|
150 |
inputs=[input_image,
|
151 |
src_prompt,
|
152 |
tar_prompt,
|
153 |
steps,
|
154 |
src_cfg_scale,
|
155 |
skip,
|
156 |
-
tar_cfg_scale
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
157 |
edit_concept,
|
158 |
sega_edit_guidance,
|
159 |
warm_up,
|
160 |
neg_guidance
|
161 |
],
|
162 |
-
outputs=[
|
163 |
)
|
164 |
|
165 |
|
|
|
50 |
sd_pipe = StableDiffusionPipeline.from_pretrained(sd_model_id).to(device)
|
51 |
sd_pipe.scheduler = DDIMScheduler.from_config(sd_model_id, subfolder = "scheduler")
|
52 |
sem_pipe = SemanticStableDiffusionPipeline.from_pretrained(sd_model_id).to(device)
|
53 |
+
latents, wts, zs = None, None, None
|
54 |
|
55 |
+
def invert_and_reconstruct(input_image,
|
|
|
56 |
src_prompt,
|
57 |
tar_prompt,
|
58 |
steps,
|
59 |
+
# src_cfg_scale,
|
60 |
skip,
|
61 |
+
tar_cfg_scale):
|
|
|
|
|
|
|
|
|
62 |
offsets=(0,0,0,0)
|
63 |
x0 = load_512(input_image, *offsets, device)
|
64 |
|
65 |
|
66 |
# invert
|
67 |
+
# wt, zs, wts = invert(x0 =x0 , prompt_src=src_prompt, num_diffusion_steps=steps, cfg_scale_src=src_cfg_scale)
|
68 |
+
wt, zs, wts = invert(x0 =x0 , prompt_src=src_prompt, num_diffusion_steps=steps)
|
69 |
latnets = wts[skip].expand(1, -1, -1, -1)
|
70 |
|
71 |
eta = 1
|
|
|
73 |
pure_ddpm_out = sample(wt, zs, wts, prompt_tar=tar_prompt,
|
74 |
cfg_scale_tar=tar_cfg_scale, skip=skip,
|
75 |
eta = eta)
|
76 |
+
return pure_ddpm_out
|
77 |
+
|
78 |
+
def edit( tar_prompt,
|
79 |
+
steps,
|
80 |
+
edit_concept,
|
81 |
+
sega_edit_guidance,
|
82 |
+
warm_up,
|
83 |
+
neg_guidance):
|
84 |
+
|
85 |
editing_args = dict(
|
86 |
editing_prompt = [edit_concept],
|
87 |
reverse_editing_direction = [neg_guidance],
|
|
|
95 |
num_images_per_prompt=1,
|
96 |
num_inference_steps=steps,
|
97 |
use_ddpm=True, wts=wts, zs=zs[skip:], **editing_args)
|
98 |
+
return sega_out.images[0]
|
99 |
+
|
100 |
+
# def edit(input_image,
|
101 |
+
# src_prompt,
|
102 |
+
# tar_prompt,
|
103 |
+
# steps,
|
104 |
+
# # src_cfg_scale,
|
105 |
+
# skip,
|
106 |
+
# tar_cfg_scale,
|
107 |
+
# edit_concept,
|
108 |
+
# sega_edit_guidance,
|
109 |
+
# warm_up,
|
110 |
+
# neg_guidance):
|
111 |
+
# offsets=(0,0,0,0)
|
112 |
+
# x0 = load_512(input_image, *offsets, device)
|
113 |
+
|
114 |
+
|
115 |
+
# # invert
|
116 |
+
# # wt, zs, wts = invert(x0 =x0 , prompt_src=src_prompt, num_diffusion_steps=steps, cfg_scale_src=src_cfg_scale)
|
117 |
+
# wt, zs, wts = invert(x0 =x0 , prompt_src=src_prompt, num_diffusion_steps=steps)
|
118 |
+
# latnets = wts[skip].expand(1, -1, -1, -1)
|
119 |
+
|
120 |
+
# eta = 1
|
121 |
+
# #pure DDPM output
|
122 |
+
# pure_ddpm_out = sample(wt, zs, wts, prompt_tar=tar_prompt,
|
123 |
+
# cfg_scale_tar=tar_cfg_scale, skip=skip,
|
124 |
+
# eta = eta)
|
125 |
+
|
126 |
+
# editing_args = dict(
|
127 |
+
# editing_prompt = [edit_concept],
|
128 |
+
# reverse_editing_direction = [neg_guidance],
|
129 |
+
# edit_warmup_steps=[warm_up],
|
130 |
+
# edit_guidance_scale=[sega_edit_guidance],
|
131 |
+
# edit_threshold=[.93],
|
132 |
+
# edit_momentum_scale=0.5,
|
133 |
+
# edit_mom_beta=0.6
|
134 |
+
# )
|
135 |
+
# sega_out = sem_pipe(prompt=tar_prompt,eta=eta, latents=latnets,
|
136 |
+
# num_images_per_prompt=1,
|
137 |
+
# num_inference_steps=steps,
|
138 |
+
# use_ddpm=True, wts=wts, zs=zs[skip:], **editing_args)
|
139 |
+
# return pure_ddpm_out,sega_out.images[0]
|
140 |
|
141 |
|
142 |
####################################
|
|
|
178 |
with gr.Row():
|
179 |
#inversion
|
180 |
steps = gr.Number(value=100, precision=0, label="Steps", interactive=True)
|
181 |
+
# src_cfg_scale = gr.Number(value=3.5, label=f"Source CFG", interactive=True)
|
182 |
# reconstruction
|
183 |
skip = gr.Number(value=36, precision=0, label="Skip", interactive=True)
|
184 |
tar_cfg_scale = gr.Number(value=15, label=f"Reconstruction CFG", interactive=True)
|
|
|
192 |
# gr.Markdown(help_text)
|
193 |
|
194 |
generate_button.click(
|
195 |
+
fn=invert_and_reconstruct,
|
196 |
inputs=[input_image,
|
197 |
src_prompt,
|
198 |
tar_prompt,
|
199 |
steps,
|
200 |
src_cfg_scale,
|
201 |
skip,
|
202 |
+
tar_cfg_scale
|
203 |
+
],
|
204 |
+
outputs=[ddpm_edited_image],
|
205 |
+
)
|
206 |
+
|
207 |
+
edit_button.click(
|
208 |
+
fn=edit,
|
209 |
+
inputs=[tar_prompt,
|
210 |
+
steps,
|
211 |
edit_concept,
|
212 |
sega_edit_guidance,
|
213 |
warm_up,
|
214 |
neg_guidance
|
215 |
],
|
216 |
+
outputs=[sega_edited_image],
|
217 |
)
|
218 |
|
219 |
|