-
Notifications
You must be signed in to change notification settings - Fork 7
/
run_editing_edit_friendly_p2p.py
177 lines (141 loc) · 8.13 KB
/
run_editing_edit_friendly_p2p.py
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
import torch
from diffusers import StableDiffusionPipeline
from diffusers import DDIMScheduler
import numpy as np
from PIL import Image
import os
from models.p2p.scheduler_dev import DDIMSchedulerDev
import json
import random
import argparse
from torch import autocast, inference_mode
from utils.utils import load_512,txt_draw
from models.edit_friendly_ddm.inversion_utils import inversion_forward_process, inversion_reverse_process
from models.edit_friendly_ddm.ptp_classes import AttentionReplace,AttentionRefine,AttentionStore
from models.edit_friendly_ddm.ptp_utils import register_attention_control
def mask_decode(encoded_mask,image_shape=[512,512]):
length=image_shape[0]*image_shape[1]
mask_array=np.zeros((length,))
for i in range(0,len(encoded_mask),2):
splice_len=min(encoded_mask[i+1],length-encoded_mask[i])
for j in range(splice_len):
mask_array[encoded_mask[i]+j]=1
mask_array=mask_array.reshape(image_shape[0], image_shape[1])
# to avoid annotation errors in boundary
mask_array[0,:]=1
mask_array[-1,:]=1
mask_array[:,0]=1
mask_array[:,-1]=1
return mask_array
def setup_seed(seed=1234):
torch.manual_seed(seed)
torch.cuda.manual_seed_all(seed)
np.random.seed(seed)
random.seed(seed)
torch.backends.cudnn.deterministic = True
torch.backends.cudnn.benchmark = False
image_save_paths={
"edit-friendly-inversion+p2p":"edit-friendly-inversion+p2p",
}
device = torch.device('cuda') if torch.cuda.is_available() else torch.device(
'cpu')
NUM_DDIM_STEPS = 50
model_id="CompVis/stable-diffusion-v1-4"
ldm_stable = StableDiffusionPipeline.from_pretrained(
model_id).to(device)
ldm_stable.scheduler = DDIMScheduler.from_config(model_id, subfolder = "scheduler")
ldm_stable.scheduler.set_timesteps(NUM_DDIM_STEPS)
ETA=1
SKIP=12
def edit_image_EF(edit_method,
image_path,
prompt_src,
prompt_tar,
source_guidance_scale=1,
target_guidance_scale=7.5,cross_replace_steps=0.4,
self_replace_steps=0.6
):
if edit_method=="edit-friendly-inversion+p2p":
image_gt = load_512(image_path)
image_gt = torch.from_numpy(image_gt).float() / 127.5 - 1
image_gt = image_gt.permute(2, 0, 1).unsqueeze(0).to(device)
with autocast("cuda"), inference_mode():
w0 = (ldm_stable.vae.encode(image_gt).latent_dist.mode() * 0.18215).float()
controller = AttentionStore()
register_attention_control(ldm_stable, controller)
wt, zs, wts = inversion_forward_process(ldm_stable, w0, etas=ETA, prompt=prompt_src, cfg_scale=source_guidance_scale, prog_bar=True, num_inference_steps=NUM_DDIM_STEPS)
controller = AttentionStore()
register_attention_control(ldm_stable, controller)
x0_reconstruct, _ = inversion_reverse_process(ldm_stable, xT=wts[NUM_DDIM_STEPS-SKIP], etas=ETA, prompts=[prompt_tar], cfg_scales=[target_guidance_scale], prog_bar=True, zs=zs[:(NUM_DDIM_STEPS-SKIP)], controller=controller)
cfg_scale_list = [source_guidance_scale, target_guidance_scale]
prompts = [prompt_src, prompt_tar]
if (len(prompt_src.split(" ")) == len(prompt_tar.split(" "))):
controller = AttentionReplace(prompts, NUM_DDIM_STEPS, cross_replace_steps=cross_replace_steps, self_replace_steps=self_replace_steps, model=ldm_stable)
else:
# Should use Refine for target prompts with different number of tokens
controller = AttentionRefine(prompts, NUM_DDIM_STEPS, cross_replace_steps=cross_replace_steps, self_replace_steps=self_replace_steps, model=ldm_stable)
register_attention_control(ldm_stable, controller)
w0, _ = inversion_reverse_process(ldm_stable, xT=wts[NUM_DDIM_STEPS-SKIP], etas=ETA, prompts=prompts, cfg_scales=cfg_scale_list, prog_bar=True, zs=zs[:(NUM_DDIM_STEPS-SKIP)], controller=controller)
with autocast("cuda"), inference_mode():
x0_dec = ldm_stable.vae.decode(1 / 0.18215 * w0[1].unsqueeze(0)).sample
x0_reconstruct_edit = ldm_stable.vae.decode(1 / 0.18215 * w0[0].unsqueeze(0)).sample
x0_reconstruct = ldm_stable.vae.decode(1 / 0.18215 * x0_reconstruct[0].unsqueeze(0)).sample
image_instruct = txt_draw(f"source prompt: {prompt_src}\ntarget prompt: {prompt_tar}")
return Image.fromarray(np.concatenate(
(
image_instruct,
np.uint8((np.array(image_gt[0].permute(1,2,0).cpu().detach())/2+ 0.5)*255),
np.uint8((np.array(x0_reconstruct_edit[0].permute(1,2,0).cpu().detach())/2+ 0.5)*255),
np.uint8((np.array(x0_dec[0].permute(1,2,0).cpu().detach())/2+ 0.5)*255)
),
1
)
)
else:
raise NotImplementedError(f"No edit method named {edit_method}")
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument('--rerun_exist_images', action= "store_true") # rerun existing images
parser.add_argument('--data_path', type=str, default="data") # the editing category that needed to run
parser.add_argument('--output_path', type=str, default="output") # the editing category that needed to run
parser.add_argument('--edit_category_list', nargs = '+', type=str, default=["0","1","2","3","4","5","6","7","8","9"]) # the editing category that needed to run
parser.add_argument('--edit_method_list', nargs = '+', type=str, default=["edit-friendly-inversion+p2p"]) # the editing methods that needed to run
args = parser.parse_args()
rerun_exist_images=args.rerun_exist_images
data_path=args.data_path
output_path=args.output_path
edit_category_list=args.edit_category_list
edit_method_list=args.edit_method_list
with open(f"{data_path}/mapping_file.json", "r") as f:
editing_instruction = json.load(f)
for key, item in editing_instruction.items():
if item["editing_type_id"] not in edit_category_list:
continue
original_prompt = item["original_prompt"].replace("[", "").replace("]", "")
editing_prompt = item["editing_prompt"].replace("[", "").replace("]", "")
image_path = os.path.join(f"{data_path}/annotation_images", item["image_path"])
editing_instruction = item["editing_instruction"]
blended_word = item["blended_word"].split(" ") if item["blended_word"] != "" else []
mask = Image.fromarray(np.uint8(mask_decode(item["mask"])[:,:,np.newaxis].repeat(3,2))).convert("L")
for edit_method in edit_method_list:
present_image_save_path=image_path.replace(data_path, os.path.join(output_path,image_save_paths[edit_method]))
if ((not os.path.exists(present_image_save_path)) or rerun_exist_images):
print(f"editing image [{image_path}] with [{edit_method}]")
setup_seed()
torch.cuda.empty_cache()
edited_image = edit_image_EF(
edit_method=edit_method,
image_path=image_path,
prompt_src=original_prompt,
prompt_tar=editing_prompt,
source_guidance_scale=1,
target_guidance_scale=7.5,
cross_replace_steps=0.4,
self_replace_steps=0.6
)
if not os.path.exists(os.path.dirname(present_image_save_path)):
os.makedirs(os.path.dirname(present_image_save_path))
edited_image.save(present_image_save_path)
print(f"finish")
else:
print(f"skip image [{image_path}] with [{edit_method}]")