yash
commited on
Commit
·
545d518
1
Parent(s):
b8b2322
first commit
Browse files- app.py +173 -0
- requirements.txt +104 -0
app.py
ADDED
@@ -0,0 +1,173 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import torch
|
2 |
+
import gradio as gr
|
3 |
+
from diffusers import StableDiffusionControlNetPipeline
|
4 |
+
from diffusers import ControlNetModel, DDIMScheduler,EulerDiscreteScheduler,EulerAncestralDiscreteScheduler
|
5 |
+
from diffusers import KDPM2DiscreteScheduler,KDPM2AncestralDiscreteScheduler,PNDMScheduler,UniPCMultistepScheduler
|
6 |
+
from diffusers import DPMSolverMultistepScheduler
|
7 |
+
import random
|
8 |
+
import numpy as np
|
9 |
+
import cv2
|
10 |
+
from PIL import Image
|
11 |
+
from diffusers.utils import load_image
|
12 |
+
|
13 |
+
|
14 |
+
|
15 |
+
def canny_image(image,th1=100,th2=200):
|
16 |
+
image = np.array(image)
|
17 |
+
image = cv2.Canny(image,th1,th2)
|
18 |
+
image = image[:, :, None]
|
19 |
+
image = np.concatenate([image, image, image], axis=2)
|
20 |
+
canny_image = Image.fromarray(image)
|
21 |
+
return canny_image
|
22 |
+
|
23 |
+
|
24 |
+
def set_pipeline(model_id_repo,scheduler):
|
25 |
+
|
26 |
+
model_ids_dict = {
|
27 |
+
"runwayml": "runwayml/stable-diffusion-v1-5",
|
28 |
+
"Realistic_Vision_V5_1_noVAE":"SG161222/Realistic_Vision_V5.1_noVAE"
|
29 |
+
}
|
30 |
+
model_id = model_id_repo
|
31 |
+
model_repo = model_ids_dict.get(model_id)
|
32 |
+
print("model_repo :",model_repo)
|
33 |
+
|
34 |
+
|
35 |
+
|
36 |
+
# load control net and stable diffusion v1-5
|
37 |
+
controlnet = ControlNetModel.from_pretrained("lllyasviel/control_v11p_sd15_canny",
|
38 |
+
# torch_dtype=torch.float16
|
39 |
+
)
|
40 |
+
|
41 |
+
pipe = StableDiffusionControlNetPipeline.from_pretrained(
|
42 |
+
model_repo,
|
43 |
+
controlnet=controlnet,
|
44 |
+
# torch_dtype=torch.float16,
|
45 |
+
use_safetensors = True
|
46 |
+
).to("cpu")
|
47 |
+
|
48 |
+
|
49 |
+
|
50 |
+
|
51 |
+
scheduler_classes = {
|
52 |
+
"DDIM": DDIMScheduler,
|
53 |
+
"Euler": EulerDiscreteScheduler,
|
54 |
+
"Euler a": EulerAncestralDiscreteScheduler,
|
55 |
+
"UniPC": UniPCMultistepScheduler,
|
56 |
+
"DPM2 Karras": KDPM2DiscreteScheduler,
|
57 |
+
"DPM2 a Karras": KDPM2AncestralDiscreteScheduler,
|
58 |
+
"PNDM": PNDMScheduler,
|
59 |
+
"DPM++ 2M Karras": DPMSolverMultistepScheduler,
|
60 |
+
"DPM++ 2M SDE Karras": DPMSolverMultistepScheduler,
|
61 |
+
}
|
62 |
+
|
63 |
+
sampler_name = scheduler # Example sampler name, replace with the actual value
|
64 |
+
scheduler_class = scheduler_classes.get(sampler_name)
|
65 |
+
|
66 |
+
if scheduler_class is not None:
|
67 |
+
print("sampler_name:",sampler_name)
|
68 |
+
pipe.scheduler = scheduler_class.from_config(pipe.scheduler.config)
|
69 |
+
else:
|
70 |
+
pass
|
71 |
+
|
72 |
+
return pipe
|
73 |
+
|
74 |
+
|
75 |
+
def img_args(
|
76 |
+
prompt,
|
77 |
+
negative_prompt,
|
78 |
+
image_canny,
|
79 |
+
controlnet_conditioning_scale = 1.0,
|
80 |
+
control_guidance_start=0.0,
|
81 |
+
control_guidance_end=1.0,
|
82 |
+
clip_skip=0,
|
83 |
+
model_id_repo = "Realistic_Vision_V5_1_noVAE",
|
84 |
+
scheduler= "Euler a",
|
85 |
+
num_inference_steps = 30,
|
86 |
+
guidance_scale = 7.5,
|
87 |
+
num_images_per_prompt = 1,
|
88 |
+
seed = 0
|
89 |
+
):
|
90 |
+
|
91 |
+
controlnet_conditioning_scale = float(controlnet_conditioning_scale)
|
92 |
+
|
93 |
+
if image_canny is None:
|
94 |
+
return
|
95 |
+
|
96 |
+
pipe = set_pipeline(model_id_repo,scheduler)
|
97 |
+
|
98 |
+
if seed == -1:
|
99 |
+
seed = random.randint(0,2564798154)
|
100 |
+
print(f"random seed :{seed}")
|
101 |
+
generator = torch.manual_seed(seed)
|
102 |
+
else:
|
103 |
+
generator = torch.manual_seed(seed)
|
104 |
+
print(f"manual seed :{seed}")
|
105 |
+
|
106 |
+
print("Prompt:",prompt)
|
107 |
+
image = pipe(prompt=prompt,
|
108 |
+
negative_prompt = negative_prompt,
|
109 |
+
image=image_canny,
|
110 |
+
control_guidance_start = control_guidance_start,
|
111 |
+
control_guidance_end = control_guidance_end,
|
112 |
+
clip_skip =clip_skip,
|
113 |
+
num_inference_steps = num_inference_steps,
|
114 |
+
guidance_scale = guidance_scale,
|
115 |
+
num_images_per_prompt = num_images_per_prompt, # default 1
|
116 |
+
generator = generator,
|
117 |
+
controlnet_conditioning_scale = controlnet_conditioning_scale
|
118 |
+
).images
|
119 |
+
return image
|
120 |
+
|
121 |
+
|
122 |
+
block = gr.Blocks().queue()
|
123 |
+
block.title = "Inpaint Anything"
|
124 |
+
with block as image_gen:
|
125 |
+
with gr.Column():
|
126 |
+
with gr.Row():
|
127 |
+
gr.Markdown("## Image Generation With Canny Controlnet")
|
128 |
+
with gr.Row():
|
129 |
+
with gr.Column():
|
130 |
+
# with gr.Row():
|
131 |
+
input_image = gr.Image(type="pil",label="Input")
|
132 |
+
prompt = gr.Textbox(placeholder="what you want to generate",label="Positive Prompt")
|
133 |
+
negative_prompt = gr.Textbox(placeholder="what you don't want to generate",label="Negative prompt")
|
134 |
+
with gr.Column():
|
135 |
+
canny_output = gr.Image(type="pil",label="Canny Input")
|
136 |
+
canny_btn = gr.Button("Canny Image", elem_id="select_btn", variant="primary")
|
137 |
+
|
138 |
+
with gr.Accordion(label="Controlnet Advance Options",open=False):
|
139 |
+
controlnet_conditioning_scale_slider = gr.Slider(label="Control Condition Scale", minimum=0.0, maximum=2.0, value=1.0, step=0.05)
|
140 |
+
control_guidance_start_slider = gr.Slider(label="Contron Guidance Start", minimum=0.0, maximum=1.0, value=0, step=0.1)
|
141 |
+
control_guidance_end_slider = gr.Slider(label="Contron Guidance Start End", minimum=0.0, maximum=1.0, value=1, step=0.1)
|
142 |
+
canny_th1 = gr.Slider(label="Canny High Threshold",minimum=0, maximum=300, value=100, step=1)
|
143 |
+
canny_th2 = gr.Slider(label="Canny Low Threshold",minimum=0, maximum=300, value=200, step=1)
|
144 |
+
with gr.Column():
|
145 |
+
out_img = gr.Gallery(label='Output', show_label=True, elem_id="gallery", preview=True)
|
146 |
+
run_btn = gr.Button("Generation", elem_id="select_btn", variant="primary")
|
147 |
+
|
148 |
+
with gr.Accordion(label="Generation Advance Options",open=False):
|
149 |
+
with gr.Row():
|
150 |
+
model_selection = gr.Dropdown(choices=["runwayml","Realistic_Vision_V5_1_noVAE"],value="Realistic_Vision_V5_1_noVAE",label="Models")
|
151 |
+
schduler_selection = gr.Dropdown(choices=["DDIM","Euler","Euler a","UniPC","DPM2 Karras","DPM2 a Karras","PNDM","DPM++ 2M Karras","DPM++ 2M SDE Karras"],value="Euler a",label="Scheduler")
|
152 |
+
guidance_scale_slider = gr.Slider(label="Guidance Scale", minimum=0, maximum=15, value=7.5, step=0.5)
|
153 |
+
num_images_per_prompt_slider = gr.Slider(label="num_images_per_prompt", minimum=0, maximum=5, value=1, step=1)
|
154 |
+
num_inference_steps_slider = gr.Slider(label="num_inference_steps", minimum=0, maximum=150, value=30, step=1)
|
155 |
+
seed_slider = gr.Slider(label="Seed", minimum=-1, maximum=256479815, value=-1, step=1)
|
156 |
+
clip_skip_slider = gr.Slider(label="Clip Skip", minimum=0, maximum=3, value=0, step=1)
|
157 |
+
|
158 |
+
canny_btn.click(fn=canny_image,inputs=[input_image,canny_th1,canny_th2],outputs=[canny_output])
|
159 |
+
run_btn.click(fn=img_args,inputs=[prompt,
|
160 |
+
negative_prompt,
|
161 |
+
canny_output,
|
162 |
+
controlnet_conditioning_scale_slider,
|
163 |
+
control_guidance_start_slider,
|
164 |
+
control_guidance_end_slider,
|
165 |
+
clip_skip_slider,
|
166 |
+
model_selection,
|
167 |
+
schduler_selection,
|
168 |
+
num_inference_steps_slider,
|
169 |
+
guidance_scale_slider,
|
170 |
+
num_images_per_prompt_slider,
|
171 |
+
seed_slider], outputs=[out_img])
|
172 |
+
image_gen.launch()
|
173 |
+
|
requirements.txt
ADDED
@@ -0,0 +1,104 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
accelerate==0.30.1
|
2 |
+
aiofiles==23.2.1
|
3 |
+
altair==5.3.0
|
4 |
+
annotated-types==0.6.0
|
5 |
+
anyio==4.3.0
|
6 |
+
attrs==23.2.0
|
7 |
+
certifi==2024.2.2
|
8 |
+
charset-normalizer==3.3.2
|
9 |
+
click==8.1.7
|
10 |
+
contourpy==1.2.1
|
11 |
+
cycler==0.12.1
|
12 |
+
diffusers==0.27.2
|
13 |
+
dnspython==2.6.1
|
14 |
+
email_validator==2.1.1
|
15 |
+
exceptiongroup==1.2.1
|
16 |
+
fastapi==0.111.0
|
17 |
+
fastapi-cli==0.0.3
|
18 |
+
ffmpy==0.3.2
|
19 |
+
filelock==3.14.0
|
20 |
+
fonttools==4.51.0
|
21 |
+
fsspec==2024.3.1
|
22 |
+
gradio==3.50.2
|
23 |
+
gradio_client==0.6.1
|
24 |
+
h11==0.14.0
|
25 |
+
httpcore==1.0.5
|
26 |
+
httptools==0.6.1
|
27 |
+
httpx==0.27.0
|
28 |
+
huggingface-hub==0.23.0
|
29 |
+
idna==3.7
|
30 |
+
importlib_metadata==7.1.0
|
31 |
+
importlib_resources==6.4.0
|
32 |
+
inquirerpy==0.3.4
|
33 |
+
Jinja2==3.1.4
|
34 |
+
jsonschema==4.22.0
|
35 |
+
jsonschema-specifications==2023.12.1
|
36 |
+
kiwisolver==1.4.5
|
37 |
+
markdown-it-py==3.0.0
|
38 |
+
MarkupSafe==2.1.5
|
39 |
+
matplotlib==3.8.4
|
40 |
+
mdurl==0.1.2
|
41 |
+
mpmath==1.3.0
|
42 |
+
networkx==3.3
|
43 |
+
numpy==1.26.4
|
44 |
+
nvidia-cublas-cu12==12.1.3.1
|
45 |
+
nvidia-cuda-cupti-cu12==12.1.105
|
46 |
+
nvidia-cuda-nvrtc-cu12==12.1.105
|
47 |
+
nvidia-cuda-runtime-cu12==12.1.105
|
48 |
+
nvidia-cudnn-cu12==8.9.2.26
|
49 |
+
nvidia-cufft-cu12==11.0.2.54
|
50 |
+
nvidia-curand-cu12==10.3.2.106
|
51 |
+
nvidia-cusolver-cu12==11.4.5.107
|
52 |
+
nvidia-cusparse-cu12==12.1.0.106
|
53 |
+
nvidia-nccl-cu12==2.20.5
|
54 |
+
nvidia-nvjitlink-cu12==12.4.127
|
55 |
+
nvidia-nvtx-cu12==12.1.105
|
56 |
+
opencv-python==4.9.0.80
|
57 |
+
orjson==3.10.3
|
58 |
+
packaging==24.0
|
59 |
+
pandas==2.2.2
|
60 |
+
pfzy==0.3.4
|
61 |
+
pillow==10.3.0
|
62 |
+
prompt-toolkit==3.0.43
|
63 |
+
psutil==5.9.8
|
64 |
+
pydantic==2.7.1
|
65 |
+
pydantic_core==2.18.2
|
66 |
+
pydub==0.25.1
|
67 |
+
Pygments==2.18.0
|
68 |
+
pyparsing==3.1.2
|
69 |
+
python-dateutil==2.9.0.post0
|
70 |
+
python-dotenv==1.0.1
|
71 |
+
python-multipart==0.0.9
|
72 |
+
pytz==2024.1
|
73 |
+
PyYAML==6.0.1
|
74 |
+
referencing==0.35.1
|
75 |
+
regex==2024.5.10
|
76 |
+
requests==2.31.0
|
77 |
+
rich==13.7.1
|
78 |
+
rpds-py==0.18.1
|
79 |
+
safetensors==0.4.3
|
80 |
+
semantic-version==2.10.0
|
81 |
+
shellingham==1.5.4
|
82 |
+
six==1.16.0
|
83 |
+
sniffio==1.3.1
|
84 |
+
starlette==0.37.2
|
85 |
+
sympy==1.12
|
86 |
+
tokenizers==0.19.1
|
87 |
+
toolz==0.12.1
|
88 |
+
torch==2.3.0
|
89 |
+
torchvision==0.18.0
|
90 |
+
tqdm==4.66.4
|
91 |
+
transformers==4.40.2
|
92 |
+
triton==2.3.0
|
93 |
+
typer==0.12.3
|
94 |
+
typing_extensions==4.11.0
|
95 |
+
tzdata==2024.1
|
96 |
+
ujson==5.10.0
|
97 |
+
urllib3==2.2.1
|
98 |
+
uvicorn==0.29.0
|
99 |
+
uvloop==0.19.0
|
100 |
+
watchfiles==0.21.0
|
101 |
+
wcwidth==0.2.13
|
102 |
+
websockets==11.0.3
|
103 |
+
xformers==0.0.26.post1
|
104 |
+
zipp==3.18.1
|