Spaces:
Running
on
Zero
Running
on
Zero
support donut
Browse files
app.py
CHANGED
@@ -3,7 +3,7 @@ import torch
|
|
3 |
from diffusers import FluxInpaintPipeline
|
4 |
import gradio as gr
|
5 |
import re
|
6 |
-
from PIL import Image
|
7 |
|
8 |
import os
|
9 |
import numpy as np
|
@@ -28,6 +28,16 @@ def adjust_to_multiple_of_32(width: int, height: int):
|
|
28 |
height = height - (height % 32)
|
29 |
return width, height
|
30 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
31 |
|
32 |
|
33 |
dtype = torch.bfloat16
|
@@ -47,7 +57,7 @@ def sanitize_prompt(prompt):
|
|
47 |
return sanitized_prompt
|
48 |
|
49 |
@spaces.GPU(duration=120)
|
50 |
-
def process_images(image, image2=None,prompt="a girl",inpaint_model="black-forest-labs/FLUX.1-schnell",strength=0.75,seed=0,progress=gr.Progress(track_tqdm=True)):
|
51 |
# I'm not sure when this happen
|
52 |
progress(0, desc="start-process-images")
|
53 |
#print("start-process-images")
|
@@ -78,9 +88,9 @@ def process_images(image, image2=None,prompt="a girl",inpaint_model="black-fores
|
|
78 |
generator = torch.Generator("cuda").manual_seed(seed)
|
79 |
generators.append(generator)
|
80 |
|
81 |
-
|
82 |
#print(f"fit {width}x{height}")
|
83 |
-
width,height = adjust_to_multiple_of_32(
|
84 |
#print(f"multiple {width}x{height}")
|
85 |
image = image.resize((width, height), Image.LANCZOS)
|
86 |
mask_image = mask_image.resize((width, height), Image.NEAREST)
|
@@ -89,11 +99,24 @@ def process_images(image, image2=None,prompt="a girl",inpaint_model="black-fores
|
|
89 |
output = pipe(prompt=prompt, image=image, mask_image=mask_image,generator=generator,strength=strength,width=width,height=height,
|
90 |
guidance_scale=0,num_inference_steps=num_inference_steps,max_sequence_length=256)
|
91 |
|
92 |
-
return output.images[0],mask_image
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
93 |
|
94 |
-
|
95 |
-
output,mask_image = process_inpaint(image["background"],mask,prompt,inpaint_model,strength,seed)
|
96 |
-
|
97 |
return output,mask_image
|
98 |
|
99 |
|
@@ -138,9 +161,10 @@ with gr.Blocks(css=css, elem_id="demo-container") as demo:
|
|
138 |
with gr.Column():
|
139 |
image = gr.ImageEditor(height=800,sources=['upload','clipboard'],transforms=[],image_mode='RGB', layers=False, elem_id="image_upload", type="pil", label="Upload",brush=gr.Brush(colors=["#fff"], color_mode="fixed"))
|
140 |
with gr.Row(elem_id="prompt-container", equal_height=False):
|
141 |
-
|
142 |
-
|
143 |
-
|
|
|
144 |
btn = gr.Button("Inpaint", elem_id="run_button",variant="primary")
|
145 |
|
146 |
image_mask = gr.Image(sources=['upload','clipboard'], elem_id="mask_upload", type="pil", label="Mask_Upload",height=400, value=None)
|
@@ -159,7 +183,7 @@ with gr.Blocks(css=css, elem_id="demo-container") as demo:
|
|
159 |
|
160 |
|
161 |
|
162 |
-
btn.click(fn=process_images, inputs=[image, image_mask,prompt,inpaint_model,strength,seed], outputs =[image_out,mask_out], api_name='infer')
|
163 |
gr.Examples(
|
164 |
examples=[
|
165 |
["examples/00538245.jpg", "examples/normal_mouth_mask.jpg","a beautiful girl,big-smile",0.75,"examples/normal_mouth_mask_result.jpg"],
|
|
|
3 |
from diffusers import FluxInpaintPipeline
|
4 |
import gradio as gr
|
5 |
import re
|
6 |
+
from PIL import Image,ImageFilter
|
7 |
|
8 |
import os
|
9 |
import numpy as np
|
|
|
28 |
height = height - (height % 32)
|
29 |
return width, height
|
30 |
|
31 |
+
def mask_to_donut(mask,size):
|
32 |
+
if size%2 ==0:
|
33 |
+
size+=1
|
34 |
+
dilation_mask = mask.filter(ImageFilter.MaxFilter(size))
|
35 |
+
|
36 |
+
white_img = Image.new('RGB', mask.size, (255,255,255))
|
37 |
+
black_img = Image.new('RGB', mask.size, (0,0,0))
|
38 |
+
white_img.paste(black_img,(0,0),dilation_mask.convert("L"))
|
39 |
+
white_img.paste(mask,(0,0),mask.convert("L"))
|
40 |
+
return white_img
|
41 |
|
42 |
|
43 |
dtype = torch.bfloat16
|
|
|
57 |
return sanitized_prompt
|
58 |
|
59 |
@spaces.GPU(duration=120)
|
60 |
+
def process_images(image, image2=None,prompt="a girl",inpaint_model="black-forest-labs/FLUX.1-schnell",strength=0.75,seed=0,donut_mask=True,donut_size=32,progress=gr.Progress(track_tqdm=True)):
|
61 |
# I'm not sure when this happen
|
62 |
progress(0, desc="start-process-images")
|
63 |
#print("start-process-images")
|
|
|
88 |
generator = torch.Generator("cuda").manual_seed(seed)
|
89 |
generators.append(generator)
|
90 |
|
91 |
+
fit_width,fit_height = convert_to_fit_size(image.size)
|
92 |
#print(f"fit {width}x{height}")
|
93 |
+
width,height = adjust_to_multiple_of_32(fit_width,fit_height)
|
94 |
#print(f"multiple {width}x{height}")
|
95 |
image = image.resize((width, height), Image.LANCZOS)
|
96 |
mask_image = mask_image.resize((width, height), Image.NEAREST)
|
|
|
99 |
output = pipe(prompt=prompt, image=image, mask_image=mask_image,generator=generator,strength=strength,width=width,height=height,
|
100 |
guidance_scale=0,num_inference_steps=num_inference_steps,max_sequence_length=256)
|
101 |
|
102 |
+
return output.images[0],mask_image,image,fit_width,fit_height
|
103 |
+
|
104 |
+
if donut_mask:
|
105 |
+
original_mask = mask
|
106 |
+
mask = mask_to_donut(mask,donut_size)
|
107 |
+
|
108 |
+
#output,mask_image,image_resized,fit_width,fit_height=image["background"],mask,image["background"],512,512
|
109 |
+
output,mask_image,image_resized,fit_width,fit_height = process_inpaint(image["background"],mask,prompt,inpaint_model,strength,seed)
|
110 |
+
|
111 |
+
if donut_mask:
|
112 |
+
mask = original_mask.resize(mask_image.size)
|
113 |
+
image_resized.paste(output,(0,0),mask.convert("L"))
|
114 |
+
output = image_resized.resize((fit_width,fit_height),Image.LANCZOS)
|
115 |
+
mask_image = mask.resize(output.size)
|
116 |
+
else:
|
117 |
+
output = output.resize((fit_width,fit_height),Image.LANCZOS)
|
118 |
+
mask_image = mask_image.resize(output.size)
|
119 |
|
|
|
|
|
|
|
120 |
return output,mask_image
|
121 |
|
122 |
|
|
|
161 |
with gr.Column():
|
162 |
image = gr.ImageEditor(height=800,sources=['upload','clipboard'],transforms=[],image_mode='RGB', layers=False, elem_id="image_upload", type="pil", label="Upload",brush=gr.Brush(colors=["#fff"], color_mode="fixed"))
|
163 |
with gr.Row(elem_id="prompt-container", equal_height=False):
|
164 |
+
prompt = gr.Textbox(label="Prompt",value="a person",placeholder="Your prompt (what you want in place of what is erased)", elem_id="prompt")
|
165 |
+
with gr.Row(equal_height=True):
|
166 |
+
donut_mask = gr.Checkbox(label="Donut Mask",value=False,info="Usually improve result,but slow.Do second example things")
|
167 |
+
donut_size = gr.Slider(label="Donut Size",minimum=1,maximum=64,step=1,value=32,info="Larger value make extreamly slow")
|
168 |
btn = gr.Button("Inpaint", elem_id="run_button",variant="primary")
|
169 |
|
170 |
image_mask = gr.Image(sources=['upload','clipboard'], elem_id="mask_upload", type="pil", label="Mask_Upload",height=400, value=None)
|
|
|
183 |
|
184 |
|
185 |
|
186 |
+
btn.click(fn=process_images, inputs=[image, image_mask,prompt,inpaint_model,strength,seed,donut_mask,donut_size], outputs =[image_out,mask_out], api_name='infer')
|
187 |
gr.Examples(
|
188 |
examples=[
|
189 |
["examples/00538245.jpg", "examples/normal_mouth_mask.jpg","a beautiful girl,big-smile",0.75,"examples/normal_mouth_mask_result.jpg"],
|