Spaces:
Running
on
Zero
Running
on
Zero
Update app.py
Browse files
app.py
CHANGED
@@ -15,44 +15,44 @@ from diffusers.utils import load_image
|
|
15 |
from pipeline_flux_control_removal import FluxControlRemovalPipeline
|
16 |
pipe = None
|
17 |
torch.set_grad_enabled(False)
|
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 |
base_model_path = 'black-forest-labs/FLUX.1-dev'
|
52 |
lora_path = 'theSure/Omnieraser'
|
53 |
transformer = FluxTransformer2DModel.from_pretrained(base_model_path, subfolder='transformer', torch_dtype=torch.bfloat16)
|
54 |
gr.Info(str(f"Model loading: {int((40 / 100) * 100)}%"))
|
55 |
-
# enable image inputs
|
56 |
with torch.no_grad():
|
57 |
initial_input_channels = transformer.config.in_channels
|
58 |
new_linear = torch.nn.Linear(
|
@@ -90,64 +90,48 @@ def set_seed(seed):
|
|
90 |
@spaces.GPU
|
91 |
def predict(
|
92 |
input_image,
|
|
|
93 |
prompt,
|
94 |
ddim_steps,
|
95 |
seed,
|
96 |
scale,
|
97 |
-
image_paths,
|
98 |
-
mask_paths
|
99 |
|
100 |
):
|
101 |
-
|
102 |
-
|
103 |
-
|
104 |
-
input_image["background"] = load_image(image_paths).convert("RGB")
|
105 |
-
input_image["layers"][0] = load_image(mask_paths).convert("RGB")
|
106 |
-
|
107 |
-
size1, size2 = input_image["background"].convert("RGB").size
|
108 |
-
icc_profile = input_image["background"].info.get('icc_profile')
|
109 |
if icc_profile:
|
110 |
gr.Info(str(f"Image detected to contain ICC profile, converting color space to sRGB..."))
|
111 |
srgb_profile = ImageCms.createProfile("sRGB")
|
112 |
io_handle = io.BytesIO(icc_profile)
|
113 |
src_profile = ImageCms.ImageCmsProfile(io_handle)
|
114 |
-
input_image
|
115 |
-
input_image
|
116 |
|
117 |
if size1 < size2:
|
118 |
-
input_image
|
119 |
else:
|
120 |
-
input_image
|
121 |
|
122 |
-
img = np.array(input_image
|
123 |
|
124 |
W = int(np.shape(img)[0] - np.shape(img)[0] % 8)
|
125 |
H = int(np.shape(img)[1] - np.shape(img)[1] % 8)
|
126 |
|
127 |
-
input_image
|
128 |
-
|
129 |
|
130 |
if seed == -1:
|
131 |
seed = random.randint(1, 2147483647)
|
132 |
set_seed(random.randint(1, 2147483647))
|
133 |
else:
|
134 |
set_seed(seed)
|
135 |
-
if image_paths is None:
|
136 |
-
img=input_image["layers"][0]
|
137 |
-
img_data = np.array(img)
|
138 |
-
alpha_channel = img_data[:, :, 3]
|
139 |
-
white_background = np.ones_like(alpha_channel) * 255
|
140 |
-
gray_image = white_background.copy()
|
141 |
-
gray_image[alpha_channel == 0] = 0
|
142 |
-
gray_image_pil = Image.fromarray(gray_image).convert('L')
|
143 |
-
else:
|
144 |
-
gray_image_pil = input_image["layers"][0]
|
145 |
base_model_path = 'black-forest-labs/FLUX.1-dev'
|
146 |
lora_path = 'theSure/Omnieraser'
|
147 |
result = pipe(
|
148 |
prompt=prompt,
|
149 |
-
control_image=input_image
|
150 |
-
control_mask=
|
151 |
width=H,
|
152 |
height=W,
|
153 |
num_inference_steps=ddim_steps,
|
@@ -156,19 +140,19 @@ def predict(
|
|
156 |
max_sequence_length=512,
|
157 |
).images[0]
|
158 |
|
159 |
-
mask_np = np.array(
|
160 |
-
red = np.array(input_image
|
161 |
red[:, :, 0] = 180.0
|
162 |
red[:, :, 2] = 0
|
163 |
red[:, :, 1] = 0
|
164 |
-
result_m = np.array(input_image
|
165 |
result_m = Image.fromarray(
|
166 |
(
|
167 |
result_m.astype("float") * (1 - mask_np.astype("float") / 512.0) + mask_np.astype("float") / 512.0 * red
|
168 |
).astype("uint8")
|
169 |
)
|
170 |
|
171 |
-
dict_res = [input_image
|
172 |
|
173 |
dict_out = [result]
|
174 |
image_path = None
|
@@ -178,21 +162,19 @@ def predict(
|
|
178 |
|
179 |
def infer(
|
180 |
input_image,
|
|
|
181 |
ddim_steps,
|
182 |
seed,
|
183 |
scale,
|
184 |
removal_prompt,
|
185 |
|
186 |
):
|
187 |
-
img_path = image_path
|
188 |
-
msk_path = mask_path
|
189 |
return predict(input_image,
|
|
|
190 |
removal_prompt,
|
191 |
ddim_steps,
|
192 |
seed,
|
193 |
scale,
|
194 |
-
img_path,
|
195 |
-
msk_path
|
196 |
)
|
197 |
|
198 |
def process_example(image_paths, mask_paths):
|
@@ -288,13 +270,8 @@ with gr.Blocks(
|
|
288 |
gr.Markdown("## 📥 Input Panel")
|
289 |
|
290 |
with gr.Group():
|
291 |
-
|
292 |
-
|
293 |
-
type="pil",
|
294 |
-
label="Upload & Annotate",
|
295 |
-
elem_id="custom-image",
|
296 |
-
interactive=True
|
297 |
-
)
|
298 |
with gr.Row(variant="compact"):
|
299 |
run_button = gr.Button(
|
300 |
"🚀 Start Processing",
|
@@ -311,21 +288,18 @@ with gr.Blocks(
|
|
311 |
step=1,
|
312 |
info="-1 for random generation"
|
313 |
)
|
314 |
-
|
315 |
-
|
316 |
-
|
317 |
-
|
318 |
-
|
319 |
-
|
320 |
-
|
321 |
-
|
322 |
-
|
323 |
-
|
324 |
-
|
325 |
-
|
326 |
-
# label="Click any example to load",
|
327 |
-
# elem_id="inline-examples"
|
328 |
-
# )
|
329 |
|
330 |
with gr.Column(scale=1, variant="panel"):
|
331 |
gr.Markdown("## 📤 Output Panel")
|
@@ -351,6 +325,7 @@ with gr.Blocks(
|
|
351 |
fn=infer,
|
352 |
inputs=[
|
353 |
input_image,
|
|
|
354 |
ddim_steps,
|
355 |
seed,
|
356 |
scale,
|
|
|
15 |
from pipeline_flux_control_removal import FluxControlRemovalPipeline
|
16 |
pipe = None
|
17 |
torch.set_grad_enabled(False)
|
18 |
+
|
19 |
+
image_examples = [
|
20 |
+
[
|
21 |
+
"example/image/3c43156c-2b44-4ebf-9c47-7707ec60b166.png",
|
22 |
+
"example/mask/3c43156c-2b44-4ebf-9c47-7707ec60b166.png"
|
23 |
+
],
|
24 |
+
[
|
25 |
+
"example/image/0e5124d8-fe43-4b5c-819f-7212f23a6d2a.png",
|
26 |
+
"example/mask/0e5124d8-fe43-4b5c-819f-7212f23a6d2a.png"
|
27 |
+
],
|
28 |
+
[
|
29 |
+
"example/image/0f900fe8-6eab-4f85-8121-29cac9509b94.png",
|
30 |
+
"example/mask/0f900fe8-6eab-4f85-8121-29cac9509b94.png"
|
31 |
+
],
|
32 |
+
[
|
33 |
+
"example/image/3ed1ee18-33b0-4964-b679-0e214a0d8848.png",
|
34 |
+
"example/mask/3ed1ee18-33b0-4964-b679-0e214a0d8848.png"
|
35 |
+
],
|
36 |
+
[
|
37 |
+
"example/image/9a3b6af9-c733-46a4-88d4-d77604194102.png",
|
38 |
+
"example/mask/9a3b6af9-c733-46a4-88d4-d77604194102.png"
|
39 |
+
],
|
40 |
+
[
|
41 |
+
"example/image/87cdf3e2-0fa1-4d80-a228-cbb4aba3f44f.png",
|
42 |
+
"example/mask/87cdf3e2-0fa1-4d80-a228-cbb4aba3f44f.png"
|
43 |
+
],
|
44 |
+
[
|
45 |
+
"example/image/55dd199b-d99b-47a2-a691-edfd92233a6b.png",
|
46 |
+
"example/mask/55dd199b-d99b-47a2-a691-edfd92233a6b.png"
|
47 |
+
]
|
48 |
|
49 |
+
]
|
50 |
+
|
51 |
|
52 |
base_model_path = 'black-forest-labs/FLUX.1-dev'
|
53 |
lora_path = 'theSure/Omnieraser'
|
54 |
transformer = FluxTransformer2DModel.from_pretrained(base_model_path, subfolder='transformer', torch_dtype=torch.bfloat16)
|
55 |
gr.Info(str(f"Model loading: {int((40 / 100) * 100)}%"))
|
|
|
56 |
with torch.no_grad():
|
57 |
initial_input_channels = transformer.config.in_channels
|
58 |
new_linear = torch.nn.Linear(
|
|
|
90 |
@spaces.GPU
|
91 |
def predict(
|
92 |
input_image,
|
93 |
+
uploaded_mask
|
94 |
prompt,
|
95 |
ddim_steps,
|
96 |
seed,
|
97 |
scale,
|
|
|
|
|
98 |
|
99 |
):
|
100 |
+
gr.Info(str(f"Set seed = {seed}"))
|
101 |
+
size1, size2 = input_image.convert("RGB").size
|
102 |
+
icc_profile = input_image.info.get('icc_profile')
|
|
|
|
|
|
|
|
|
|
|
103 |
if icc_profile:
|
104 |
gr.Info(str(f"Image detected to contain ICC profile, converting color space to sRGB..."))
|
105 |
srgb_profile = ImageCms.createProfile("sRGB")
|
106 |
io_handle = io.BytesIO(icc_profile)
|
107 |
src_profile = ImageCms.ImageCmsProfile(io_handle)
|
108 |
+
input_image = ImageCms.profileToProfile(input_image["background"], src_profile, srgb_profile)
|
109 |
+
input_image.info.pop('icc_profile', None)
|
110 |
|
111 |
if size1 < size2:
|
112 |
+
input_image = input_image.convert("RGB").resize((1024, int(size2 / size1 * 1024)))
|
113 |
else:
|
114 |
+
input_image = input_image.convert("RGB").resize((int(size1 / size2 * 1024), 1024))
|
115 |
|
116 |
+
img = np.array(input_image.convert("RGB"))
|
117 |
|
118 |
W = int(np.shape(img)[0] - np.shape(img)[0] % 8)
|
119 |
H = int(np.shape(img)[1] - np.shape(img)[1] % 8)
|
120 |
|
121 |
+
input_image = input_imageresize((H, W))
|
122 |
+
uploaded_mask = uploaded_mask.resize((H, W))
|
123 |
|
124 |
if seed == -1:
|
125 |
seed = random.randint(1, 2147483647)
|
126 |
set_seed(random.randint(1, 2147483647))
|
127 |
else:
|
128 |
set_seed(seed)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
129 |
base_model_path = 'black-forest-labs/FLUX.1-dev'
|
130 |
lora_path = 'theSure/Omnieraser'
|
131 |
result = pipe(
|
132 |
prompt=prompt,
|
133 |
+
control_image=input_image.convert("RGB"),
|
134 |
+
control_mask=uploaded_mask.convert("RGB"),
|
135 |
width=H,
|
136 |
height=W,
|
137 |
num_inference_steps=ddim_steps,
|
|
|
140 |
max_sequence_length=512,
|
141 |
).images[0]
|
142 |
|
143 |
+
mask_np = np.array(uploaded_mask.convert("RGB"))
|
144 |
+
red = np.array(input_image).astype("float") * 1
|
145 |
red[:, :, 0] = 180.0
|
146 |
red[:, :, 2] = 0
|
147 |
red[:, :, 1] = 0
|
148 |
+
result_m = np.array(input_image)
|
149 |
result_m = Image.fromarray(
|
150 |
(
|
151 |
result_m.astype("float") * (1 - mask_np.astype("float") / 512.0) + mask_np.astype("float") / 512.0 * red
|
152 |
).astype("uint8")
|
153 |
)
|
154 |
|
155 |
+
dict_res = [input_image, uploaded_mask, result_m, result]
|
156 |
|
157 |
dict_out = [result]
|
158 |
image_path = None
|
|
|
162 |
|
163 |
def infer(
|
164 |
input_image,
|
165 |
+
uploaded_mask,
|
166 |
ddim_steps,
|
167 |
seed,
|
168 |
scale,
|
169 |
removal_prompt,
|
170 |
|
171 |
):
|
|
|
|
|
172 |
return predict(input_image,
|
173 |
+
uploaded_mask
|
174 |
removal_prompt,
|
175 |
ddim_steps,
|
176 |
seed,
|
177 |
scale,
|
|
|
|
|
178 |
)
|
179 |
|
180 |
def process_example(image_paths, mask_paths):
|
|
|
270 |
gr.Markdown("## 📥 Input Panel")
|
271 |
|
272 |
with gr.Group():
|
273 |
+
image_input = gr.Image(label="Upload Image", type="pil", image_mode="RGB")
|
274 |
+
uploaded_mask = gr.Image(label="Upload Mask", type="pil", image_mode="L")
|
|
|
|
|
|
|
|
|
|
|
275 |
with gr.Row(variant="compact"):
|
276 |
run_button = gr.Button(
|
277 |
"🚀 Start Processing",
|
|
|
288 |
step=1,
|
289 |
info="-1 for random generation"
|
290 |
)
|
291 |
+
with gr.Column(variant="panel"):
|
292 |
+
gr.Markdown("### 🖼️ Example Gallery", elem_classes=["example-title"])
|
293 |
+
example = gr.Examples(
|
294 |
+
examples=image_examples,
|
295 |
+
inputs=[
|
296 |
+
image_input, uploaded_mask
|
297 |
+
],
|
298 |
+
outputs=[inpaint_result, gallery],
|
299 |
+
examples_per_page=10,
|
300 |
+
label="Click any example to load",
|
301 |
+
elem_id="inline-examples"
|
302 |
+
)
|
|
|
|
|
|
|
303 |
|
304 |
with gr.Column(scale=1, variant="panel"):
|
305 |
gr.Markdown("## 📤 Output Panel")
|
|
|
325 |
fn=infer,
|
326 |
inputs=[
|
327 |
input_image,
|
328 |
+
uploaded_mask
|
329 |
ddim_steps,
|
330 |
seed,
|
331 |
scale,
|