Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
@@ -3,14 +3,12 @@ from diffusers import ControlNetModel, StableDiffusionXLControlNetPipeline, Auto
|
|
3 |
from diffusers.utils import load_image
|
4 |
from transformers import DPTImageProcessor, DPTForDepthEstimation
|
5 |
import torch
|
6 |
-
import mediapy
|
7 |
import sa_handler
|
8 |
import pipeline_calls
|
9 |
|
10 |
|
11 |
|
12 |
-
#
|
13 |
-
|
14 |
depth_estimator = DPTForDepthEstimation.from_pretrained("Intel/dpt-hybrid-midas").to("cuda")
|
15 |
feature_processor = DPTImageProcessor.from_pretrained("Intel/dpt-hybrid-midas")
|
16 |
|
@@ -29,9 +27,11 @@ pipeline = StableDiffusionXLControlNetPipeline.from_pretrained(
|
|
29 |
use_safetensors=True,
|
30 |
torch_dtype=torch.float16,
|
31 |
).to("cuda")
|
|
|
32 |
pipeline.enable_model_cpu_offload()
|
33 |
pipeline.enable_vae_slicing()
|
34 |
|
|
|
35 |
sa_args = sa_handler.StyleAlignedArgs(share_group_norm=False,
|
36 |
share_layer_norm=False,
|
37 |
share_attention=True,
|
@@ -43,50 +43,56 @@ handler = sa_handler.Handler(pipeline)
|
|
43 |
handler.register(sa_args, )
|
44 |
|
45 |
|
46 |
-
|
47 |
-
|
48 |
-
# run ControlNet depth with StyleAligned
|
49 |
def style_aligned_controlnet(ref_style_prompt, depth_map, ref_image, img_generation_prompt):
|
50 |
-
|
51 |
-
|
52 |
-
|
53 |
-
|
54 |
-
|
55 |
-
|
56 |
-
|
57 |
-
|
58 |
-
|
59 |
-
|
60 |
-
|
61 |
-
|
62 |
-
|
63 |
-
|
64 |
-
|
65 |
-
|
66 |
-
|
67 |
-
|
68 |
-
|
|
|
|
|
69 |
with gr.Blocks() as demo:
|
70 |
gr.HTML('<h1 style="text-align: center;">Style-aligned with ControlNet Depth</h1>')
|
71 |
with gr.Row():
|
72 |
|
73 |
with gr.Column(variant='panel'):
|
|
|
74 |
ref_style_prompt = gr.Textbox(
|
75 |
label='Reference style prompt',
|
76 |
info="Enter a Prompt to generate the reference image", placeholder='a poster in <style name> style'
|
77 |
)
|
|
|
78 |
depth_map = gr.Checkbox(label='Depth-map',)
|
|
|
79 |
ref_style_image = gr.Image(visible=False, label='Reference style image')
|
80 |
|
81 |
with gr.Column(variant='panel'):
|
|
|
82 |
ref_image = gr.Image(label="Upload the reference image",
|
83 |
type='filepath' )
|
|
|
84 |
img_generation_prompt = gr.Textbox(
|
85 |
label='ControlNet Prompt',
|
86 |
info="Enter a Prompt to generate images using ControlNet and Style-aligned",
|
87 |
)
|
88 |
-
|
89 |
btn = gr.Button("Generate", size='sm')
|
|
|
90 |
gallery = gr.Gallery(label="Style-Aligned ControlNet - Generated images",
|
91 |
elem_id="gallery",
|
92 |
columns=5,
|
@@ -101,7 +107,7 @@ with gr.Blocks() as demo:
|
|
101 |
api_name="style_aligned_controlnet")
|
102 |
|
103 |
|
104 |
-
|
105 |
gr.Examples(
|
106 |
examples=[
|
107 |
['A poster in a papercut art style.', False, 'example_image/A.png', 'Letter A in a papercut art style.'],
|
@@ -116,5 +122,5 @@ with gr.Blocks() as demo:
|
|
116 |
fn=style_aligned_controlnet,
|
117 |
)
|
118 |
|
119 |
-
|
120 |
demo.launch()
|
|
|
3 |
from diffusers.utils import load_image
|
4 |
from transformers import DPTImageProcessor, DPTForDepthEstimation
|
5 |
import torch
|
|
|
6 |
import sa_handler
|
7 |
import pipeline_calls
|
8 |
|
9 |
|
10 |
|
11 |
+
# Initialize models
|
|
|
12 |
depth_estimator = DPTForDepthEstimation.from_pretrained("Intel/dpt-hybrid-midas").to("cuda")
|
13 |
feature_processor = DPTImageProcessor.from_pretrained("Intel/dpt-hybrid-midas")
|
14 |
|
|
|
27 |
use_safetensors=True,
|
28 |
torch_dtype=torch.float16,
|
29 |
).to("cuda")
|
30 |
+
# Configure pipeline for CPU offloading and VAE slicing
|
31 |
pipeline.enable_model_cpu_offload()
|
32 |
pipeline.enable_vae_slicing()
|
33 |
|
34 |
+
# Initialize style-aligned handler
|
35 |
sa_args = sa_handler.StyleAlignedArgs(share_group_norm=False,
|
36 |
share_layer_norm=False,
|
37 |
share_attention=True,
|
|
|
43 |
handler.register(sa_args, )
|
44 |
|
45 |
|
46 |
+
# Function to run ControlNet depth with StyleAligned
|
|
|
|
|
47 |
def style_aligned_controlnet(ref_style_prompt, depth_map, ref_image, img_generation_prompt):
|
48 |
+
try:
|
49 |
+
if depth_map == True:
|
50 |
+
image = load_image(ref_image)
|
51 |
+
depth_image = pipeline_calls.get_depth_map(image, feature_processor, depth_estimator)
|
52 |
+
else:
|
53 |
+
depth_image = load_image(ref_image).resize((1024, 1024))
|
54 |
+
controlnet_conditioning_scale = 0.8
|
55 |
+
num_images_per_prompt = 3 # adjust according to VRAM size
|
56 |
+
latents = torch.randn(1 + num_images_per_prompt, 4, 128, 128).to(pipeline.unet.dtype)
|
57 |
+
latents[1:] = torch.randn(num_images_per_prompt, 4, 128, 128).to(pipeline.unet.dtype)
|
58 |
+
images = pipeline_calls.controlnet_call(pipeline, [ref_style_prompt, img_generation_prompt],
|
59 |
+
image=depth_image,
|
60 |
+
num_inference_steps=50,
|
61 |
+
controlnet_conditioning_scale=controlnet_conditioning_scale,
|
62 |
+
num_images_per_prompt=num_images_per_prompt,
|
63 |
+
latents=latents)
|
64 |
+
return [images[0], depth_image] + images[1:], gr.Image(value=images[0], visible=True)
|
65 |
+
except Exception as e:
|
66 |
+
raise gr.Error(f"Error in generating images:{e}")
|
67 |
+
|
68 |
+
# Create a Gradio UI
|
69 |
with gr.Blocks() as demo:
|
70 |
gr.HTML('<h1 style="text-align: center;">Style-aligned with ControlNet Depth</h1>')
|
71 |
with gr.Row():
|
72 |
|
73 |
with gr.Column(variant='panel'):
|
74 |
+
# Textbox for reference style prompt
|
75 |
ref_style_prompt = gr.Textbox(
|
76 |
label='Reference style prompt',
|
77 |
info="Enter a Prompt to generate the reference image", placeholder='a poster in <style name> style'
|
78 |
)
|
79 |
+
# Checkbox for using controller depth-map
|
80 |
depth_map = gr.Checkbox(label='Depth-map',)
|
81 |
+
# Image display for the generated reference style image
|
82 |
ref_style_image = gr.Image(visible=False, label='Reference style image')
|
83 |
|
84 |
with gr.Column(variant='panel'):
|
85 |
+
# Image upload option for uploading a reference image for controlnet
|
86 |
ref_image = gr.Image(label="Upload the reference image",
|
87 |
type='filepath' )
|
88 |
+
# Textbox for ControlNet prompt
|
89 |
img_generation_prompt = gr.Textbox(
|
90 |
label='ControlNet Prompt',
|
91 |
info="Enter a Prompt to generate images using ControlNet and Style-aligned",
|
92 |
)
|
93 |
+
# Button to trigger image generation
|
94 |
btn = gr.Button("Generate", size='sm')
|
95 |
+
# Gallery to display generated images
|
96 |
gallery = gr.Gallery(label="Style-Aligned ControlNet - Generated images",
|
97 |
elem_id="gallery",
|
98 |
columns=5,
|
|
|
107 |
api_name="style_aligned_controlnet")
|
108 |
|
109 |
|
110 |
+
# Example inputs for the Gradio interface
|
111 |
gr.Examples(
|
112 |
examples=[
|
113 |
['A poster in a papercut art style.', False, 'example_image/A.png', 'Letter A in a papercut art style.'],
|
|
|
122 |
fn=style_aligned_controlnet,
|
123 |
)
|
124 |
|
125 |
+
# Launch the Gradio demo
|
126 |
demo.launch()
|