Spaces:
Running
on
Zero
Running
on
Zero
Update app.py
Browse files
app.py
CHANGED
@@ -1,70 +1,89 @@
|
|
1 |
import gradio as gr
|
2 |
import numpy as np
|
|
|
|
|
3 |
import random
|
4 |
-
from diffusers import
|
|
|
|
|
|
|
|
|
5 |
import torch
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
6 |
|
7 |
-
|
8 |
|
9 |
-
|
10 |
-
|
11 |
-
|
12 |
-
|
13 |
-
|
14 |
-
|
15 |
-
|
16 |
-
|
|
|
17 |
|
18 |
MAX_SEED = np.iinfo(np.int32).max
|
19 |
-
MAX_IMAGE_SIZE =
|
20 |
|
21 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
22 |
|
23 |
if randomize_seed:
|
24 |
seed = random.randint(0, MAX_SEED)
|
25 |
-
|
26 |
generator = torch.Generator().manual_seed(seed)
|
27 |
-
|
28 |
-
image = pipe(
|
29 |
-
prompt = prompt,
|
30 |
-
negative_prompt = negative_prompt,
|
31 |
-
guidance_scale = guidance_scale,
|
32 |
-
num_inference_steps = num_inference_steps,
|
33 |
-
width = width,
|
34 |
-
height = height,
|
35 |
-
generator = generator
|
36 |
-
).images[0]
|
37 |
-
|
38 |
-
return image
|
39 |
|
40 |
-
|
41 |
-
|
42 |
-
|
43 |
-
|
44 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
45 |
|
46 |
-
css="""
|
47 |
#col-container {
|
48 |
margin: 0 auto;
|
49 |
max-width: 520px;
|
50 |
}
|
51 |
"""
|
52 |
|
53 |
-
if torch.cuda.is_available():
|
54 |
-
power_device = "GPU"
|
55 |
-
else:
|
56 |
-
power_device = "CPU"
|
57 |
-
|
58 |
with gr.Blocks(css=css) as demo:
|
59 |
-
|
60 |
with gr.Column(elem_id="col-container"):
|
61 |
-
gr.Markdown(
|
62 |
-
# Text-to-Image
|
63 |
-
|
|
|
|
|
64 |
""")
|
65 |
-
|
66 |
with gr.Row():
|
67 |
-
|
68 |
prompt = gr.Text(
|
69 |
label="Prompt",
|
70 |
show_label=False,
|
@@ -72,20 +91,22 @@ with gr.Blocks(css=css) as demo:
|
|
72 |
placeholder="Enter your prompt",
|
73 |
container=False,
|
74 |
)
|
75 |
-
|
76 |
run_button = gr.Button("Run", scale=0)
|
77 |
-
|
|
|
78 |
result = gr.Image(label="Result", show_label=False)
|
79 |
|
|
|
80 |
with gr.Accordion("Advanced Settings", open=False):
|
81 |
-
|
82 |
negative_prompt = gr.Text(
|
83 |
label="Negative prompt",
|
84 |
max_lines=1,
|
85 |
placeholder="Enter a negative prompt",
|
86 |
-
|
87 |
)
|
88 |
-
|
89 |
seed = gr.Slider(
|
90 |
label="Seed",
|
91 |
minimum=0,
|
@@ -93,54 +114,47 @@ with gr.Blocks(css=css) as demo:
|
|
93 |
step=1,
|
94 |
value=0,
|
95 |
)
|
96 |
-
|
97 |
randomize_seed = gr.Checkbox(label="Randomize seed", value=True)
|
98 |
-
|
99 |
with gr.Row():
|
100 |
-
|
101 |
width = gr.Slider(
|
102 |
label="Width",
|
103 |
minimum=256,
|
104 |
maximum=MAX_IMAGE_SIZE,
|
105 |
step=32,
|
106 |
-
value=
|
107 |
)
|
108 |
-
|
109 |
height = gr.Slider(
|
110 |
label="Height",
|
111 |
minimum=256,
|
112 |
maximum=MAX_IMAGE_SIZE,
|
113 |
step=32,
|
114 |
-
value=
|
115 |
)
|
116 |
-
|
117 |
with gr.Row():
|
118 |
-
|
119 |
guidance_scale = gr.Slider(
|
120 |
label="Guidance scale",
|
121 |
minimum=0.0,
|
122 |
-
maximum=
|
123 |
step=0.1,
|
124 |
-
value=
|
125 |
)
|
126 |
-
|
127 |
num_inference_steps = gr.Slider(
|
128 |
label="Number of inference steps",
|
129 |
minimum=1,
|
130 |
-
maximum=
|
131 |
step=1,
|
132 |
-
value=
|
133 |
)
|
134 |
-
|
135 |
-
|
136 |
-
|
137 |
-
|
138 |
-
|
139 |
-
|
140 |
-
run_button.click(
|
141 |
-
fn = infer,
|
142 |
-
inputs = [prompt, negative_prompt, seed, randomize_seed, width, height, guidance_scale, num_inference_steps],
|
143 |
-
outputs = [result]
|
144 |
)
|
145 |
|
146 |
-
demo.queue().launch()
|
|
|
1 |
import gradio as gr
|
2 |
import numpy as np
|
3 |
+
import PIL.Image
|
4 |
+
from PIL import Image
|
5 |
import random
|
6 |
+
from diffusers import ControlNetModel, StableDiffusionXLControlNetPipeline, StableDiffusionXLPipeline, AutoencoderKL
|
7 |
+
from diffusers import DDIMScheduler, EulerAncestralDiscreteScheduler
|
8 |
+
#from diffusers.utils import load_image
|
9 |
+
|
10 |
+
import cv2
|
11 |
import torch
|
12 |
+
import spaces
|
13 |
+
|
14 |
+
|
15 |
+
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
16 |
+
|
17 |
+
controlnet = ControlNetModel.from_pretrained(
|
18 |
+
#"2vXpSwA7/test_controlnet2/CN-anytest_v4-marged_am_dim256.safetensors",
|
19 |
+
"xinsir/controlnet-scribble-sdxl-1.0",
|
20 |
+
torch_dtype=torch.float16
|
21 |
+
#from_tf=False,
|
22 |
+
#variant="safetensors"
|
23 |
+
)
|
24 |
|
25 |
+
vae = AutoencoderKL.from_pretrained("madebyollin/sdxl-vae-fp16-fix", torch_dtype=torch.float16)
|
26 |
|
27 |
+
pipe = StableDiffusionXLControlNetPipeline.from_pretrained(
|
28 |
+
"yodayo-ai/holodayo-xl-2.1",
|
29 |
+
controlnet=controlnet,
|
30 |
+
vae=vae,
|
31 |
+
torch_dtype=torch.float16,
|
32 |
+
)
|
33 |
+
|
34 |
+
pipe.scheduler = EulerAncestralDiscreteScheduler.from_config(pipe.scheduler.config)
|
35 |
+
pipe.to(device)
|
36 |
|
37 |
MAX_SEED = np.iinfo(np.int32).max
|
38 |
+
MAX_IMAGE_SIZE = 1216
|
39 |
|
40 |
+
|
41 |
+
@spaces.GPU
|
42 |
+
def infer(prompt, negative_prompt, seed, randomize_seed, width, height, guidance_scale, num_inference_steps, image: PIL.Image.Image) -> PIL.Image.Image:
|
43 |
+
width, height = image['composite'].size
|
44 |
+
ratio = np.sqrt(1024. * 1024. / (width * height))
|
45 |
+
new_width, new_height = int(width * ratio), int(height * ratio)
|
46 |
+
image = image['composite'].resize((new_width, new_height))
|
47 |
+
print(image)
|
48 |
|
49 |
if randomize_seed:
|
50 |
seed = random.randint(0, MAX_SEED)
|
51 |
+
|
52 |
generator = torch.Generator().manual_seed(seed)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
53 |
|
54 |
+
output_image = pipe(
|
55 |
+
prompt=prompt + ", masterpiece, best quality, very aesthetic, absurdres",
|
56 |
+
negative_prompt=negative_prompt,
|
57 |
+
image=image,
|
58 |
+
controlnet_conditioning_scale=1.0,
|
59 |
+
guidance_scale=guidance_scale,
|
60 |
+
num_inference_steps=num_inference_steps,
|
61 |
+
width=new_width,
|
62 |
+
height=new_height,
|
63 |
+
generator=generator
|
64 |
+
).images[0]
|
65 |
+
|
66 |
+
return output_image
|
67 |
+
|
68 |
|
69 |
+
css = """
|
70 |
#col-container {
|
71 |
margin: 0 auto;
|
72 |
max-width: 520px;
|
73 |
}
|
74 |
"""
|
75 |
|
|
|
|
|
|
|
|
|
|
|
76 |
with gr.Blocks(css=css) as demo:
|
77 |
+
|
78 |
with gr.Column(elem_id="col-container"):
|
79 |
+
gr.Markdown("""
|
80 |
+
# Text-to-Image Demo
|
81 |
+
using :
|
82 |
+
[Holodayo XL 2.1](https://huggingface.co/yodayo-ai/holodayo-xl-2.1),
|
83 |
+
[scribble](https://huggingface.co/xinsir/controlnet-scribble-sdxl-1.0)
|
84 |
""")
|
85 |
+
|
86 |
with gr.Row():
|
|
|
87 |
prompt = gr.Text(
|
88 |
label="Prompt",
|
89 |
show_label=False,
|
|
|
91 |
placeholder="Enter your prompt",
|
92 |
container=False,
|
93 |
)
|
94 |
+
|
95 |
run_button = gr.Button("Run", scale=0)
|
96 |
+
|
97 |
+
image = gr.ImageEditor(type="pil", image_mode="L", crop_size=(512, 512))
|
98 |
result = gr.Image(label="Result", show_label=False)
|
99 |
|
100 |
+
|
101 |
with gr.Accordion("Advanced Settings", open=False):
|
102 |
+
|
103 |
negative_prompt = gr.Text(
|
104 |
label="Negative prompt",
|
105 |
max_lines=1,
|
106 |
placeholder="Enter a negative prompt",
|
107 |
+
value="nsfw, (low quality, worst quality:1.2), very displeasing, 3d, watermark, signature, ugly, poorly drawn"
|
108 |
)
|
109 |
+
|
110 |
seed = gr.Slider(
|
111 |
label="Seed",
|
112 |
minimum=0,
|
|
|
114 |
step=1,
|
115 |
value=0,
|
116 |
)
|
117 |
+
|
118 |
randomize_seed = gr.Checkbox(label="Randomize seed", value=True)
|
119 |
+
|
120 |
with gr.Row():
|
|
|
121 |
width = gr.Slider(
|
122 |
label="Width",
|
123 |
minimum=256,
|
124 |
maximum=MAX_IMAGE_SIZE,
|
125 |
step=32,
|
126 |
+
value=1024,#832,
|
127 |
)
|
128 |
+
|
129 |
height = gr.Slider(
|
130 |
label="Height",
|
131 |
minimum=256,
|
132 |
maximum=MAX_IMAGE_SIZE,
|
133 |
step=32,
|
134 |
+
value=1024,#1216,
|
135 |
)
|
136 |
+
|
137 |
with gr.Row():
|
|
|
138 |
guidance_scale = gr.Slider(
|
139 |
label="Guidance scale",
|
140 |
minimum=0.0,
|
141 |
+
maximum=20.0,
|
142 |
step=0.1,
|
143 |
+
value=7,
|
144 |
)
|
145 |
+
|
146 |
num_inference_steps = gr.Slider(
|
147 |
label="Number of inference steps",
|
148 |
minimum=1,
|
149 |
+
maximum=28,
|
150 |
step=1,
|
151 |
+
value=28,
|
152 |
)
|
153 |
+
|
154 |
+
run_button.click(#lambda x: None, inputs=None, outputs=result).then(
|
155 |
+
fn=infer,
|
156 |
+
inputs=[use_image, prompt, negative_prompt, seed, randomize_seed, width, height, guidance_scale, num_inference_steps,image],
|
157 |
+
outputs=[result]
|
|
|
|
|
|
|
|
|
|
|
158 |
)
|
159 |
|
160 |
+
demo.queue().launch()
|