Spaces:
Build error
Build error
Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,88 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
import jax
|
3 |
+
import numpy as np
|
4 |
+
import jax.numpy as jnp
|
5 |
+
from flax.jax_utils import replicate
|
6 |
+
from flax.training.common_utils import shard
|
7 |
+
from PIL import Image
|
8 |
+
from diffusers import FlaxStableDiffusionControlNetPipeline, FlaxControlNetModel
|
9 |
+
import cv2
|
10 |
+
|
11 |
+
# load control net and stable diffusion v1-5
|
12 |
+
controlnet, controlnet_params = FlaxControlNetModel.from_pretrained(
|
13 |
+
"Nahrawy/controlnet-VIDIT-FAID", dtype=jnp.bfloat16, revision="615ba4a457b95a0eba813bcc8caf842c03a4f7bd"
|
14 |
+
)
|
15 |
+
pipe, params = FlaxStableDiffusionControlNetPipeline.from_pretrained(
|
16 |
+
"runwayml/stable-diffusion-v1-5", controlnet=controlnet, revision="flax", dtype=jnp.bfloat16
|
17 |
+
)
|
18 |
+
|
19 |
+
def create_key(seed=0):
|
20 |
+
return jax.random.PRNGKey(seed)
|
21 |
+
|
22 |
+
def process_mask(image):
|
23 |
+
mask = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
|
24 |
+
mask = cv2.resize(mask,(512,512))
|
25 |
+
return mask
|
26 |
+
|
27 |
+
|
28 |
+
|
29 |
+
def infer(prompts, negative_prompts, image):
|
30 |
+
params["controlnet"] = controlnet_params
|
31 |
+
|
32 |
+
num_samples = 1 #jax.device_count()
|
33 |
+
rng = create_key(0)
|
34 |
+
rng = jax.random.split(rng, jax.device_count())
|
35 |
+
im = process_mask(image)
|
36 |
+
mask = Image.fromarray(im)
|
37 |
+
|
38 |
+
prompt_ids = pipe.prepare_text_inputs([prompts] * num_samples)
|
39 |
+
negative_prompt_ids = pipe.prepare_text_inputs([negative_prompts] * num_samples)
|
40 |
+
processed_image = pipe.prepare_image_inputs([mask] * num_samples)
|
41 |
+
|
42 |
+
p_params = replicate(params)
|
43 |
+
prompt_ids = shard(prompt_ids)
|
44 |
+
negative_prompt_ids = shard(negative_prompt_ids)
|
45 |
+
processed_image = shard(processed_image)
|
46 |
+
print(processed_image[0].shape)
|
47 |
+
output = pipe(
|
48 |
+
prompt_ids=prompt_ids,
|
49 |
+
image=processed_image,
|
50 |
+
params=p_params,
|
51 |
+
prng_seed=rng,
|
52 |
+
num_inference_steps=50,
|
53 |
+
neg_prompt_ids=negative_prompt_ids,
|
54 |
+
jit=True,
|
55 |
+
).images
|
56 |
+
|
57 |
+
output_images = pipe.numpy_to_pil(np.asarray(output.reshape((num_samples,) + output.shape[-3:])))
|
58 |
+
return output_images
|
59 |
+
|
60 |
+
e_images = ['examples/0.png',
|
61 |
+
'examples/1.png'
|
62 |
+
'examples/2.png']
|
63 |
+
e_prompts = ['a dog in the middle of the road, shadow on the ground,light direction north-east',
|
64 |
+
'a skyscraper in the middle of an intersection, shadow on the ground, light direction east',
|
65 |
+
'a red rural house, light temperature 5500, shadow on the ground, light direction south-west']
|
66 |
+
e_negative_prompts = ['monochromatic, unrealistic, bad looking, full of glitches'*3]
|
67 |
+
examples = []
|
68 |
+
for image, prompt, negative_prompt in zip(e_images, e_prompts, e_negative_prompts):
|
69 |
+
examples.append([prompt, negative_prompt, image])
|
70 |
+
|
71 |
+
with gr.Blocks() as demo:
|
72 |
+
gr.Markdown(title)
|
73 |
+
prompts = gr.Textbox(label='prompts')
|
74 |
+
negative_prompts = gr.Textbox(label='negative_prompts')
|
75 |
+
with gr.Row():
|
76 |
+
with gr.Column():
|
77 |
+
in_image = gr.Image(label="Depth Map Conditioning")
|
78 |
+
with gr.Column():
|
79 |
+
out_image = gr.Image(label="Generated Image")
|
80 |
+
with gr.Row():
|
81 |
+
btn = gr.Button("Run")
|
82 |
+
gr.Examples(examples=examples,
|
83 |
+
inputs=[prompts,negative_prompts, in_image],
|
84 |
+
outputs=out_image,
|
85 |
+
cache_examples=True)
|
86 |
+
btn.click(fn=infer, inputs=[prompts,negative_prompts, in_image] , outputs=out_image)
|
87 |
+
|
88 |
+
demo.launch()
|