Spaces:
Runtime error
Runtime error
Jordan Legg
commited on
Commit
Β·
13ab5d1
1
Parent(s):
f0decf0
twin model
Browse files
app.py
CHANGED
@@ -1,11 +1,10 @@
|
|
1 |
import gradio as gr
|
2 |
import numpy as np
|
3 |
import random
|
4 |
-
import spaces
|
5 |
import torch
|
6 |
from PIL import Image
|
7 |
from torchvision import transforms
|
8 |
-
from diffusers import DiffusionPipeline
|
9 |
|
10 |
# Define constants
|
11 |
dtype = torch.bfloat16
|
@@ -13,7 +12,11 @@ device = "cuda" if torch.cuda.is_available() else "cpu"
|
|
13 |
MAX_SEED = np.iinfo(np.int32).max
|
14 |
MAX_IMAGE_SIZE = 2048
|
15 |
|
16 |
-
# Load the
|
|
|
|
|
|
|
|
|
17 |
pipe = DiffusionPipeline.from_pretrained("black-forest-labs/FLUX.1-schnell", torch_dtype=dtype)
|
18 |
pipe.enable_model_cpu_offload()
|
19 |
pipe.vae.enable_slicing()
|
@@ -27,11 +30,13 @@ def preprocess_image(image, image_size):
|
|
27 |
transforms.Normalize([0.5], [0.5])
|
28 |
])
|
29 |
image = preprocess(image).unsqueeze(0).to(device, dtype=dtype)
|
|
|
30 |
return image
|
31 |
|
32 |
def encode_image(image, vae):
|
33 |
with torch.no_grad():
|
34 |
latents = vae.encode(image).latent_dist.sample() * 0.18215
|
|
|
35 |
return latents
|
36 |
|
37 |
@spaces.GPU()
|
@@ -61,20 +66,33 @@ def infer(prompt, init_image=None, seed=42, randomize_seed=False, width=1024, he
|
|
61 |
return fallback_image, seed
|
62 |
else:
|
63 |
# img2img case
|
64 |
-
|
|
|
65 |
init_image = init_image.convert("RGB")
|
66 |
init_image = preprocess_image(init_image, vae_image_size)
|
67 |
-
|
68 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
69 |
latents = torch.nn.functional.interpolate(latents, size=(height // 8, width // 8))
|
70 |
-
|
71 |
-
|
72 |
-
|
|
|
|
|
|
|
|
|
73 |
latents = conv(latents)
|
74 |
|
75 |
-
latents = latents.permute(0, 2, 3, 1).contiguous().view(-1,
|
|
|
76 |
|
77 |
try:
|
|
|
78 |
# Determine if 'timesteps' is required for the transformer
|
79 |
if hasattr(pipe.transformer, 'forward') and hasattr(pipe.transformer.forward, '__code__') and 'timesteps' in pipe.transformer.forward.__code__.co_varnames:
|
80 |
timestep = torch.tensor([num_inference_steps], device=device, dtype=dtype)
|
@@ -86,6 +104,7 @@ def infer(prompt, init_image=None, seed=42, randomize_seed=False, width=1024, he
|
|
86 |
return fallback_image, seed
|
87 |
|
88 |
try:
|
|
|
89 |
image = pipe(
|
90 |
prompt=prompt,
|
91 |
height=height,
|
@@ -95,6 +114,7 @@ def infer(prompt, init_image=None, seed=42, randomize_seed=False, width=1024, he
|
|
95 |
guidance_scale=0.0,
|
96 |
latents=latents
|
97 |
).images[0]
|
|
|
98 |
except Exception as e:
|
99 |
print(f"Pipeline call with latents failed with error: {e}")
|
100 |
return fallback_image, seed
|
@@ -210,9 +230,8 @@ with gr.Blocks(css=css) as demo:
|
|
210 |
cache_examples="lazy"
|
211 |
)
|
212 |
|
213 |
-
|
214 |
-
|
215 |
-
fn=infer,
|
216 |
inputs=[prompt, init_image, seed, randomize_seed, width, height, num_inference_steps],
|
217 |
outputs=[result, seed]
|
218 |
)
|
|
|
1 |
import gradio as gr
|
2 |
import numpy as np
|
3 |
import random
|
|
|
4 |
import torch
|
5 |
from PIL import Image
|
6 |
from torchvision import transforms
|
7 |
+
from diffusers import DiffusionPipeline, AutoencoderKL
|
8 |
|
9 |
# Define constants
|
10 |
dtype = torch.bfloat16
|
|
|
12 |
MAX_SEED = np.iinfo(np.int32).max
|
13 |
MAX_IMAGE_SIZE = 2048
|
14 |
|
15 |
+
# Load the initial VAE model for preprocessing
|
16 |
+
vae_model_name = "CompVis/stable-diffusion-v1-4" # Example VAE model
|
17 |
+
vae = AutoencoderKL.from_pretrained(vae_model_name).to(device)
|
18 |
+
|
19 |
+
# Load the FLUX diffusion pipeline with optimizations
|
20 |
pipe = DiffusionPipeline.from_pretrained("black-forest-labs/FLUX.1-schnell", torch_dtype=dtype)
|
21 |
pipe.enable_model_cpu_offload()
|
22 |
pipe.vae.enable_slicing()
|
|
|
30 |
transforms.Normalize([0.5], [0.5])
|
31 |
])
|
32 |
image = preprocess(image).unsqueeze(0).to(device, dtype=dtype)
|
33 |
+
print("Image processed successfully.")
|
34 |
return image
|
35 |
|
36 |
def encode_image(image, vae):
|
37 |
with torch.no_grad():
|
38 |
latents = vae.encode(image).latent_dist.sample() * 0.18215
|
39 |
+
print("Image encoded successfully.")
|
40 |
return latents
|
41 |
|
42 |
@spaces.GPU()
|
|
|
66 |
return fallback_image, seed
|
67 |
else:
|
68 |
# img2img case
|
69 |
+
print("Initial image provided, starting preprocessing...")
|
70 |
+
vae_image_size = 1024 # Using FLUX VAE sample size for preprocessing
|
71 |
init_image = init_image.convert("RGB")
|
72 |
init_image = preprocess_image(init_image, vae_image_size)
|
73 |
+
|
74 |
+
print("Starting encoding of the image...")
|
75 |
+
latents = encode_image(init_image, vae)
|
76 |
+
|
77 |
+
print(f"Latents shape after encoding: {latents.shape}")
|
78 |
+
|
79 |
+
# Ensure the latents size matches the expected input size for the FLUX model
|
80 |
+
print("Interpolating latents to match model's input size...")
|
81 |
latents = torch.nn.functional.interpolate(latents, size=(height // 8, width // 8))
|
82 |
+
|
83 |
+
latent_channels = 16 # Using FLUX VAE latent channels
|
84 |
+
print(f"Latent channels from VAE: {latent_channels}, expected by FLUX model: {pipe.vae.config.latent_channels}")
|
85 |
+
|
86 |
+
if latent_channels != pipe.vae.config.latent_channels:
|
87 |
+
print(f"Adjusting latent channels from {latent_channels} to {pipe.vae.config.latent_channels}")
|
88 |
+
conv = torch.nn.Conv2d(latent_channels, pipe.vae.config.latent_channels, kernel_size=1).to(device, dtype=dtype)
|
89 |
latents = conv(latents)
|
90 |
|
91 |
+
latents = latents.permute(0, 2, 3, 1).contiguous().view(-1, pipe.vae.config.latent_channels)
|
92 |
+
print(f"Latents shape after permutation: {latents.shape}")
|
93 |
|
94 |
try:
|
95 |
+
print("Sending latents to the FLUX transformer...")
|
96 |
# Determine if 'timesteps' is required for the transformer
|
97 |
if hasattr(pipe.transformer, 'forward') and hasattr(pipe.transformer.forward, '__code__') and 'timesteps' in pipe.transformer.forward.__code__.co_varnames:
|
98 |
timestep = torch.tensor([num_inference_steps], device=device, dtype=dtype)
|
|
|
104 |
return fallback_image, seed
|
105 |
|
106 |
try:
|
107 |
+
print("Generating final image with the FLUX pipeline...")
|
108 |
image = pipe(
|
109 |
prompt=prompt,
|
110 |
height=height,
|
|
|
114 |
guidance_scale=0.0,
|
115 |
latents=latents
|
116 |
).images[0]
|
117 |
+
print("Image generation completed.")
|
118 |
except Exception as e:
|
119 |
print(f"Pipeline call with latents failed with error: {e}")
|
120 |
return fallback_image, seed
|
|
|
230 |
cache_examples="lazy"
|
231 |
)
|
232 |
|
233 |
+
run_button.click(
|
234 |
+
infer,
|
|
|
235 |
inputs=[prompt, init_image, seed, randomize_seed, width, height, num_inference_steps],
|
236 |
outputs=[result, seed]
|
237 |
)
|