Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
|
@@ -32,48 +32,56 @@ ip_adapter_path = "./"
|
|
| 32 |
print("Files in root directory:", os.listdir("."))
|
| 33 |
print("Files in ./unet/ directory:", os.listdir("./unet") if os.path.exists("./unet") else "No ./unet/ directory")
|
| 34 |
|
| 35 |
-
# Check if weights exist or download them
|
| 36 |
-
kolors_weights = model_path + "
|
| 37 |
if not os.path.exists(kolors_weights):
|
| 38 |
-
|
| 39 |
-
if not os.path.exists(
|
| 40 |
-
|
| 41 |
-
os.
|
| 42 |
-
|
| 43 |
-
|
| 44 |
-
|
| 45 |
-
|
| 46 |
-
|
| 47 |
-
|
| 48 |
-
|
| 49 |
-
|
| 50 |
-
|
| 51 |
-
|
| 52 |
-
|
| 53 |
-
|
| 54 |
-
|
| 55 |
-
|
| 56 |
-
|
| 57 |
-
|
| 58 |
-
|
| 59 |
-
|
| 60 |
-
|
| 61 |
-
|
| 62 |
-
|
| 63 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 64 |
raise FileNotFoundError(f"IP-Adapter weights not found at {ip_adapter_path}")
|
| 65 |
|
| 66 |
# Initialize model with empty weights
|
| 67 |
with init_empty_weights():
|
| 68 |
pipe = StableDiffusionXLPipeline.from_pretrained(
|
| 69 |
-
"Kwai-Kolors/Kolors-
|
| 70 |
torch_dtype=dtype,
|
| 71 |
safety_checker=None,
|
| 72 |
)
|
| 73 |
|
| 74 |
# Load and dispatch model with accelerate
|
| 75 |
pipe = load_checkpoint_and_dispatch(pipe, model_path, device_map="cpu", offload_folder=None)
|
| 76 |
-
pipe.load_ip_adapter("
|
| 77 |
|
| 78 |
def generate_image(uploaded_image, prompt):
|
| 79 |
img = cv2.cvtColor(np.array(uploaded_image), cv2.COLOR_RGB2BGR)
|
|
@@ -101,7 +109,7 @@ interface = gr.Interface(
|
|
| 101 |
fn=generate_image,
|
| 102 |
inputs=[gr.Image(type="pil", label="Upload Reference Image"), gr.Textbox(label="Enter Prompt", placeholder="e.g., A photorealistic astronaut in space")],
|
| 103 |
outputs=[gr.Textbox(label="Status"), gr.Image(label="Generated Image")],
|
| 104 |
-
title="Face Reference Image Generator (Kolors
|
| 105 |
description="Upload an image with a face, enter a prompt, and generate a new image preserving the reference face."
|
| 106 |
)
|
| 107 |
|
|
|
|
| 32 |
print("Files in root directory:", os.listdir("."))
|
| 33 |
print("Files in ./unet/ directory:", os.listdir("./unet") if os.path.exists("./unet") else "No ./unet/ directory")
|
| 34 |
|
| 35 |
+
# Check if base model weights exist or download them
|
| 36 |
+
kolors_weights = model_path + "diffusers_weights.safetensors"
|
| 37 |
if not os.path.exists(kolors_weights):
|
| 38 |
+
kolors_weights = model_path + "diffusion_pytorch_model.fp16.safetensors"
|
| 39 |
+
if not os.path.exists(kolors_weights):
|
| 40 |
+
kolors_weights_unet = "./unet/diffusion_pytorch_model.fp16.safetensors"
|
| 41 |
+
if not os.path.exists(kolors_weights_unet):
|
| 42 |
+
print("Preloading failed. Attempting runtime download with retry...")
|
| 43 |
+
os.makedirs("./unet", exist_ok=True)
|
| 44 |
+
max_retries = 3
|
| 45 |
+
correct_url = "https://huggingface.co/Kwai-Kolors/Kolors/raw/main/unet/diffusion_pytorch_model.fp16.safetensors"
|
| 46 |
+
for attempt in range(max_retries):
|
| 47 |
+
try:
|
| 48 |
+
print(f"Download attempt {attempt + 1} of {max_retries}")
|
| 49 |
+
urllib.request.urlretrieve(correct_url, kolors_weights_unet)
|
| 50 |
+
print("Kolors base weights downloaded to", kolors_weights_unet)
|
| 51 |
+
model_path = "./unet/"
|
| 52 |
+
kolors_weights = kolors_weights_unet
|
| 53 |
+
break
|
| 54 |
+
except urllib.error.HTTPError as e:
|
| 55 |
+
print(f"Download attempt {attempt + 1} failed: HTTP Error {e.code} - {e.reason}")
|
| 56 |
+
if attempt < max_retries - 1:
|
| 57 |
+
time.sleep(5)
|
| 58 |
+
else:
|
| 59 |
+
raise FileNotFoundError(f"Failed to download Kolors base weights after {max_retries} attempts: HTTP Error {e.code} - {e.reason}. Verify the URL or contact support.")
|
| 60 |
+
except Exception as e:
|
| 61 |
+
print(f"Download attempt {attempt + 1} failed: {e}")
|
| 62 |
+
if attempt < max_retries - 1:
|
| 63 |
+
time.sleep(5)
|
| 64 |
+
else:
|
| 65 |
+
raise FileNotFoundError(f"Failed to download Kolors base weights after {max_retries} attempts: {e}. Check network access or contact support.")
|
| 66 |
+
else:
|
| 67 |
+
model_path = "./unet/"
|
| 68 |
+
kolors_weights = kolors_weights_unet
|
| 69 |
+
|
| 70 |
+
# Check if IP-Adapter weights exist (preloaded)
|
| 71 |
+
if not os.path.exists(ip_adapter_path + "ipa-faceid-plus.bin"):
|
| 72 |
raise FileNotFoundError(f"IP-Adapter weights not found at {ip_adapter_path}")
|
| 73 |
|
| 74 |
# Initialize model with empty weights
|
| 75 |
with init_empty_weights():
|
| 76 |
pipe = StableDiffusionXLPipeline.from_pretrained(
|
| 77 |
+
"Kwai-Kolors/Kolors-IP-Adapter-FaceID-Plus",
|
| 78 |
torch_dtype=dtype,
|
| 79 |
safety_checker=None,
|
| 80 |
)
|
| 81 |
|
| 82 |
# Load and dispatch model with accelerate
|
| 83 |
pipe = load_checkpoint_and_dispatch(pipe, model_path, device_map="cpu", offload_folder=None)
|
| 84 |
+
pipe.load_ip_adapter("Kwai-Kolors/Kolors-IP-Adapter-FaceID-Plus", subfolder=None, weight_name="ipa-faceid-plus.bin")
|
| 85 |
|
| 86 |
def generate_image(uploaded_image, prompt):
|
| 87 |
img = cv2.cvtColor(np.array(uploaded_image), cv2.COLOR_RGB2BGR)
|
|
|
|
| 109 |
fn=generate_image,
|
| 110 |
inputs=[gr.Image(type="pil", label="Upload Reference Image"), gr.Textbox(label="Enter Prompt", placeholder="e.g., A photorealistic astronaut in space")],
|
| 111 |
outputs=[gr.Textbox(label="Status"), gr.Image(label="Generated Image")],
|
| 112 |
+
title="Face Reference Image Generator (Kolors-IP-Adapter-FaceID-Plus)",
|
| 113 |
description="Upload an image with a face, enter a prompt, and generate a new image preserving the reference face."
|
| 114 |
)
|
| 115 |
|