test gradio

app.py

@@ -7,44 +7,34 @@ from diffusers import StableDiffusionXLPipeline
 from PIL import Image
 import torch
 from diffusers import AutoPipelineForText2Image, DDIMScheduler
-from transformers import CLIPVisionModelWithProjection
+from diffusers import AutoPipelineForText2Image
 from diffusers.utils import load_image
+import torch
 
 token = os.getenv("HF_TOKEN")
 login(token=token)
 
-image_encoder = CLIPVisionModelWithProjection.from_pretrained(
-    "h94/IP-Adapter",
-    subfolder="models/image_encoder",
-    torch_dtype=torch.float16,
-)
 
-pipeline = AutoPipelineForText2Image.from_pretrained(
-    "stabilityai/stable-diffusion-xl-base-1.0",
-    torch_dtype=torch.float16,
-    image_encoder=image_encoder,
-)
-pipeline.scheduler = DDIMScheduler.from_config(pipeline.scheduler.config)
-pipeline.load_ip_adapter(
-  "h94/IP-Adapter",
-  subfolder="sdxl_models",
-  weight_name=["ip-adapter-plus_sdxl_vit-h.safetensors"]
-)
-pipeline.set_ip_adapter_scale([0.6])
-pipeline.enable_model_cpu_offload()
+pipeline = AutoPipelineForText2Image.from_pretrained("stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16).to("cuda")
+pipeline.load_ip_adapter("h94/IP-Adapter", subfolder="sdxl_models", weight_name="ip-adapter_sdxl.bin")
+
 
 
 @spaces.GPU
 def generate_image(prompt, reference_image, controlnet_conditioning_scale):
-    reference_image = Image.open(reference_image)
+    style_image = Image.open(reference_image)
     # reference_image.resize((512, 512))
-    pipeline.set_ip_adapter_scale([controlnet_conditioning_scale])
+    scale = {
+        "up": {"block_0": [0.0, controlnet_conditioning_scale, 0.0]},
+    }
+    pipeline.set_ip_adapter_scale(scale)
 
     image = pipeline(
         prompt=prompt,
-        ip_adapter_image=[reference_image],
-        negative_prompt="monochrome, lowres, bad anatomy, worst quality, low quality",
-        num_inference_steps=50, num_images_per_prompt=1,
+        ip_adapter_image=style_image,
+        negative_prompt="",
+        guidance_scale=5,
+        num_inference_steps=30,
     ).images[0]
 
     return image