Update app.py

app.py

@@ -1,13 +1,18 @@
 import gradio as gr
-from diffusers import AutoPipelineForText2Image, AutoencoderKL
+from diffusers import AutoPipelineForText2Image, AutoPipelineForImage2Image, AutoencoderKL
 from diffusers.utils import load_image
 import torch
 
 vae = AutoencoderKL.from_pretrained("madebyollin/sdxl-vae-fp16-fix", torch_dtype=torch.float16)
+
 text_pipeline = AutoPipelineForText2Image.from_pretrained("stabilityai/stable-diffusion-xl-base-1.0", vae=vae, torch_dtype=torch.float16, variant="fp16", use_safetensors=True).to("cuda")
 text_pipeline.load_ip_adapter("h94/IP-Adapter", subfolder="sdxl_models", weight_name="ip-adapter_sdxl.bin")
 text_pipeline.set_ip_adapter_scale(0.6)
 
+image_pipeline = AutoPipelineForImage2Image.from_pretrained("stabilityai/stable-diffusion-xl-base-1.0", vae=vae, torch_dtype=torch.float16, variant="fp16", use_safetensors=True).to("cuda")
+image_pipeline.load_ip_adapter("h94/IP-Adapter", subfolder="sdxl_models", weight_name="ip-adapter_sdxl.bin")
+image_pipeline.set_ip_adapter_scale(0.6)
+
 
 def text_to_image(ip, prompt, neg_prompt, width, height, ip_scale, strength, guidance, steps):
     ip.thumbnail((1024, 1024))
@@ -28,6 +33,27 @@ def text_to_image(ip, prompt, neg_prompt, width, height, ip_scale, strength, gui
     return images[0]
 
 
+def image_to_image(ip, image, prompt, neg_prompt, width, height, ip_scale, strength, guidance, steps):
+    ip.thumbnail((1024, 1024))
+    image.thumbnail((1024, 1024))
+    
+    image_pipeline.set_ip_adapter_scale(ip_scale)
+    
+    images = image_pipeline(
+        prompt=prompt,
+        image=image,
+        ip_adapter_image=ip,
+        negative_prompt=neg_prompt,
+        width=width,
+        height=height,
+        strength=strength,
+        guidance_scale=guidance,
+        num_inference_steps=steps,
+    ).images
+    
+    return images[0]
+
+
 with gr.Blocks() as demo:
     gr.Markdown("""
     # IP-Adapter Playground
@@ -62,5 +88,6 @@ with gr.Blocks() as demo:
         steps_slider = gr.Slider(50, 100, value=75, step=1, label="Steps")
 
     text_button.click(text_to_image, inputs=[text_ip, text_prompt, neg_prompt, width_slider, height_slider, ip_scale_slider, strength_slider, guidance_slider, steps_slider], outputs=output_image)
+    image_button.click(image_to_image, inputs=[image_ip, image_image, image_prompt, neg_prompt, width_slider, height_slider, ip_scale_slider, strength_slider, guidance_slider, steps_slider], outputs=output_image)
 
 demo.launch()