Update app.py

app.py

@@ -13,6 +13,7 @@ import numpy as np
 import cv2
 
 vae = AutoencoderKL.from_pretrained("madebyollin/sdxl-vae-fp16-fix", torch_dtype=torch.float16)
+
 controlnet = ControlNetModel.from_pretrained(
     "diffusers/controlnet-canny-sdxl-1.0",
     torch_dtype=torch.float16
@@ -26,21 +27,15 @@ pipe = StableDiffusionXLControlNetPipeline.from_pretrained(
     variant="fp16",
     use_safetensors=True
 )
+
 pipe.to("cuda")
-generator = torch.Generator(device="cuda")
 
-#pipe.enable_model_cpu_offload()
 
-def infer(use_custom_model, model_name, custom_lora_weight, image_in, prompt, negative_prompt, preprocessor, controlnet_conditioning_scale, guidance_scale, seed):
-    if use_custom_model:
-        custom_model = model_name
 
-        # This is where you load your trained weights
-        pipe.load_lora_weights(custom_model, use_auth_token=True)
-    
-    prompt = prompt
-    negative_prompt = negative_prompt
+#pipe.enable_model_cpu_offload()
 
+def infer(use_custom_model, model_name, custom_lora_weight, image_in, prompt, negative_prompt, preprocessor, controlnet_conditioning_scale, guidance_scale, steps, seed, progress=gr.Progress(track_tqdm=True)):
+    
     if preprocessor == "canny":
 
         image = load_image(image_in)
@@ -50,6 +45,16 @@ def infer(use_custom_model, model_name, custom_lora_weight, image_in, prompt, ne
         image = image[:, :, None]
         image = np.concatenate([image, image, image], axis=2)
         image = Image.fromarray(image)
+    
+    if use_custom_model:
+        custom_model = model_name
+
+        # This is where you load your trained weights
+        pipe.load_lora_weights(custom_model, use_auth_token=True)
+    
+    prompt = prompt
+    negative_prompt = negative_prompt
+    generator = torch.Generator(device="cuda").manual_seed(seed)
 
     if use_custom_model:
         lora_scale=custom_lora_weight
@@ -60,8 +65,8 @@ def infer(use_custom_model, model_name, custom_lora_weight, image_in, prompt, ne
             image=image, 
             controlnet_conditioning_scale=controlnet_conditioning_scale,
             guidance_scale = guidance_scale,
-            num_inference_steps=50,
-            generator=generator.manual_seed(seed),
+            num_inference_steps=steps,
+            generator=generator,
             cross_attention_kwargs={"scale": lora_scale}
         ).images
     else:
@@ -71,8 +76,8 @@ def infer(use_custom_model, model_name, custom_lora_weight, image_in, prompt, ne
             image=image, 
             controlnet_conditioning_scale=controlnet_conditioning_scale,
             guidance_scale = guidance_scale,
-            num_inference_steps=50,
-            generator=generator.manual_seed(seed),
+            num_inference_steps=steps,
+            generator=generator,
         ).images
 
     images[0].save(f"result.png")
@@ -88,9 +93,9 @@ css="""
 """
 with gr.Blocks(css=css) as demo:
     with gr.Column(elem_id="col-container"):
-        gr.Markdown("""
-# SD-XL Control LoRas
-Use StableDiffusion XL with ControlNet pretrained LoRas 
+        gr.HTML("""
+<h2 style="text-align: center;>SD-XL Control LoRas</h2>
+<p style="text-align: center;">Use StableDiffusion XL with <a href="https://huggingface.co/collections/diffusers/sdxl-controlnets-64f9c35846f3f06f5abe351f">Diffusers' SDXL ControlNets</a></p>
 
         """)
         
@@ -100,6 +105,7 @@ Use StableDiffusion XL with ControlNet pretrained LoRas
                 prompt = gr.Textbox(label="Prompt")
                 negative_prompt = gr.Textbox(label="Negative prompt", value="extra digit, fewer digits, cropped, worst quality, low quality, glitch, deformed, mutated, ugly, disfigured")
                 guidance_scale = gr.Slider(label="Guidance Scale", minimum=1.0, maximum=10.0, step=0.1, value=7.5)
+                steps = gr.Slider(label="Inference Steps", minimum="25", maximum="50", step=1, value=25)
             with gr.Column():
                 preprocessor = gr.Dropdown(label="Preprocessor", choices=["canny"], value="canny", interactive=False, info="For the moment, only canny is available")  
                 controlnet_conditioning_scale = gr.Slider(label="Controlnet conditioning Scale", minimum=0.1, maximum=0.9, step=0.01, value=0.5, type="float")
@@ -113,7 +119,7 @@ Use StableDiffusion XL with ControlNet pretrained LoRas
 
     submit_btn.click(
         fn = infer,
-        inputs = [use_custom_model, model_name, custom_lora_weight, image_in, prompt, negative_prompt, preprocessor, controlnet_conditioning_scale, guidance_scale, seed],
+        inputs = [use_custom_model, model_name, custom_lora_weight, image_in, prompt, negative_prompt, preprocessor, controlnet_conditioning_scale, guidance_scale, steps, seed],
         outputs = [result]
     )