Update app.py

app.py

@@ -62,7 +62,7 @@ checkpoint = "microsoft/Phi-3.5-mini-instruct"
 #vae = AutoencoderKL.from_pretrained("ford442/sdxl-vae-bf16")
 vaeXL = AutoencoderKL.from_pretrained("stabilityai/sdxl-vae", safety_checker=None, use_safetensors=False) #, device_map='cpu') #.to(torch.bfloat16) #.to(device=device, dtype=torch.bfloat16)
 
-pipe = StableDiffusion3Pipeline.from_pretrained("ford442/stable-diffusion-3.5-medium-bf16").to(device=device, dtype=torch.bfloat16)
+pipe = StableDiffusion3Pipeline.from_pretrained("ford442/stable-diffusion-3.5-large-bf16").to(device=device, dtype=torch.bfloat16)
 #pipe = StableDiffusion3Pipeline.from_pretrained("ford442/stable-diffusion-3.5-medium-bf16").to(torch.device("cuda:0"))
 #pipe = StableDiffusion3Pipeline.from_pretrained("ford442/RealVis_Medium_1.0b_bf16", torch_dtype=torch.bfloat16)
 #pipe = StableDiffusion3Pipeline.from_pretrained("stabilityai/stable-diffusion-3.5-medium", token=hftoken, torch_dtype=torch.float32, device_map='balanced')
@@ -76,9 +76,9 @@ pipe = StableDiffusion3Pipeline.from_pretrained("ford442/stable-diffusion-3.5-me
 #pipe = torch.compile(pipe)
 # pipe.scheduler = EulerAncestralDiscreteScheduler.from_config(pipe.scheduler.config, beta_schedule="scaled_linear")
 
-refiner = StableDiffusionXLImg2ImgPipeline.from_pretrained("ford442/stable-diffusion-xl-refiner-1.0-bf16",vae = vaeXL, requires_aesthetics_score=True) #.to(torch.bfloat16)
+#refiner = StableDiffusionXLImg2ImgPipeline.from_pretrained("ford442/stable-diffusion-xl-refiner-1.0-bf16",vae = vaeXL, requires_aesthetics_score=True) #.to(torch.bfloat16)
 #refiner = StableDiffusionXLImg2ImgPipeline.from_pretrained("stabilityai/stable-diffusion-xl-refiner-1.0", vae=vae, torch_dtype=torch.float32, requires_aesthetics_score=True, device_map='balanced')
-refiner.scheduler=EulerAncestralDiscreteScheduler.from_config(refiner.scheduler.config)
+#refiner.scheduler=EulerAncestralDiscreteScheduler.from_config(refiner.scheduler.config)
 #refiner.enable_model_cpu_offload()
 
 #pipe.to(device=device, dtype=torch.bfloat16)
@@ -261,6 +261,7 @@ def infer(
     #torch.save(generated_latents, latent_path)
     #upload_to_ftp(latent_path)
     #refiner.scheduler.set_timesteps(num_inference_steps,device)
+    '''
     pipe.to(torch.device('cpu'))
     refiner.to(device=device, dtype=torch.bfloat16)
     refine = refiner(
@@ -276,16 +277,17 @@ def infer(
     refine.save(refine_path,optimize=False,compress_level=0)
     upload_to_ftp(refine_path)
     refiner.to(torch.device('cpu'))
+    '''
     upscaler_2.to(torch.device('cuda'))
     with torch.no_grad():
-        upscale2 = upscaler_2(refine, tiling=True, tile_width=256, tile_height=256)
+        upscale2 = upscaler_2(sd_image, tiling=True, tile_width=256, tile_height=256)
     print('-- got upscaled image --')
     upscaler_2.to(torch.device('cpu'))
     downscale2 = upscale2.resize((upscale2.width // 4, upscale2.height // 4),Image.LANCZOS)
     upscale_path = f"sd35_upscale_{seed}.png"
     downscale2.save(upscale_path,optimize=False,compress_level=0)
     upload_to_ftp(upscale_path)
-    return refine, seed, enhanced_prompt
+    return sd_image, seed, enhanced_prompt
 
 examples = [
     "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k",
@@ -335,7 +337,7 @@ def repeat_infer(
 
 with gr.Blocks(theme=gr.themes.Origin(),css=css) as demo:
     with gr.Column(elem_id="col-container"):
-        gr.Markdown(" # Text-to-Text-to-Image StableDiffusion 3.5 Medium (with refine)")
+        gr.Markdown(" # Text-to-Text-to-Image StableDiffusion 3.5 Large")
         expanded_prompt_output = gr.Textbox(label="Expanded Prompt", lines=5)  # Add this line
         with gr.Row():
             prompt = gr.Text(
@@ -343,7 +345,6 @@ with gr.Blocks(theme=gr.themes.Origin(),css=css) as demo:
                 show_label=False,
                 max_lines=1,
                 placeholder="Enter your prompt",
-                value="A captivating Christmas scene.",
                 container=False,
             )
             options = [True, False]