up

app.py

@@ -22,8 +22,7 @@ from diffusers import StableDiffusionUpscalePipeline
 from diffusers import LDMSuperResolutionPipeline
 import cv2
 import onnxruntime
-# import xformers
-# from xformers.ops import MemoryEfficientAttentionFlashAttentionOp
+from split_image import split
 
 def removeFurniture(input_img1,
             input_img2,
@@ -306,11 +305,20 @@ def upscale_image(img, rows, cols, seed, prompt, negative_prompt, xformers, cpu_
     # Save the merged image
     return final_img
 
-
-def upscale( image, prompt):
-    print("upscale", image, prompt)   
+    
+def upscale( image, prompt, negative_prompt, rows, guidance, iterations, xformers_input, cpu_offload_input, attention_slicing_input):
+    print("upscale", image, prompt, negative_prompt, rows, guidance, iterations, xformers_input, cpu_offload_input, attention_slicing_input)   
     # return upscale1(image, prompt)
-    return upscale_image(image,rows=4,cols=4,seed=-1,prompt=prompt,negative_prompt="jpeg artifacts, lowres, bad quality, watermark",xformers=True,cpu_offload=True,attention_slicing=True,iterations=10)
+    return upscale_image(image,
+                         rows=rows,cols=rows,
+                         seed=-1,
+                         prompt=prompt,
+                         guidance=guidance,
+                         negative_prompt=negative_prompt,
+                         xformers=xformers_input,
+                         cpu_offload=cpu_offload_input,
+                         attention_slicing=attention_slicing_input,
+                         iterations=iterations)
 
 modes = {
     '1': '1',
@@ -359,66 +367,19 @@ with gr.Blocks() as app:
             gr.Button("Upscale").click(
                 upscale, 
                 inputs=[
-                    gr.Image(),
-                    gr.Textbox(label="prompt",value="empty room")
+                    gr.Image(label="Source Image to upscale"),
+                    gr.Textbox(label="prompt",value="empty room"),
+                    gr.Textbox(label="negative prompt",value="jpeg artifacts, lowres, bad quality, watermark, text"),
+                    gr.Number(value=2, label="Tile grid dimension amount (number of rows and columns) - X by X "),
+                    gr.Slider(2, 15, 7, step=1, label='Guidance Scale: How much the AI influences the Upscaling.'),
+                    gr.Slider(10, 100, 10, step=1, label='Number of Iterations'),
+                    gr.Checkbox(value=True,label="Enable Xformers memory efficient attention"),                    
+                    gr.Checkbox(value=True,label="Enable sequential CPU offload"),
+                    gr.Checkbox(value=True,label="Enable attention slicing")
                     ], 
                 outputs=gr.Image())
 
 
-
-    # with gr.Row():
-        
-    #     with gr.Column(scale=55):
-    #       with gr.Group():
-    #           with gr.Row():
-    #             prompt = gr.Textbox(label="Prompt", show_label=False, max_lines=2,placeholder=f"Enter prompt")
-    #             generate = gr.Button(value="Generate")
-
-    #           gallery = gr.Gallery(label="Generated images", show_label=False)
-    #       state_info = gr.Textbox(label="State", show_label=False, max_lines=2)
-    #       error_output = gr.Markdown(visible=False)
-
-    #     with gr.Column(scale=45):
-    #       inf_mode = gr.Radio(label="Inference Mode", choices=list(modes.values())[:4], value=modes['txt2img']) # TODO remove [:3] limit
-          
-    #       with gr.Group(visible=False) as i2i_options:
-    #         image = gr.Image(label="Image", height=128, type="pil")
-    #         inpaint_info = gr.Markdown("Inpainting resizes and pads images to 512x512", visible=False)
-    #         upscale_info = gr.Markdown("""Best for small images (128x128 or smaller).<br>
-    #                                     Bigger images will be sliced into 128x128 tiles which will be upscaled individually.<br>
-    #                                     This is done to avoid running out of GPU memory.""", visible=False)
-    #         strength = gr.Slider(label="Transformation strength", minimum=0, maximum=1, step=0.01, value=0.5)
-
-    #       with gr.Group():
-    #         neg_prompt = gr.Textbox(label="Negative prompt", placeholder="What to exclude from the image")
-
-    #         n_images = gr.Slider(label="Number of images", value=1, minimum=1, maximum=4, step=1)
-    #         with gr.Row():
-    #           guidance = gr.Slider(label="Guidance scale", value=7.5, maximum=15)
-    #           steps = gr.Slider(label="Steps", value=current_steps, minimum=2, maximum=100, step=1)
-
-    #         with gr.Row():
-    #           width = gr.Slider(label="Width", value=768, minimum=64, maximum=1024, step=8)
-    #           height = gr.Slider(label="Height", value=768, minimum=64, maximum=1024, step=8)
-
-    #         seed = gr.Slider(0, 2147483647, label='Seed (0 = random)', value=0, step=1)
-    #         with gr.Accordion("Memory optimization"):
-    #           attn_slicing = gr.Checkbox(label="Attention slicing (a bit slower, but uses less memory)", value=attn_slicing_enabled)
-    #           # mem_eff_attn = gr.Checkbox(label="Memory efficient attention (xformers)", value=mem_eff_attn_enabled)
-
-    # inf_mode.change(on_mode_change, inputs=[inf_mode], outputs=[i2i_options, inpaint_info, upscale_info, strength], queue=False)
-    # steps.change(on_steps_change, inputs=[steps], outputs=[], queue=False)
-    # attn_slicing.change(lambda x: switch_attention_slicing(x), inputs=[attn_slicing], queue=False)
-    # # mem_eff_attn.change(lambda x: switch_mem_eff_attn(x), inputs=[mem_eff_attn], queue=False)
-
-    # inputs = [inf_mode, prompt, n_images, guidance, steps, width, height, seed, image, strength, neg_prompt]
-    # outputs = [gallery, error_output]
-    # prompt.submit(inference, inputs=inputs, outputs=outputs)
-    # generate.click(inference, inputs=inputs, outputs=outputs)
-
-    # app.load(update_state_info, inputs=state_info, outputs=state_info, every=0.5, show_progress=False)
-
-
 app.queue()
 app.launch(debug=True,share=True, height=768)
 

requirements.txt

@@ -2,11 +2,12 @@ diffusers==0.15.0
 xformers==0.0.16
 transformers==4.28.0
 torchvision==0.14.1
-git+https://github.com/huggingface/accelerate.git
+accelerate
 opencv-python-headless==4.7.0.72
 scipy==1.10.0
 python-docx
 triton
 altair<5
 gradio
-onnxruntime==1.12.0
+onnxruntime==1.12.0
+split_image