Spaces:
				
			
			
	
			
			
		Runtime error
		
	
	
	
			
			
	
	
	
	
		
		
		Runtime error
		
	
		layerdiffusion
		
	commited on
		
		
					Commit 
							
							·
						
						5a58ccb
	
1
								Parent(s):
							
							06fccba
								
Update README.md
Browse files
    	
        README.md
    CHANGED
    
    | 
         @@ -38,7 +38,7 @@ If you do not have required computation devices and still wants an online soluti 
     | 
|
| 38 | 
         | 
| 39 | 
         
             
            We currently release two models `paints_undo_single_frame` and `paints_undo_multi_frame`. Let's call them single-frame model and multi-frame model.
         
     | 
| 40 | 
         | 
| 41 | 
         
            -
            The single-frame model takes one image and an `operation step` as input, and outputs one single image. Assuming that an artwork can always be created with 1000 human operations (for example, one brush stroke is one operation), and the `operation step` is an int number from 0 to 999. The number 0 is the finished final artwork, and the number 999 is the first brush stroke drawn on the pure white canvas. You can understand this model as an "undo" (or called Ctrl+Z) model. You input the final image, and indicate how many times you want to "Ctrl+Z", and the model will give you a "simulated" screenshot after those "Ctrl+Z"s are pressed. If your `operation step` is 100, then it means you want to simulate "Ctrl+Z" 100 times on this image get the appearance after the 100-th "Ctrl+Z".
         
     | 
| 42 | 
         | 
| 43 | 
         
             
            The multi-frame model takes two images as inputs and output 16 intermediate frames between the two input images. The result is much more consistent than the single-frame model, but also much slower, less "creative", and limited in 16 frames.
         
     | 
| 44 | 
         | 
| 
         | 
|
| 38 | 
         | 
| 39 | 
         
             
            We currently release two models `paints_undo_single_frame` and `paints_undo_multi_frame`. Let's call them single-frame model and multi-frame model.
         
     | 
| 40 | 
         | 
| 41 | 
         
            +
            The single-frame model takes one image and an `operation step` as input, and outputs one single image. Assuming that an artwork can always be created with 1000 human operations (for example, one brush stroke is one operation), and the `operation step` is an int number from 0 to 999. The number 0 is the finished final artwork, and the number 999 is the first brush stroke drawn on the pure white canvas. You can understand this model as an "undo" (or called Ctrl+Z) model. You input the final image, and indicate how many times you want to "Ctrl+Z", and the model will give you a "simulated" screenshot after those "Ctrl+Z"s are pressed. If your `operation step` is 100, then it means you want to simulate "Ctrl+Z" 100 times on this image to get the appearance after the 100-th "Ctrl+Z".
         
     | 
| 42 | 
         | 
| 43 | 
         
             
            The multi-frame model takes two images as inputs and output 16 intermediate frames between the two input images. The result is much more consistent than the single-frame model, but also much slower, less "creative", and limited in 16 frames.
         
     | 
| 44 | 
         |