size error in Harmonization mode, discrepancy btw noise and input image shapes

[nikjetchev]

Dear all

I am trying to use the harmonization example. I specified images as required.

However, I get an error in SinGAN_generate(), the shapes do not match on the code line

z_in = noise_amp(z_curr)+I_prev*

I wonder, should the harmonization image I insert have exactly the same size as the training image? Or is there a way to make it work with any size image for harmonization, the model architecture seems flexible enough. But I am a bit confused by the overall code structure and where these tensor size changes should be refactored exactly

[xuxiangsun]

Sir，I also want to use the harmonization. But the first problem is that i do know how to get naively pasted images and its mask. How do you realize it? @nikjetchev

[tamarott]

In the current version of the code, the harmonized image should have the same size of the original image.

z_curr is the current noise input sample. in order to work with arbitrary image harmonization size you should change it size according to the image size of each scale of the model.

[xuxiangsun]

In the current version of the code, the harmonized image should have the same size of the original image.

z_curr is the current noise input sample. in order to work with arbitrary image harmonization size you should change it size according to the image size of each scale of the model.

Dear madam,Can you share with me the method you use to get naively pasted images and its mask?I don't know how to realize it. Please……

[tamarott]

you can create the naively pasted version and the mask manually, using photoshop/paint.net or similar programs