Training a new model with different scaling factor

[chhsieh0309]

Hi there,

If I want to train a new ESRGAN model, a 2.5x upsampling/downsampling factor, for example, how do I get the corresponding pre-trained PSNR model?

[xinntao]

I think you can use the weights of pre-trained X4 ESRGAN to initialize the X2.5 model, except the upsampling layer.

[zhangcl]

@chhsieh0309 have you successfully trained the model with 2.5x scale factor? I was able to train with scale factor of integer values (e.g. 2x, 4x), but not sure how to train a model with non-integer scale factor (e.g. 1.5x, 2.5x).

[jorjiang]

@chhsieh0309 have you successfully trained the model with 2.5x scale factor? I was able to train with scale factor of integer values (e.g. 2x, 4x), but not sure how to train a model with non-integer scale factor (e.g. 1.5x, 2.5x).

why not just use 4x and then downsample to 2.5