MinusOneOneDenoiserWrapper

class deepinv.models.MinusOneOneDenoiserWrapper(model, xmin=0.0, xmax=1.0)

Bases: Module

A wrapper for denoisers trained on \([x_{\mathrm{min}}, x_{\mathrm{max}}]\) images to be used with math:[-1, 1] images, i.e. on diffusion sampling iterates.

Parameters:

• denoiser (deepinv.models.Denoiser) – the denoiser to be wrapped.

• xmin (float) – minimum value of the denoiser training range. Default to 0.0.

• xmax (float) – maximum value of the denoiser training range. Default to 1.0.