Regarding inpainting performance

[hsollee0408]

Hi, I was wondering if you’ve had a chance to look into inpainting as well. I tried implementing a naive inpainting approach, but the performance wasn't quite what I was hoping for. I'd appreciate any thoughts or guidance you might have.

[jeongsol-kim]

Hi, yes we can solve inpainting problem.
Apply a projection operation for the data-consistency update. Maintain following computations such as interpolation and noise addition. Projection is defined by z0y = (1-mask)*z0t + mask*y where the masked region to be filled is assumed to have zeros