Expected behavior: eyes and mouth conform to src image.
Actually: generated face keeps reverting to a generalized face, often not even close to the direction of the src, It will stat looking like what it is supposed to but then revert. I am on 144k iter and my faces are wrong.
Expected: Scr & Dst images to become defined and clear to see in training.
Actually: After 144k iter src & dts models are still very blurry and occationally warp to odd dimensions that carry over to the combined image.
Dims are all default values, mask training on, eye and mouth priority on, warp on.
Expected behavior: eyes and mouth conform to src image. Actually: generated face keeps reverting to a generalized face, often not even close to the direction of the src, It will stat looking like what it is supposed to but then revert. I am on 144k iter and my faces are wrong.
Expected: Scr & Dst images to become defined and clear to see in training. Actually: After 144k iter src & dts models are still very blurry and occationally warp to odd dimensions that carry over to the combined image.
Dims are all default values, mask training on, eye and mouth priority on, warp on.