>>4469085
CFAs are awful for detail rendering. they basically get confused because each pixel only contains partial information (even film scans have full info for each pixel). this is a small part of why film managed to render detail better for nature and portraits. the other part is the finer the detail on film, the lower the contrast. on digital contrast is constant down to the nyquist limit (aka where the camera spits out rainbow garbage for the noise reduction algorithm to attempt to fix) which makes things look flat and textures look harsh. notably film is nearly immune to the "sub pixel detail" problem of digital where subtle textures totally disappear. on film the grains can be six times smaller than a pixel on the nikon ZF, so it's effectively 6x the resolution with a contrast falloff effect.

back to the xtrans problem. the bigger the CFA's pattern, the greater the resolution loss because the less information from adjacent pixels can be used to infer the real color and luminance of each individual digibox. bayer 2x2 is already pretty bad. xtrans is a larger pattern and slightly blurrier, most noticeably on red textures. being aps-c also hurts it because digital cameras with smaller pixels need sharper and sharper lenses to render smaller and smaller details, contributing to the shittiness. on film most people shoot at least full frame but go for 6x6.

this size advantage adds even more goodness because using a less corrected lens with a larger format renders a smoother but still looking image with better bokeh and more graceful in focus/out of focus transitions than using a highly corrected "clinical" lens like an XF 23mm f/1.4 WR or literally anything for micro four thirds.