At the top right corner of the image in https://discuss.pixls.us/t/playraw-venise-la-serenissime/8571, there is an overwhelming amount of sun. Wherever the rays touch, I see more and more of the rectangles as I push the image. A wild guess is that it has to do with the demosaicing; otherwise, something on the camera or in between the glass and the scene…
From my entry.
From my hessian norm filter (normalized). Edit: filter done on RT output: DSCF1393.RAF.pp3 (11.0 KB).
Somehow I know this effect from another picture. Afaik it is not caused by demosaicing, but by someting in between the camera and the lens. But I’m not sure, because it’s a while ago, I’ve seen this…
Indeed as Info said, it’s a known issue of the xtans CFA, under some circumstances the light rays hit the sensor at certain angle and cause this pattern.
First search link of “masked phase detect pixels” led me to this DPREVIEW forum reply for topic “Sony A7r2 - horizontal banding ??”:
What you’re seeing are artifacts from the phase detect pixels on the image sensor. That is why you don’t see them with the Nikon D810 images.
These phase detect pixels just like regular pixels except they are half masked in order to detect the phase of incoming light. The camera still uses them to form part of the image, but must make adjustments for the difference in sensitivity of these pixels.
It seems like the adjustments break down in some way when pixels are very close to being overexposed, possibly because they have a different clipping point. […]
@afre I think that’s a bit different problem – or perhaps the same but with different symptoms due to different CFA and PDAF pixel layouts (I’m totally ignorant on the topic, in fact…). If you search “PDAF striping” you will find a lot of info on dpreview (including a couple of references to the workaround implemented in RT)
