Heya,
I made a curious discovery while playing with vkdts filmsim. In photos I took outside today the filmsim introduces a striping artifact. When disabling “printing” and only viewing the negative they seem to be missing. Any any ideas on what might be the cause? Happens both in latest stable and nightly. Even when pushing various local contrast enhancements in darktable I could not see these artifacts.
hm don’t have any evidence that the splatting code would produce stripes in the weights or anything, some sanity checking looks all good here. from what i can tell the stripes are present in the blue channel of the input data. rcd seems to successfully work around this, maybe by larger context window?
