Well, the mystery deepens. I am still not convinced that the stripes are primarily caused by the difference between AC vs. DC or something like that. For starters, here’s an image captured only with rather constant sunlight illumination (again, as above, top left RGB image, top right red channel, bot left green channel, bot right blue channel):
That is - here’s no artifical, possibly varying illumination involved at all. However, on close examination, you will notice the characteristic horizontal noise stripes, increasing in noise intensity toward the right image edge. They are noticable even in the green channel. (As it is one of the example images I made the raw data available for download (test .dng-file) a long time ago,you can have an own look into this with your favorite raw developer.)
So even if you use the HQ sensor in a classical image setting, you get this type of noise.
By the way, the color gains of this image are very similar to the color gains obtained with your “cob led” setup:
red/green/blue gains - 3.12 1.0 1.56
So… - as these stripes seem to occur with ordinary daylight illumination, I do not think that we see here the effect of varying illumination intensity caused by the scanner’s illumination. I still suspect the applied color gains to play a role here.
To give some more arguments: for sure the light output of your illumination should stay as constant as possible. Otherwise you certainly will encounter funny stripes in your captured image.
But, the way whitelight LEDs are working, one would not expect to see too much variations between the color channels. That is, if intensity variations happen because, for example, your power supply is not up to the task, all three channels should show a similar brightness variation. That is, we should mainly expect an overall brightness variation.
However we are currently discussing colored stripes - that is, stripes were the channels exhibit different noise levels, independent from each other. At least with your “cob LED” vs “lepro LED”, the image captured with the lower red gain did show less noise stripes. That’s my current state of analysis.
Here’s by the way an old capture of you, displayed like the images above:
If I remember correctly, you once stated that this was captured with the HQ sensor. Note that there are barely any noise stripes visible. Can you recall the circumstances of this capture? It differs at least in exposure time with respect to your other pictures (
1/50 vs. 1/600).