Good evening Steve,
Thank you for your answer and the link to that interesting article.
I am very aware that in reality a spectrogram of a LED light is not that smooth and is very peaky because of the way it's light is emmited via narrow bands in the light spectrum.
One should be able to use a meter like the Sekonic C-700R SPECTROMASTER or other professional meters to see pretty precise how the spectrogram looks like.
(I studied math, science and photography, so I know a few things about light and technic, at least I think so, but I am eager to learn more every day so any link to interesting articles is always very welcome)
Maybe more about this in another topic.
But what I cannot explain in the case of this problem I have, is this.
I have done some extra tests to illustrate that the problem, at least to my knowledge, lies solely in the camera.
Follow what I did, please.
The only light source in all the tests were the IKEA LED lights.
- I have set the IKEA LED lights to neutral and saved a custom WB setting in slot 1
- made several shots one after another without pixel shift and all had exactly the same settings, exposure and had the same neutral color, no difference at all
- did the same test with pixel shift enabled and made many photo's, again exactly the same results, all were the same and perfect
- made a whole serie of correct exposed images with different shutter speeds, from 1/200 of a second up to 4 seconds, no difference at all and all looked perfect, not the slightest change in exposure or color, all with pixel shift enabled
In all these circumstances (more than 60 test shots were made) the result was: :----------------------------->perfect whites and a very neutral image without AND with pixel shift, even with shutter times up to 1/200st of a second with increments in 1/3 of a stop up to 4".
Then I did the next test:
- changed the IKEA LED light to warm white and made a new custom WB and stored it in K2
- made several shots one after another without pixel shift and all had exactly the same settings and had the same neutral color, no difference at all between them, even with that warm LED light this time
- did the same test with pixel shift enabled and made many photo's, all pictures had the same yellow color and the same abnormal histogram as seen in picture 3 of the screenshots in my earlier post
- made a whole serie of images with different shutter speeds, again from 1/200 of a second to 4 seconds, no difference as above and all stayed equally yellow when pixel shift was enabled
In all these circumstances with the warm LED light, only where pixel shift was enabled, this time the result was: :----------------------------->a completly yellow image and very abnormal looking histogram, but no variations between them individualy.
Now think with me please. If the LED lights would cause possibly so much fluctuations, at least one image should look different, but none did.
But, all yellow looking images looked good in RawTherapee, without any change. Just importing and instantly they showed up with the right colors like all the other normal pictures on my camera.
To me, this is a problem that is caused only in-camera; otherwise the RAW data would show the same yellow results in an external editor as RT and this is absolutely not the case.
RT is set to camera WB, so things should stay the same.
More important, from each ps image, the four individual pixel shift images all look perfect in RT and there is no visible difference between any of those four when you compare them.
Only the miniatures that are imported with the RAW files indeed also exhibits the same yellow color.
As soon as I open the RAW files in RT, the miniatures in the browser panel of RT are replaced with normal colors, see also my earlier post and pictures.
All images, once inside RT, look exactly the same, pixel shifted or non pixel shifted, with warm LED light or neutral LED light.
Only my camera shows these strange colors under exactly these circumstances on my K1 display.
Also Windows explorer shows the same DNG miniatures as yellow pictures with the problematic files, I suppose because they come directly from the camera together with the RAW files?
Don't forget that I already wrote that while taking the ps shots, when they show up as 1...2...3...4 (from top left to bottom right) on my K1 display, all four images appear as perfectly normal colors.
(I know how to look at colors because as a professional photographer I have made about twelve thousand professional prints for a photo studio and worked in a studio for almost ten years)
If the properties of those cheap lights should play any important role in this problem, how is it possible that every individual file of a ps image is perfectly exposed and has perfect colors?
Why does it not happen anymore as soon as I change those same LED's from warm to neutral? Once they are set to neutral, the pictures with ps are all fine, also in-camera I mean.
How can these colors become so very different, but only after the camera has done it's math? This is almost certain a bug.
I can switch off pixel shift and have a perfect image, switch it back on and... bam, a yellow image, off again, normal image, on again, well you get the picture...
Every single time again I get the same results, with no color variation even between the "faulty" ones. I did about 120 shots to test if the results were constant and they are.
Nothing has changed between different pictures, except the pixel shift on or off and the change from warm white to neutral LED setting.
Shutter speeds seems to have not the slightest effect, ISO neither, I tested this also to rule out every other aspect.
Not that I want to push things here to get my answer, but it doesn't make sense to me and I always want to understand why things are happening, it is so intriguing to me.
If someone who knows how the order of calculations is done in a camera like my K1, maybe we can find out what possibly can happen and why only under such specific circumstances.
After all, It is not a random affair that is happening now and then and with each time different results, like you would expect from the complex LED's behaviour.
If I can get constant results with these lights with shutter speeds ranging from 1/200 sec to 4 sec, then I think we can conclude that these light sources are pretty stable.
This problem is reproducable every time again and the colors are perfect, or completely wrong and yellow with always the same wrong histogram as posted before.
If a non pixel shifted image can be exposed correct with a perfect set custom manual WB, then a pixel shifted one directly taken after it under the same circumstances should look exactly the same with that same WB, except a little bit sharper, with a little bit less noise and a little more dynamic range. I don't think anyone can argue with that.
And indeed if I make a whole serie of the same image, there is no difference between any of them.
But again, this is no longer a problem for me, I know what to do and how it only seems to be a problem with the way my camera shows those specific files under those very specific circumstances.
I can perfectly use them in RT and they look exactly like the non ps versions that I took right before and without any work that need to be done. Just importing in RT and they look fine.
Sorry for the long post and thank you all for your patience.
Again Steve, thank you for your help so far.
Have a very good night.
Sven
Added information:
To exclude the possibility that RT should perhaps also play a role in this, I opened the same yellow files in Camera raw 8.5 and the images instantly looked perfect as soon as they were open.
This rules out that there is any problem with the file itself. It can only be the camera.