I have 2 images (raw), the first one looks normal, the second one with some color artifacts. Shot one after another, the camera and the guy position was slightly changed (rotation) between images.
I opened the second image with a few editors, the artifacts are present, so I assume that the camera introduced this.
The lens had a hood and lens protection glass.
Do you have a clue what this might be? 
Introduced by the protection glass?
This is called Moire, and you can read more about it here: Moiré pattern - Wikipedia
It’s a Moiré pattern which is a type of aliasing. The fine detail in the jacket is too fine for the sensor. In technical terms, you’ve exceeded the sensor Nyquist limit.
The lens MTF may be too high for the sensor and may need an optical low pass filter.
You may be able to reduce the aliasing by adding a digital low pass filter but it’s not always possible.
A couple more papers explaining what’s happening with less maths
Just guessing here - would VNG4 demosaic do enough to mitigate this?
Edit: I think a diffusion filter should help with Moiré. Something like a Black Pro Mist 1/8.
That was my guess, but I didn’t want to answer not having solid knowledge.
Moiré has been well explained here and I don’t need to add to that. BUT, I wonder if electronic shutter vs mechanical shutter makes any difference to Moiré. Probably not, but just thinking out loud.
Maybe read the articles posted…
EDIT: No mechanical or electronic shutter makes no different to moiré. I was just having a brain fart idea at 4am and thought I would share it. Sorry if this noise in the thread upset anyone.
The issue is that an alias once created looks exactly like the wanted signal. If you filter the image to reduce the aliasing you also reduce the wanted fine detail. You need to reduce the fine detail before it hits the sensor by optical filtering or changing zoom level.
There are some AI tools that claim to recognise aliasing and create a fixed image, but classical sampling theory goes against this.
Here are some hints, how to remove those moiré patterns in post-processing:
in Darktable Diffuse or sharpen can deal with this effect pretty well.
Just set it to bloom preset with at least 4 iterations and change the blend mode to chromaticity.
Local masking of the moire parts should also be used otherwise this method reduces color contrast of the whole image.
Thank you all for the valuable advice. So, basically, diffusion & chromaticity blending will fix this.
I tried also to change the demosaic method but it didn’t help in this case (usually I’m using default RCD)
I tried the diffuse and sharpen method. On this image it works perfectly. But I then I tried it on some other images (without moirés) and on some I produced an unwanted glow effect on top of the already teasered loss of chromacity.
So use it with care and open eyes.
Switching the demosaicing isn’t enough. You have to use heavy chroma denoising as well.
All moire related problems are pretty difficult to correct. As mentioned some demosaicers work better than others; vng4, lmmse or amze are often more stable. rawtherapee offers IGV which is also well suited but gives less details.
All try-to-fix methods after demosaicing come with a hefty quality loss. @Iain and @garagecoder from gmic team worked on this some time ago, not sure how far they got. Maybe we can do some fresh dt work on this …?
BTW - i have only a very limited collection of raw images with such moire problems - would be very interested to have more 
No, it may remove the aliasing (and wanted parts of the image). It won’t work on every image and you may need to soften it so much it becomes unusable.
