CMOS camera image processing problem

[centroid]

I successfully imaged for 12 years with SX CCD cameras. See examples here: https://davesimaging.wixsite.com/mysite

 

I now have a 294c CMOS based camera, and if I demosaic (debayer) a single image (sub), I get colour, nowhere near as accurate as with CCD, but that is adjustable.

 

However, the problem is, if I align and 'stack' (sum) a batch of images, I get a mono result, which won't debayer to an RGB image.

 

Almost certainly me doing something wrong somewhere, but It has me mystified.

 

Also, if I take a single de-bayered colour image into photoshop, it appears as a mono (greyscale) image, albeit, in Astroart it appears in colour.

 

I'm not sure what the white Like/Answer/Share box is below, not seen that on the forum before?

Like

 

 

Answer

 

Share

 

 

Edited January 14, 2022 by centroid

[PeterCPC]

What are you using to stack the images?

[vlaiv]

4 minutes ago, centroid said:

However, the problem is, if I align and 'stack' (sum) a batch of images, I get a mono result, which won't debayer to an RGB image.

You need to stack debayered version of your subs where colors are already split

4 minutes ago, centroid said:

I'm not sure what the white Like/Answer/Share box is below, not seen that on the forum before?

Maybe you copy/pasted above from some other website you are using and by accident you also copied their "social tools" box?

[centroid]

9 minutes ago, PeterCPC said:

What are you using to stack the images?

Astroart 8

[centroid]

21 minutes ago, vlaiv said:

You need to stack debayered version of your subs where colors are already split

Maybe you copy/pasted above from some other website you are using and by accident you also copied their "social tools" box?

Ah, that is probably the answer Vlaiv. The images were non debayered. I have another 'play' this afternoon.

CMOS  cameras do seem to produce same strange colours, compared to CCD.

It has to be down to the camera processing, as my Canon EOS 5d MKlV is perfect, as were all my DSLRs, and all CMOS of course.

Yes, I did copy and paste the post.

Edited January 14, 2022 by centroid
Typo

[vlaiv]

1 minute ago, centroid said:

CMOS  cameras do seem to produce same strange colours, compared to CCD.

It has to be down to the camera processing, as my Canon EOS 5d MKlV perfect, as were all my DSLRs, and all CMOS of coure.

That is down to color calibration.

CCDs don't produce accurate colors either - but are closer to real colors when color calibration is not performed. DSLRs perform color calibration internally because people expect accurate colors from their daytime images.

In order to produce accurate colors - one needs to create color transform matrix and apply that to raw linear data to produce linear RGB data. Then in order to maintain that color information - images need to processed in certain way. Almost no one is doing that in astrophotography.

When we stretch data we usually desaturate it a bit with that stretch - and that is why people end up boosting their saturation.

For example - here is DXOMark Canon 750d specs:

https://www.dxomark.com/Cameras/Canon/EOS-750D---Measurements

That is color matrix that produces accurate colors from RAW values for this camera.

When using dedicated astro camera - you get RAW values that need to be converted to sRGB tripled values for accurate color rendition.

[centroid]

Thanks Vlaiv, very informative as always 👍

That explains why we see many odd  looking colour images posted from CMOS colour astro cameras.

CCD astro cameras may not produce colour equivalent to that of our DSLRs, but in my experience were quite acceptable. Certainly not the case with  my re-badged Touptek 294c, but the colour I can adjust to suit.

Edited January 14, 2022 by centroid

[Adam J]

47 minutes ago, centroid said:

Ah, that is probably the answer Vlaiv. The images were non debayered. I have another 'play' this afternoon.

CMOS  cameras do seem to produce same strange colours, compared to CCD.

It has to be down to the camera processing, as my Canon EOS 5d MKlV is perfect, as were all my DSLRs, and all CMOS of course.

Yes, I did copy and paste the post.

Despite Canon calling it RAW it is not, they apply a different multiplier to each channel prior to readout. Hence boosting to the R and B pixels. Not well documented but if like me you have removed the Bayer matrix from one of their sensors you would know that a checker board pattern in the RGGB form is still present in flats despite all the pixels now being physically identical with no Bayer filter. Some experimentation leads to the conclusion that they do not process the pixels identically.  

Adam 

[vlaiv]

5 minutes ago, Adam J said:

Despite Canon calling it RAW it is not, they apply a different multiplier to each channel prior to readout. Hence boosting to the R and B pixels. Not well documented but if like me you have removed the Bayer matrix from one of their sensors you would know that a checker board pattern in the RGGB form is still present in flats despite all the pixels now being physically identical with no Bayer filter. Some experimentation leads to the conclusion that they do not process the pixels identically.  

Adam 

I think you can actually extract raw (or close to) raw with DCRaw and choice of color mode.

Have you tried that?

[Adam J]

5 minutes ago, vlaiv said:

I think you can actually extract raw (or close to) raw with DCRaw and choice of color mode.

Have you tried that?

This was about 5 years ago now, my memory is not the best but I looked into it quite extensively at the time. Also it was a 1000D so not exactly a modern processor. 

Edited January 14, 2022 by Adam J