Why do choice of colour space affect required gain/exposure

[mickelind]

Hi!

I have just started doing my first attempts at deep sky imaging using my old ASI1290MC camera. Initially I took some images of M77 using the RGB32 colour space and by looking at the display histogram stretch (and the final result) I found that using gain=300 and exposure 20s gave good results.

Now I want to try using the RAW color space (in order to try some other tools for post processing) but when testing I saw that the image became darker (also visible in the histogram) and that I need more gain/longer exposure.

Can someone explain why this is happening?

Mikael

[admin]

Hi Mikael,

it's not usual for the brightness to change in this way when you change from RGB to RAW. One thing that is possible though is that some ZWO cameras have a 'high speed mode' option that can brighten the image. If you switched to RAW16 then the high speed mode option would no longer take effect, leading to a darker image. RAW16 is the right choice for any deep sky imaging - using 8 bit modes (RAW8, RGB32) will reduce the image quality by throwing away useful data except at very high gains.

cheers,

Robin

[mickelind]

Hi!

Thanks for the feedback, I did not realise that this might be related to the type of camera I was using, I thought it was something that I just did not understand about capturing in RGB vs RAW in general.

Since I have been using 2x binning during capture I looked up how this is implemented in ZWO cameras and found this information: https://astronomy-imaging-camera.com/tu ... ntals.html from ZWO.

Apparently ZWO accumulates pixel values when using RAW8 (and I guess RGB) but averages values when using RAW16 so this explains why RAW8 images will be brighter compared to RAW16. I really do not understand why they use different approaches for different formats though but that is another question.

/Mikael

[admin]

Hi Mikael,

ah, yes, it sounds like the binning choices made by ZWO are the cause in this case.

Using the averaging binning in RAW16 makes sense, since most cameras are really only 12 bit, so can only produce 4096 output values. These are all scaled up to the 16 bit range by multiplying the raw values (0..4095) by 16 to get the range 0..65520. As you will see, only the values that are multiples of 16 get used in the output data in this case. By using averaging binning, you start using the values in between, which means you can see the benefit of binning with no loss of data.

For RAW8, there isn't a good option - if you average then you throw away data (every time you really want a 1/2 value, you have to round either up or down). If you add pixel values then you soon get saturated pixels and lose data that way.

cheers,

Robin

[mickelind]

Ah, I see, makes sense, thanks for a really good explanation!

/Mikael