Colour balance and flats and RAW files

[SteveJP (Steve)]

Hi,

I'm imaging with an ASI071MC Pro OSC camera, generally using Sharpcap, although I'm also playing around with SGP as well. I'm calibrating, stacking and processing with Pixinsight.

There are a couple of things I don't understand, so perhaps someone can enlighten me.

Firstly, Sharpcap implements the Red and Blue colour balance controls provided in the ZWO API, so when I'm taking flats, especially sky ones, I can adjust for a nice neutral grey so that the stacked image is not affected by the colour of the flat. SGP on the other hand doesn't provide any colour balance controls (presumably since it doesn't debayer anyway), so you get I think the ZWO default colour balance - what ever that is. Is this not a problem with SGP? How do you get neutral flats?

I should add, that I have Sharpcap set to RAW16, but the colour balance still affects the colours in the FTS file (after debayer). Actually I don't understand why the Sharpcap colour balance settings affect the FITs RAW file. What is the point of a RAW file anyway, when it's the camera itself that is adjusting the colour balance?

Of course, none of this would matter if the colour of the flats made no difference to the stacked image. It certainly does with Pixinsight. Why should the colour of flats be taken into account anyway when the frames are being flattened.

Thanks
Steve

[multiweb (Marc)]

I'm not 100% sure but I'd say the color of your flat doesn't affect the color in your light frames. It's divided in. My flats depending on what EL sheet I use vary from pink to blue. I've never noticed any issue with color balance down the line with stacking and processing. Some others might chime in.

[glend (Glen)]

Not trying to be unkind, but the simple solution seems to be, don't use Sharpcap. If you have SGP I don't see why you would use Sharpcap. And frankly, if your just starting our, and using a OSC camera, i would forget about flats for now. Flats are not a requirement, but a refinement.

[SteveJP (Steve)]

Originally Posted by glend

The simple solution seems to be, don't use Sharpcap. If you have SGP I don't see why you would use Sharpcap.

Well, for one thing, with Sharpcap I can set the colour balance, with SGP you just get pot luck as far as I can see?

[SteveJP (Steve)]

Originally Posted by multiweb

I'm not 100% sure but I'd say the color of your flat doesn't affect the color in your light frames. It's divided in. My flats depending on what EL sheet I use vary from pink to blue. I've never noticed any issue with color balance down the line with stacking and processing. Some others might chime in.

Hi, I'll have to check if "divide" is being used. I certainly notice a difference however.

Also, I wonder if what you say is true only when doing NB with a Mono camera when you never debayer? Do you mean when taking OSC images and flats?

[SteveJP (Steve)]

Originally Posted by glend

Not trying to be unkind, but the simple solution seems to be, don't use Sharpcap. If you have SGP I don't see why you would use Sharpcap. And frankly, if your just starting our, and using a OSC camera, i would forget about flats for now. Flats are not a requirement, but a refinement.

Hi Glen,
The problem is not Sharpcap, which has some really nice features such as probably the best polar alignment method going, but trying to understand the affect of colour balance, particularly when using flats.
Steve

[RickS (Rick)]

Don't expect the calibrated data to look colour balanced. Once you do BackgroundNeutralization and ColorCalibration (or PhotometricColorCalibration) it will look fine.

A quick way to see what the data looks like prior to colour calibration is an unlinked screen stretch (unclick the icon that looks like a chain link on the ScreenTransferFunction process.)

Cheers,
Rick.

[SteveJP (Steve)]

Originally Posted by RickS

Don't expect the calibrated data to look colour balanced. Once you do BackgroundNeutralization and ColorCalibration (or PhotometricColorCalibration) it will look fine.

A quick way to see what the data looks like prior to colour calibration is an unlinked screen stretch (unclick the icon that looks like a chain link on the ScreenTransferFunction process.)

Cheers,
Rick.

Hi Rick, Thanks for the tip about the "unlinked" stretch. I hadn't used that before. I tried it on a recent M45 after DBE and didn't see much difference, but maybe the colour cal was ok anyway. I do generally try BackgroundNeutralization after DBE, but often the difference is almost imperceptable - I guess that's not a bad thing. I do use the PhotometricColorCalibration and it is really impressive - seems like cheating however :-)
Thanks
Steve

[RickS (Rick)]

Originally Posted by SteveJP

Hi Rick, Thanks for the tip about the "unlinked" stretch. I hadn't used that before. I tried it on a recent M45 after DBE and didn't see much difference, but maybe the colour cal was ok anyway. I do generally try BackgroundNeutralization after DBE, but often the difference is almost imperceptable - I guess that's not a bad thing. I do use the PhotometricColorCalibration and it is really impressive - seems like cheating however :-)

Hi Steve,

Unlinked STF stretch will do a decent job of matching channels even the the background is out of whack but once you've neutralized the background with either DBE or BGN it won't make as much of a difference (and BGN won't make much difference after DBE because removing the gradients does effectively neutralize the background as well.)

I don't consider PCC cheating The correct ratio between colours isn't fixed and depends on atmospheric conditions, light pollution, etc. so it's best to calculate separately for each data set.

Cheers,
Rick.

[sil (Steve)]

Steve, I agree with Rick. Don't worry too much about the difference or using PCC. PCC still uses approximate values to colour balance but still close at least. PCC is good to compare your colour calibration against to make sure you've done your data well and it gets you quicker to a pic to put on facebook. Using the "shortcuts" is valid to use depending on your needs. Especially when you've changed hardware or trying something new you want to see if your data will produce a good image and that its worth the time going through a long manual workflow. I use APP to process jpgs to an image before I chuck raws into Pixinsight to work on. Sometimes I use the APP output as my target for registration in PI. They are just tools use their strengths.

[Capella_Ben (Ben)]

I have a ASI071MC-Pro as well and was pretty confused by the colour balance at first.


What I ended up doing is to leave the colour balance at 50 in SharpCap for both channels and do all the clibration in PixInsight. This makes the preview in SharpCap look pretty horrible, but I generally just go by the histogram anyway. I also found out that turning off the debayer in SharpCap gave you a better idea of the signal quality.


Same for Flats; leave the balance at default 50. I use a tablet with a white screen to create them and they don't turn out white. I don't think it matters what the colour is. I don't debayer the flats in PI before calibrating (don't think you are supposed to).



I found that PCC is great for this camer as well.

[SteveJP (Steve)]

Originally Posted by RickS

Hi Steve,

Unlinked STF stretch will do a decent job of matching channels even the the background is out of whack but once you've neutralized the background with either DBE or BGN it won't make as much of a difference (and BGN won't make much difference after DBE because removing the gradients does effectively neutralize the background as well.)

I don't consider PCC cheating The correct ratio between colours isn't fixed and depends on atmospheric conditions, light pollution, etc. so it's best to calculate separately for each data set.

Cheers,
Rick.

Thanks Rick. I know it isn't really cheating, but it's just so cool!
Cheers

[SteveJP (Steve)]

Originally Posted by sil

Steve, I agree with Rick. Don't worry too much about the difference or using PCC. PCC still uses approximate values to colour balance but still close at least. PCC is good to compare your colour calibration against to make sure you've done your data well and it gets you quicker to a pic to put on facebook. Using the "shortcuts" is valid to use depending on your needs. Especially when you've changed hardware or trying something new you want to see if your data will produce a good image and that its worth the time going through a long manual workflow. I use APP to process jpgs to an image before I chuck raws into Pixinsight to work on. Sometimes I use the APP output as my target for registration in PI. They are just tools use their strengths.

Thanks sil.
I'm still learning the best methods.
Rgds
Steve

[SteveJP (Steve)]

Originally Posted by Capella_Ben

I have a ASI071MC-Pro as well and was pretty confused by the colour balance at first.


What I ended up doing is to leave the colour balance at 50 in SharpCap for both channels and do all the clibration in PixInsight. This makes the preview in SharpCap look pretty horrible, but I generally just go by the histogram anyway. I also found out that turning off the debayer in SharpCap gave you a better idea of the signal quality.


Same for Flats; leave the balance at default 50. I use a tablet with a white screen to create them and they don't turn out white. I don't think it matters what the colour is. I don't debayer the flats in PI before calibrating (don't think you are supposed to).



I found that PCC is great for this camer as well.

Hi
I've used 50/50 as well, but last time I imaged I used auto colour balance for Sky flats and they came out a nice grey. Your right, you don't debayer before calibration. The only reason I did that was to find our what colour the flat really was in an effort to determine why the colour seemed to affect the calibrated frames. Several people including yourself think the colour of the flat should not matter, so I'm going to have to go back and re-test this. If I find anything interesting, I'll re-post.
Cheers
Steve