What should a good flat look like?

[DanLXIX]

I've just bought one of these light boxes to have a go at creating flats. Haven't used flats yet in the couple of attempts I've had and I know I really need them.

I've just quickly done a test. I used the wrapping to diffuse the light and shot it in aperture priority at ISO800 (which is what I've been shooting at). Is this what a flat should look like or have I made a rookie error somewhere?

 

[alacant]

38 minutes ago, DanLXIX said:

Is this what a flat should look like

Hi. The flat will depend on the telescope, camera, filters (you don't tell us!)...anything in the imaging setup where light passes or is reflected so unless someone has exactly the same setup as you when you took the flat then it's going to be difficult to say. It's also hard to know what the original is like but I had a go with your .jpg file. It looks OK -something isn't quite centred  but hey, it doesn't matter so much- and should work fine with the light frames you took unless you changed something in the imaging setup between taking the light and flat frames.

Stack anyway and see what happens;) HTH.

[DanLXIX]

1 hour ago, alacant said:

Hi. The flat will depend on the telescope, camera, filters (you don't tell us!)...anything in the imaging setup where light passes or is reflected so unless someone has exactly the same setup as you when you took the flat then it's going to be difficult to say. It's also hard to know what the original is like but I had a go with your .jpg file. It looks OK -something isn't quite centred  but hey, it doesn't matter so much- and should work fine with the light frames you took unless you changed something in the imaging setup between taking the light and flat frames.

Stack anyway and see what happens;) HTH.

 

I'm using the stuff in my signature with no filters. Looking at the histogram on the RAW file, the colour temperature is well off and I have to drag it right over to bring the histrogram back into line, so I don't know if that will cause any problems?

[alacant]

2 minutes ago, DanLXIX said:

I'm using the stuff in my signature with no filters

Guess: The Nikkor 18-55?

[DanLXIX]

15 minutes ago, alacant said:

Guess: The Nikkor 18-55?

No, the camera is on the telescope. Is that not correct?

[petevasey]

Looks promising albeit rather on the blue side, no doubt due to the LED's used in the panel.  Maybe once you have stacked your master flat, it would be worth adjusting the colour to give a more neutral tint.  If you have Photoshop the 'WhiteCal' function of the free plugin Deep Sky Colors could help.  (The other plugins there are useful, particularly Hasta la Vista Green!)

I made my own light boards some years ago using electroluminescent panel material.  They do a good job, but your solution is much cheaper and probably good for telescopes up to 8" diameter.  Hugely cheaper than commercial telescope light panels for example these where an 8" panel costs €155.  Also very handy that it runs off USB Power from your computer - no separate external power supply needed.

Cheers,

Peter

PS While I was writing this, you added the histogram information.  So if you can adjust the master using that, then that will be ok.

[DanLXIX]

12 minutes ago, petevasey said:

Looks promising albeit rather on the blue side, no doubt due to the LED's used in the panel.  Maybe once you have stacked your master flat, it would be worth adjusting the colour to give a more neutral tint.  If you have Photoshop the 'WhiteCal' function of the free plugin Deep Sky Colors could help.  (The other plugins there are useful, particularly Hasta la Vista Green!)

I made my own light boards some years ago using electroluminescent panel material.  They do a good job, but your solution is much cheaper and probably good for telescopes up to 8" diameter.  Hugely cheaper than commercial telescope light panels for example these where an 8" panel costs €155.  Also very handy that it runs off USB Power from your computer - no separate external power supply needed.

Cheers,

Peter

PS While I was writing this, you added the histogram information.  So if you can adjust the master using that, then that will be ok.

WhiteCal sounds like the WB selector tool in Lightroom, which gives me the same results, so it looks like if I adjust the colour temperature in camera before shooting the flats I shouldn't have to do anything after.

[alacant]

39 minutes ago, DanLXIX said:

camera is on the telescope

Ah, OK. Just take several nefs with the camera on A and stack them. Aim for an exposure of say 2 seconds. AFAIK, the colour is irrelevant. HTH. 

[david_taurus83]

As above, the colour is irrelevant. The master flat is only used to remove optical imperfections, mainly vignetting and dust/dirt artefacts in your light frames.

[Ken82]

1 hour ago, david_taurus83 said:

As above, the colour is irrelevant. The master flat is only used to remove optical imperfections, mainly vignetting and dust/dirt artefacts in your light frames.

Very true the flats are only ever used in grayscale so colour data is not important. 

[ollypenrice]

If a flat doesn't look like a mess you don't need it. But it will and you do. A credible flat, when stretched, will have darker corners and a brighter middle and there will almost certainly be dust bunnies scattered about. I never believe any flat which doesn't look like this and do them again. It is important to understand that, unless your rig has a very serious problem, an unstretched flat will look pretty flat. The drop off in illumination between corners and middle might only be 5 or 10% when you measure the pixel values using a suitable programme to 'mouse over' the image. In our most unevenly lit rig it is an enormous 23% yet flats correct it.

This gives me an opportunity to thank Peter Vesey, who posted above, for explaining to me as a beginner that I could use the black and white point values in Atik's Artemis Capture to see at a glance the range of brightnesses in my flats. Ten years on, that's what I do. Thanks, Peter.

Olly

[petevasey]

29 minutes ago, Ken82 said:

Very true the flats are only ever used in grayscale so colour data is not important. 

Not in this case tho' because a one shot colour DSLR is being used.  Therefore a blue flat will weaken the blue in the image.  That can of course be corrected in further processing, but surely better to have a neutral flat to begin with?

Glad to have been of assistance, Olly ?

Cheers,

Peter

[alacant]

1 hour ago, petevasey said:

blue flat will weaken the blue

Hi

Why would you debayer raw flat frames?

TIA:)

[Ken82]

2 hours ago, petevasey said:

Not in this case tho' because a one shot colour DSLR is being used.  Therefore a blue flat will weaken the blue in the image.  That can of course be corrected in further processing, but surely better to have a neutral flat to begin with?

Glad to have been of assistance, Olly ?

Cheers,

Peter

Flats are stacked together in grayscale regardless of OSC DSLR etc etc. You should not ever have to  debayer flats ! 

[petevasey]

5 minutes ago, Ken82 said:

Flats are stacked together in grayscale regardless of OSC DSLR etc etc. You should not ever have to  debayer flats ! 

I never said anything about debayering.  I've never had to manually debayer RAW (CR2) files from my Canon - DSS does this within the program, presumably bias, dark and flat calibrating the target images first then debayering them before stacking.  When viewed within Canon's Digital Photo Professional program the images are already debayered on screen.  If I'm not using DSS, I convert all the RAW files to TIFFs in DPP before further processing.  But if there is excessive blue within the RAW flat, that will also weaken the blue within the image.  Maybe I didn't make myself clear earlier - when I said 'blue flat'  I was referring to the original coloured flat shown at the beginning of this thread which had a strong blue bias - look at the histogram.   To me the later balanced version is much more suitable.

Peter

[DanLXIX]

11 hours ago, petevasey said:

 But if there is excessive blue within the RAW flat, that will also weaken the blue within the image.  Maybe I didn't make myself clear earlier - when I said 'blue flat'  I was referring to the original coloured flat shown at the beginning of this thread which had a strong blue bias - look at the histogram.   To me the later balanced version is much more suitable.

Peter

I can easily correct this in camera - I just need to adjust the white balance so no post processing will be needed.

[Thalestris24]

I think when viewing the flats onscreen (like the blue one above) you are viewing the jpeg part of the raw file. When stacking flats in dss it uses the raw data and ignores the colour information. It only cares how bright or dark pixel values are.

Louise

[petevasey]

26 minutes ago, Thalestris24 said:

I think when viewing the flats onscreen (like the blue one above) you are viewing the jpeg part of the raw file. When stacking flats in dss it uses the raw data and ignores the colour information. It only cares how bright or dark pixel values are.

Louise

True.  But the pixel values in the RAW frame will depend on the levels of Red, Green and Blue light passing through the Bayer matrix to the pixels behind the filters.  The colour information lies within those pixels and the debayered colour will depend upon the pixel values of each group of four.  Therefore when they are subtracted from the image files, the final blue balance will be incorrect unless the flat is neutral.  When I take flats for my monochrome CCD which of course uses external colour filters so I have to take a flat for each filter, I endeavour to have similar average values for each flat to counteract any colour bias from my light source.  It would appear that the light box being used in this thread has a blue bias, which will then affect the final calibrated image.

Peter 

[Thalestris24]

41 minutes ago, petevasey said:

True.  But the pixel values in the RAW frame will depend on the levels of Red, Green and Blue light passing through the Bayer matrix to the pixels behind the filters.  The colour information lies within those pixels and the debayered colour will depend upon the pixel values of each group of four.  Therefore when they are subtracted from the image files, the final blue balance will be incorrect unless the flat is neutral.  When I take flats for my monochrome CCD which of course uses external colour filters so I have to take a flat for each filter, I endeavour to have similar average values for each flat to counteract any colour bias from my light source.  It would appear that the light box being used in this thread has a blue bias, which will then affect the final calibrated image.

Peter 

DSS says:
"Is it possible to use colored flat frames?
The short answer is yes.
The overall tint of your flat frames is not really important because DeepSkyStacker is processing each channel separately and is applying the flat frames to each channel accordingly.
Of course if your flat frames are fully red only the red channel of your light frames will be properly calibrated but otherwise as long as the peak in each channel is between 1/3 and 2/3 of the maximum you should have no problem using slightly colored flat frames."

So, although it's probably better to have a good dslr/osc flats' colour balance, so long as the peaks are within the prescribed limits, it shouldn't matter much. It may be different if doing r,g,b with separate flats on a mono camera. FWIW I bought a led panel a few years ago but, in the end, I've found taking flats against an evenly lit wall to be better. However, I can't recall having any problems because the panel was strong in the blue (which I'm sure it was!). Of course, there may be variations in the results you get according to different software being used for the stacking. Also, I've used APT for image acquisition and creating flats from a dslr or osc. In that case I just make sure the luminance peak is at about 40% of the histogram max width and ignored the colour peaks. I don't think I ever had a problem with flats affecting colour.

Louise

[DanLXIX]

So is the best way to deal with this is not shoot the flats with a custom white balance, but shoot them in monochrome?

[alacant]

14 hours ago, DanLXIX said:

custom white balance

Hi. The white balance setting has no effect on the raw. Raw is raw and the image is monochrome. The colour balance information is however recorded along with the photograph in case you want to debayer and view it in colour on e.g. your camera preview screen. Just shoot raw flat frames to calibrate your light frames; what you already have will very likely work fime. HTH.

 

[ollypenrice]

OSC is easier, right? ?

lly

[Rob63]

1 hour ago, ollypenrice said:

OSC is easier, right? ?

lly

LOL, you just couldn't resist it Olly, could you.

[Thalestris24]

3 hours ago, ollypenrice said:

OSC is easier, right? ?

lly

It is in some respects - but not in others! Mono luminance or Mono Ha is easiest (for me) really

Louise