Daytime G2V colour calibration

[RikM]

I can't find the original thread where this was suggested, sorry.

To work out the relative weightings, I pointed the scope at a sheet of white card in the garden, mid afternoon, and took short exposures through each filter. I just used the small aperture in the dustcap of my 150P and covered it with a bit of tin foil with a small hole in the middle. I ended up using 0.2 sec exposures. I took three subs through each filter and compared the average pixel vales using the histogram in Maxim; I took the midpoint of the curve. Then took the average of the three subs as my value for that filter.

I got roughly 4,400 for red, 4,700 for green and 6,000 for blue. I wasn't expecting blue to be the highest but there you are; I do have a LPR filter fitted in front of the coma corrector. That gives me ratios of 1.4 : 1.3 : 1.0.

As a test image (and first go at an LRGB, third shot ever with a CCD!)

L: 14 x 300s

R: 3 x 300s (binned 2x2)

G: 3 x 300s (binned 2x2)

B: 3 x 300s (binned 2x2)

Colour combined in Maxim using the weightings above.

It's very noisy and the stacking / alignment isn't quite right. The filterwheel rotated slightly when I changed filters because it wasn't done up tightly enough but that aside, how does the colour balance look?

Did I botch anything with this method?

[rfdesigner]

well it looks pretty nicely balanced.

However if you want to be really really picky, then you need to take account of atmospheric extinction.. that is how much more the blue light get attenuated at the altitude you're imaging compared to the red. This is where imaging a G2V star local to the target is optimal.

Also remember that your calibration frame may have been out.

The 'white' paper will not be pure white.. it may have a slight hint of blue or yellow. Also the sunlight may in reality be slightly bluer but with it not being exactly at the zenith when you did your calibration again it will be fractionally out.

Please bear in mind.. I'm splitting hairs to some extent, what you've produced strikes as a very natural colour balance, I've done similar things in the past with OSC with fairly satisfying results.. certainly much better than just guessing or playing with the 'look' of an image.

Derek

[steppenwolf]

I like your image although from personal taste (and this is just personal taste and nothing more!), I think it needs more blue to get the colour right.

Paper is often manufactured with a blue cast to it as the eye is fooled into thinking that the paper is 'ultra white' as oppose to 'warm' which indicated red/yellow.

My preferred daytime test is not to use paper at all but rather to use any colourful daytime scene and shoot it with an IR cut filter in place at 100% for each colour and then adjust the weightings during processing to get the most realistic rendition on the monitor that you intend to use for future processing. Once you have the correct colours on screen, you can calculate the required weightings for future captures.

[Shibby]

I use a grey card for daytime photography and it works very well:

http://www.amazon.co...45020374&sr=8-3

Of course, as has been said, the colour of your images will be affected by the atmosphere & light pollution. The balance in that image does look good, though!

[RikM]

I've got a 12" pop-up light tent that I use at work for product photography. Could I use the backdrop out of that as a white card?

[RikM]

I just did this properly with a G2V star (16 Cyg a) and got RGB values of 1.32:1.30:1.0, so not a million miles away from what I had before.

To balance things up, is it possible to shoot say 5x5min in red and green and 4x5min in blue and then combine at 1:1:1?

[Shibby]

I may be wrong, but I think that would just give you less noise, rather than extra weight, to the red channel.

[RikM]

But if the red and green data is getting stretched—relative to blue—when I combine at the correct weightings, would the extra subs in those channels be a help or would it make no difference? (I am talking about the principle rather than this specific number of subs).

[dmahon]

Would you be better shooting 5x5mins in R&G and 5x4mins in B (as opposed to 4x5mins).

[RikM]

Would you be better shooting 5x5mins in R&G and 5x4mins in B (as opposed to 4x5mins).

Probably.

I just wondered if it worked the other way. I read a presentation that said it was SNR that determined true colour balance and so it is possible to do it either way. Keeping all the subs the same length means only one set of darks.

[Shibby]

But if the red and green data is getting stretched—relative to blue—when I combine at the correct weightings, would the extra subs in those channels be a help or would it make no difference? (I am talking about the principle rather than this specific number of subs).

Yeah, I suppose if you were combining with your calculated weightings then having extra data for the more stretched channels would be a good idea. Seems to make logical sense, anyway...