PDA

View Full Version here: : Cant get the colour right????


E_ri_k
15-05-2013, 10:07 PM
Hey guys. I'm stumped as to what is going on here:screwy:
I posted an image of M83 a few weeks ago. I doubled the data on it the other night, but I can't process it......Or what I'm thinking is I have missed a step, and a can't work out what??? I load it into PI, stretch it, DBE, then curves and levels, remove the green. But I can't get any colour out of it. Boosting saturation makes it look crappy. I did my original image and it was fine, but I must be missing something?

I attached some poor quality JPEG's of what I'm working with. They both look the same. So if I managed to get a good result, I don't understand what I am missing. The first image is my original image, second is a stretched version before any processing. The last one is my new image with more data. It sounds stupid, but I don't know what I did in the first place???

The more I try to get colour out of it, the yellower it gets.
Hope this makes sense.

Erik

RobF
15-05-2013, 10:41 PM
My suggestion would be the DBE Erik. I sometimes find ABE can similarly strip colour when you use it early in the processing chain to control vignetting or colour cal.

Might be worth going back to the original masters and tweaking the histogram for colour balance, then stretching, etc. There are colour cal functions of course, but another option is:

1. Run Statistics module
2. Find the Median values for R, G and B
3. Call up Histogram Transform
Put the corresponding Median value in the Midtone for each colour
On the RGB/K tab, pull the midtone slider out the right a little
Run the HT
You should have better colour control now - effectively you're aligning the peaks of the R,G and B.

Be interested to hear how you go or what other people suggest.

E_ri_k
15-05-2013, 11:07 PM
Thanks Rob. I'll give it a go. I did find that this time DBE messed with the colour.

I'll let you know how I go!
Erik

Octane
15-05-2013, 11:18 PM
I would suggest cropping out the non-overlapping (redundant) portions of the image using DynamicCrop before doing any processing on the image, too.

H

E_ri_k
16-05-2013, 11:49 AM
Thanks H, yeah I always crop out any overlapping stuff before DBE. I think I'm getting somewhere :) Having a play with the colour levels. Just need to get rid of the pink tone.

Erik

RickS
16-05-2013, 12:03 PM
Did you try ColorCalibration, Erik? That's worth a go. Another option that I often use is eXcalibrator but that's a bit fiddly (but the new version is much easier to use).

Cheers,
Rick.

E_ri_k
16-05-2013, 10:16 PM
Hey Rick, yeah I did use the Colour Calibration tool, after some basic stretching. Maybe I did things in the wrong order this time? I'm making progress though. It just seems unusually fiddly and hard compared to last time?

Erik

RickS
16-05-2013, 10:36 PM
Erik,

ColorCalibration is normally applied to linear data. Perhaps that's the problem?

Cheers,
Rick.

rmuhlack
16-05-2013, 10:41 PM
Erik, I had a go in PI using the third image in your original post.

Processing steps:

1. Crop image
2. DBE
3. Background neutralization
4. Color Calibration
5. Local Histogram Equalization (Contrast limit=2, amount=0.2)
6. Extract luminance, stretch and black clip, and then apply this as a mask, to only reveal the galaxy and stars.
7. Curves to increase saturation of galaxy and stars (using the mask keeps the background from being affected). subtle boost to luminance as well (also with mask applied)
8. Color saturation to boost red (Ha regions) (also with mask applied)
9. remove green noise with SCNR (normally would do this before curve adjustment, but I forgot to do it earlier

E_ri_k
17-05-2013, 10:19 AM
Looks good Richard! Thanks for having a go. It goes to show I'm doing something in the wrong order with my processing.



It may be, thank you for pointing that out Rick :) I didn't know that.
Erik

E_ri_k
17-05-2013, 10:30 AM
Thanks Rick and Richard. It's coming up really good now. The problem was totally me using the ColourCalibration later on in the process...not at the start.

Erik