AstroArt Colour Synthesis Help

[samtheeagle]

I'm trying out the AstroArt 5 demo, with a view to purchasing the full version for my rapidly approaching birthday I've seen many very positive comments about the software on the forums here, and my first impressions are that it's very good. However... I'm really having trouble getting getting anything like acceptable images with it I know that the data is good enough (for my standards) as DSS stack and PS processing is giving me results I'm happy with, but I cannot get anything close to a good image with AA5.

I'm getting varying issues, but most seem to stem from the CCD Colour Synthesis stage. A stack of RAW files is giving me the grainy grey bayer image, which is showing me a really nice amount of detail. But as soon as I convert to colour all the detail goes, and whatever I try I can't tease it back out again. I'm also finding the colour to be wrong. For example, my Rosette image is coming out blue I've had other times when the colour image has come out with all kinds of strange mottled rgb patches across the image.

I've been reading the help file, and it's pretty good at stating the obvious, but I'm not finding it all that helpful as to what order to do things, or WHY I should do X or Y. I'm sure that the data and the software should be able to give me a good image, but I can't figure out how to do it

As I said, I really like a lot of things about this software already, but if I can't get a good image out of it, well, it's not going to be a very good birthday prezzy now is it?

Hope some of you existing users can offer me some pearls of wisdom...

[ianaiken]

Have you experimented with the CCD Colour Synthesis options and changed the values for XY Offset? These determine the colour matrix used in your raw files, and getting this right should give you a correct coloured image.

[samtheeagle]

Yes, I've played with the XY offsets, trying all the combinations. I have only left it on RGB thus far, as that's what the RAW contains, I think. I'll randomly try other things now The odd thing is no matter what I set the offset values to the colour output always seems to come out with blue hue.

[samtheeagle]

Can anybody help me out with some AA5 advice???

Here are some example images snipped from the demo version to illustrate what I'm getting. This first image is the raw stack of light frames only, no calibration frames applied at all. I'll get on to that :

This is what I'm getting after colour synthesis. Blue. When the Rosette is very definitely Red. And it comes out Red with this data using DSS.

Then for the real icing on the cake, this is the sort of insanity I'm getting when I attempt to use the dark and flat frames with this data. I've tried everything I can think of, twiddled every option I can find, but it's always the same. Washed out and blue.

I really thought that investing in some well regarded astro processing software would be a good idea, but right now I'm thinking why on earth would I spend money on this? I can't get a good result out of it to save my life. At least the freeware apps give me images like I'd expect.

I'm sure it's a failing on my part, but can anybody englighten me as to what it might be?

[nytecam]

There's severe [and large] Moire patterning in that final image - are the flat/darks processed ok? But it doesn't solve the blue cast. Any Youtube AA demos to check against?

[ianaiken]

Is that pattern due to your darks or flats or only appears when you use both?

Can you post your master dark and flat files in original format?

[ollypenrice]

First make a master flat. Stack your flats in the Images box and use a master bias as the dark frame for those flats.

When this master flat pops open, debayer it in Image - Shift and change both axes to 0.5. This will give you a clean flat.

Now do the rest in 2 stages. First, in Preprocessing, load your lights, the right dark for them and the master flat. You can add the hot pixel filter at this stage if you like. Do NOT combine the images and do NOT apply CCD colour synthesis, just run the programme with images kept and saved. They will be saved with the P_Fit suffix.

Next reset the preprocessing window and load up only the P_Fit suffixed images which are now calibrated. Go for sigma combine, align by translation and rotation, check the CCD colour synthesis option and hot pixel if you didn't already do that.

I check Normalize Background and Confirm Each Image if I haven't seen all the subs.

I think the 2 stage process is not really necessary in AA5 but try it anyway. Debayering before stacking damages colour information.

Olly

[ianaiken]

Olly,

The process you describe here sounds as though you can calibrate images prior to combining them, which would mean you can use a different set of calibration frames for the lights and then combine all the P_Fit files once all lights are calibrated.

Interesting point on debayer before stacking. In PI my work-flow is to debayer once calibrated prior to integrating.

[ollypenrice]

Olly,

The process you describe here sounds as though you can calibrate images prior to combining them, which would mean you can use a different set of calibration frames for the lights and then combine all the P_Fit files once all lights are calibrated.

Interesting point on debayer before stacking. In PI my work-flow is to debayer once calibrated prior to integrating.

Good point there, Ian. (Sorry, OP, this refers to an earlier discussion with Ian.) Yes, that would allow you to combine different sub lengths in a single integration. Obvious when you think about it! Rather dim of me not have thought of that.

Olly

[ianaiken]

And on a tear down setup different flats also rather than different sub lengths. You need to think about it 30 times and stack the thoughts for it to make sense (Edit: I've just seen the None option - brilliant!)

Back to OP - with AA5 I would go back and process like Olly suggested. When you come to the CCD Synthesis it's very important to get it right.

You should see a matrix pattern on your image when zoomed in if you have it wrong plus the colours may also look wrong. I found out that it's not good enough to just look at the colours, I thought I had the right colours on mine until I zoomed in and had the matrix pattern. With a bit trial and error I was able to get the right setting.

I think the pattern is RGGB for Canon. So leave it set as RGB and then leave both offsets at 0.

[samtheeagle]

Yay, some more responses! I was starting to think I was on my own for a while there I found an excellent AA4 Flats tutorial video by Harry Page over lunch, which has cleared up some of the issues. I wasn't removing the colour information from the flats files. I'm too used to DSS just taking all the files I throw at it and sorting them out for me I knew it was going to be down to my process, I was just struggling to find out what I should be doing.

I'll have a go with the process as defined by Olly tonight I got considerably less moire patterning when I used a better flat, though I didn't have all the files to hand to perfect it. The main issue I had then was that the colour was still blue. If I split the colour synthesised image into separate RGB channels the blue one was by far the strongest, which makes me think something is a bit confuzzled.

[ianaiken]

What light source did you use for your flats? A flat ultimately will be greyscale , colour information is not important its the intensity of light that is afaik.

[samtheeagle]

I've got an EL panel. When I view a single flat it's kind of sky blue in colour. But I've think I've got flats sorted out now, with the shift 0.5 magic I just need to sort out my bias frames and darks, and hopefully I'll be laughing.

Thanks for the advice, all. I'll report back on how things go.

[samtheeagle]

Ok, so I'm being hamstrung by the demo version not being able to save... So I'm not going to be able to properly replicate Olly's process. But I'm getting the idea now I think. I was just writing a long bit about how I was still getting a blue hue and other weirdness, but I've just re-read Olly's process and just picked up on the "None" option!

So, I'm prolly stuck for now without the ability to save, but I think I'm confident enough now to buy the full version, at which point I'll be primed and ready to go

[harry page]

Hi

I use AA5 a lot , well worth the money I believe FLO stock it

Harry

[samtheeagle]

Indeed they do, but when I asked recently they were out of stock SCS Astro also seem to be resellers, and they had stock when I asked a couple of days back. Fingers crossed. I'd really like the boxed version rather than download, just hope I can get it before Friday now

[harry page]

Hi

Fabio just sent me mine , not giving it away though

Harry

[samtheeagle]

Awwww, come on now Harry, you wouldn't want to deprive a man of his birthday prezzy would you?

[astrodude]

Can anyone one tell me a good starting point for unmoved dslr CCD colour synthesis setting in AstroArt. I am trying to process the Orion nebula, but it comes out a very bright red. What would be a good ratio for rub to try?