One shot colour camera check

[Johnboy]

Hi all,

I bought a Brightstar Mammut one shot colour camera a few weeks ago but I'm not sure that is is actually colour. I know it sounds daft but the mono and colour cameras are basically the same but the colour has a CMY Bayer matrix across the sensor. The thing is I have not yet had any colour images from the camera. I tried a test that I had seen on the web whereby I held the camera about an inch from a totally green background and took several images of varying lengths, but after debayering all I got was a fuzzy pixelated grey picture. I also tried cyan and yellow screens with the same results. Should I have had a colour picture doing this? Am I doing something wrong here or is the camera actually a mono?

Any help as always is greatly appreciated.

Thanks,

John

[martin_h]

What programme are you using to capture the images and what are you using to debayer them?

[Johnboy]

Hi

To capture and debayer the images I used the program that came with the camera. (NightWatch Studio, capturing in RAW and CMY mode) I also tried debayering in DSS. I am totally new to imaging so the problem could well be (and probably is) my inexperience. Could you please tell me what method you would use if you tried this test so I could follow it.

Thanks you for your help.

Best regards

John

[steppenwolf]

Not familiar with the camera in question but this could be a matter of the 'offset' not being correct when you de-Bayer the image. The de-Bayering algorithm needs to know what colour the first pixel is, say, at the top left hand corner of the sensor and de-Bayering software normally has an X/Y offset tick box or set of permutations to ensure the colours produced are correct. If you look closely at the image, can you still see a crosshatch of lines in a matrix? If you can, this is one of the signs of the offset being incorrectly set in the de-Bayering software.

Can you send me a FITS file of an original frame (not de-Bayered) to have a look at?

steve@nightskyimages.co.uk

[Johnboy]

Thanks Steve,

I will get an image over to you. I am really enjoying your book by the way and finding it very useful. It's definitely written with someone like me in mind!

Best regards

John

[DoctorD]

Hi

You could try the camera in daytime - you'd soon see colour if it was there.

HTH

Paul

[rfdesigner]

When you look at a pure single colour the raw (B&W) image should show a checker board type pattern. If it's even then it's mono.

Derek

[Johnboy]

Thanks for all your suggestions folks.

In between the rain I managed to take a daytime image and sent it over to Steve who very kindly worked on it for me. The camera produces CMYG colours rather than RGB which is the reason the de-mosaic program I was trying failed to produce the correct colours. All's fine now (but no doubt I'll be back with more questions! ).

Thanks again,

John

[AstroAndy]

Hello folks

I seem to have the same problem with the same camera. After eading all this, I'm still not sure of how to go about getting (correct) color images..maybe I'm already getting them, and just don't know it (as I have produced some color images, but it was always after messing with the color controls, and not "innate" color of the object after @ least 3-5 hrs of exposure). So, it would help to have a step by step sort of manual for dummies (e.g. open Night Watch S., load fit, click button x, then button y, set value w). I usually stack all my frames, then try to get the color image by using the CYMG button (I also tried the method where you duplicate a layer, use "Soft" and "Lum" to gain color, but no dice). Also, while I'm on the subject, how to use the deinterlace function correctly (before, after calibration and stacking, pic by pic)? When I use DSS, the little color icon appears beside my images, so that's ok, but how do I get an objects color w/out "coloring" the whole pic? My next plan is to shoot in RGB mode to see if there is, and if so, what the difference is. So far, I have gone thru all the color buttons, and ended up w/ some weird effects (colorwise).

[Johnboy]

Hi AstroAndy,

I'm glad I'm not alone! Whatever I do while debayering I always end up with the whole picture coloured. I would have thought the 'black' of space would stay black and the colour in the object would show it's natural colours. The Brightstar Mammut has a Bayer pattern which is CMYG rather than RGB. I took a daytime shot of some leaves and got a nice green colour, but as always the whole picture was in green! Is it me? Probably, but why should a one shot colour camera be so hard to use?

Also I have found that I have to change the matrix 'offset' values to different settings with each picture I try and colourize. For example the leaves mentioned above were set to x=1 and y=2 but with other pictures this setting gives some really strange results. I thought once that was set that would be the same for every picture.

Please let me know if you have any success and I'll do the same for you. Everyone has been really helpful which I very much appreciated but at the moment I'm feeling very deflated.

Best regards,

John

[steppenwolf]

I thought once that was set that would be the same for every picture.

That is indeed the case, offset is offset! If you are having to adjust this from image to image depending on content then you are really just chasing a false colour so I suspect that we haven't got there yet with the de-mosaic yet.

Can you send me a raw image of your 'leaves' please? The whole image being the colour you are seeking is pretty much confirmation that the de-mosaic has not worked properly and you are just 'choosing' a colour result that generally meets your expectations hence green leaves. It *could* just be a colour balance issue I guess but let's have a look at a RAW first.

[AstroAndy]

Syeve has a point there..if I work on single exposures, I get some nice color, if I don't, I'll "tweak it into what I expect..From what I understand, one first has to stack and calibrate the RAW data (fits), and then debayer them..so far, I do that with Bight Watch Studio (NWS). I haven't found the deinterlace function in DSS yet, although DSS stacks better than NWS). An aditional problem is, the interlace, or bayer grid) gets brighter the longer the exposure, so I HAVE to deinterlace, otherwise I can't do the alignment in the calibration, since I can't choose alignment stars thru the bright grid. Using DSS, I get lines thru my pix. My next plan is to shoot in RGB, and see.

[AstroAndy]

Hi John,

what I found was if you CYMG RAW convert one unprocessed exposure, you get "true color", and you can take that as a guide when you stack and colorize all images. Unfortunately, I don't know whether this is true for all images of all exposure lengths (but I would think so). Below is an image of M57 I shot yesterday, A single image of approx. 2 minutes, the original .fit file was not processed, the only thing done to it was a CMYG RAW conversion.

Andy

[Johnboy]

Hi Andy,

I wish I could get something that is as good as your image. I have attached a 20 second exposure of M82 from a couple of nights ago and no matter what I try I can't get any colour at all. I have even stacked about 100 images and tried to debayer them using different offsets but get nothing. Would a single 20 sec exposure show any hint of colour? Can you do anything with my image and if so please tell me what you did!!

Thanks for taking the time to help.

Best wishes,

John

m82 20secs1L20.fit

[AstroAndy]

Hi John,

Colors are inherently faint in 1) short exposures, 2) some objects..M82 @ that time would be hard (I did M31 on a Pentax KX, 63x 30s), then you can pull out some color, it also depends on your equipment..I'll give your image a go later on today, and see what I can come up with. Even if you stack 100 images at say 20s, all you do is improve SNR (signal to noise radio), the trick is catching photons, and that means, in my humble opinion abt1min - I've had trouble w/ one of my fave objects, M51, but now got a good "live" looking image". What I've found works best for my conditions (again w/ M51) is 3m, as many frames as I can, w/ all the calibration data (I\ve done from 1-10m, but @ 10, I get too much gradient, 5 is good, 3 is the best compromise).

Andy

[AstroAndy]

Hi John

Here are my quick and dirty tries, I'll put them up as pscreenshots so you can see the values, one is purposely overdone..amazing what color you can get out of it. All of this was done w/ Night Watch Studio. If you stretch them a little bit, you won't get a difference between YbCbr and RGB output (don't ask me what happened to the stars, though, lol).

Andy

[steppenwolf]

That's interesting as these are very similar to the results that I got with John's data although my version has horizontal matrix lines. There is undoubtedly some colour in there but John's camera is not producing consistent output which is odd:-

[Johnboy]

Hi Andy,

Thanks for looking at my image.

Like Steve says my camera appears to be producing inconsistent data. I have copied your values from the screenshots and have a very different image. (Screenshot attached). No matter what I do it seems to change with each capture somehow.

Thanks for all your help with this.

Regards,

John

[AstroAndy]

Np..lately I've been spending time w/ the RAW DPP settings in DSS, I have to go thru every setting and combination to see what works best to cut out NWS altogether (if I work with both, DSS & NWS, I have the additional step of converting an FTS file into a FITS file, although they should be the same, but putting an FTS file into NWS will give you unworkable effects). Another prob. is that interlaced CCD.

[Johnboy]

Hi Andy

Thanks for keeping me updated with your ideas and progress. I'm on hold at the moment because I think I have tried all ways of de-bayering my images (with various software packages) but the odd and inconsistent results have now got me baffled. Being new to this I have come to the end of my knowledge. I know once the offset values are correct they shouldn't be changed again, but one images' setting is not correct for another. I also did a daytime shot of some coloured containers that I set up in my garden, but can't get any colour whatsoever from it. (Yet the image of M82 above did show some colour).

John

[AstroAndy]

Hi John,

You and me both..I'm trying to develop a workflow on how to best process..I've gone through all the settings in DSS, different debayer methods, and still not come up with a conclusive workflow, however, slowly a handle is gotten on doing it w/ NWS. Basically I stack everything first, calibrating with abt 23 darks flats and bias, and aligning, then I colorize the stack. That way, I've gotten some decent M51s (more colorful than the one above). The thing is, the more subs you take, the more the interlacing will disappear, one can also deinterlace and then colorize (when there are only a few subs). I'm sure the answer is somewhere in DSS (I prefer to use it instead of NWS), so i'll keep looking. I cvan still do some decent preprocessing in DSS, but matrix lines remain. I myself am not that advanced, still feeling my way, next I'm gonna check the specs. on the mammut, maybe it'll give me any clues. I also think that once you find the right offset, it shouldn't change.

[AstroAndy]

Looks like my last comment was swallowed.

[AstroAndy]

So far I've found that if you color convert 1 image with the right values it gives you a base line of the actual color of the object, then stack, you can "customize" the stack so the same colors apply, only stronger; also, don't forget you have the color balance. Another way (as I haven't worked out DSS fully yet, is to stack in DSS, then save as fit (or reconvert if it is fts, w/ something like imageJ), then import the fits (not the fts) and color convert in NWS. Still not ideal, but workable.

[Johnboy]

Hi Andy,

I have had a *little* success. I loaded 5 x 20 second RAW images of M82 into NWS and managed to get a colour picture. (Albeit a rubbish one, really!) Working on one image at a time I firstly debayered, then I adjusted the colour to get something a little stronger. After that I did a stretch using the 'Histogram' option which actually made the image look dreadful but after adjusting the histogram sliders I got something quite acceptable. After making sure all images were stretched exactly the same I then stacked them. Not too bad but some of the stars are of varying colours, (Maybe darks, flats etc may help here but I'm nowhere near that far yet!) The problem I have now is when I load this stacked image into Photoshop all I get is a very dark image which no amount of stretching will solve. I know there is a learning curve but this seems crazy, but hey I'm nearer than I was last week. (At least I think so!) Considering I have only used 5 x 20 seconds do you think things will improve the more data is collected or should I be getting something better even at this early stage? I will upload the image I have so far later for all to laugh look at.

I have to say Bern at Modern Astronomy has been (and is continuing to be) exceptionally helpful throughout this.

Not sure if you know this Andy, but the 'Make Pixels Square' option is the last thing to do in the processing chain as doing this before converting to colour destroys the colour data.

HTH!

Regards,

John

[AstroAndy]

Hi John,

Didn't see these posts before, lol..anyway, yes, I know abt. the pixel squaring, however, when I color convert img by img., NWS won't stack w/ my darks, flats and bias (diff. channels after conversion..I'd be sitting there forever (unless I have few, say 20m exposures, then it's doable). Gotta get ready for tonight's observing, maybe will post the results and steps later (way later, lol). Darks, flats and bias r easy, for darks, just cover the scope (camera), take an exposure @ the same time and temperature )and optical train, pointed @ the object) and presto majesto, you have a dark..fr what I hear, abt 20 of them will be ok, bias, same, except set the cam (scope covered) on it's fastest speed )0.001s, and in sequence, rip off abt. 30, goes quick. Flats, in clear twilight, point the scope uncovered @ an even part of sky, and set it to 1/250 of a sec )0.004), a quick 20 abt., you should be on your way. They really help. The other night, I everything came together brill., incl. guiding, and I was able to get a few 10, 15, and even 20m exposures. Now all I gotta do is process them right, lol. Clear skies.

Andy