This is my first attempt at LRGB imaging with my QHY9. My filters arrived safely and in great condition, thanks Martin Pugh.
The image is L=10 min RGB = 5 min each in 30sec subs.
The image appears to have some form of vertical banding which I am hoping is read noise as I took no calibration files for this image. (a little to eager to see a result)
The image was taken last night with quite a bright moon and the combining is quite new to me also. I am hoping for some tips from all you who know what is the best way and software to combine the images. I used Maxim 5 this time and Photoshop for final stretch but really didn't do much more.
Your assistance would be appreciated.
Certainly looks like natural colouring - not overally saturated, however from my monitor (not calibrated - work PC) the stars look a little over stretched.
Having said that it is till a nice image. What IR/UV filter was used?
Well Doug, where does one start (its difficult not to write a novel on this topic - perhaps its time I start making money and producing my own tutorial DVDs )...it can be daunting going from one shot colour to a mono camera with filters. Once you've been there however, its difficult to go back.
Have you worked out the G2V filter weights? Sure, manufacturers will indicate you can use 1:1:1 colour weights, but it's rarely the case that this is indeed correct. Its usually close however, so depends on how accurate or particular you want to get with your imaging.
I'd say you need to go longer on your RGB data. 30 second subs ain't that great, even for binned 2x2 data. While many focus on luminance, its easy to forget the importance of quality RGB. If time is an issue, sure, bin the data, but this can come at a cost. For wide field work where the arcsec/pixel is high - 1x1 all the way. There is little value of binning.
With regards to the combine process. A few ways to do it. Obviously treat the subs through each filter like any other sub. Closely observe its quality against others in the stack. Discard ones impacted by seeing or significant guiding problems. Aeroplane trails etc are fine to leave, they'll be rejected as outlier pixels by the combine algorithm. Create master R,G,B frames. I usually create the luminance master in the process as well. Once, you've got the luminance, register/align the R,G,B masters to it. At this point, if the R,G,B is binned, then it will be upscaled to match the luminance resolution. Unbinned RGB is simply registered. I don't register the RGB subs together and register it again to the luminance. Double registrations regardless of the form should be avoided at all time. The more times data is registered the further resolution can be lost due to the resampling. While in the case of RGB, its not a huge issue, but for luminance - take care.
With the R,G,B subs aligned, they need to be normalised. A process that equalises the background across the images. If not done, the colour weights you apply will not balance the images accurately. Very rarely will the background be equal across the R,G,B masters - mostly due to differing conditions i.e. western sky may have a high reading of sky glow or the moon more illuminated. MaximDL and other tools do this well. However, if you want more control - do it yourself manually. Why would you do it manually? - Because, automated normalisation tools don't always understand the data your dealing with. Gradients are a classic example of this. Anyway, using MaximDL, measure the average ADU in the same location across the R,G,B masters. This location should be void of nebulosity. If a gradient is present, measure the darkest point. Take note of the values between the masters. The goal is to make them equal ADU's around the value of 50. You can use another value, say 100, but the point being is they need to be equal. So, an example; R ADU is 954 counts, G ADU is 1282 counts, B ADU is 1183 counts. (Green tends to be high if light polution exists). So to get to 50 ADU, using MaximDL's pixel math function, you subtract from all pixels in the image - R:-904;G:-1232;B:-1133. Once done go back over the R,G,B master and measure the area again - it should now be as close as possible to 50 ADU across the images. They are now normalised are ready for the RGB combine. You don't need to perform this for every image you create, but if you're having problem in balancing the colours its worth the effort. Analysing your data is key to success as you'll begin to understand how far it can be pushed.
With the RGB master created, you can commence using this as a base layer in photoshop or as a colour, softlight etc blend in more advanced colour processing techniques to obtain the richness you desire. It may help to start off with straight RGB images and not add any luminance in the first instance (at least until you get the colour weights under control as luminance can alter this). Later you can start getting "crafty" and create synthetic luminance from a R,G,B masters to boost original luminance signal or you can try RRGB blends using a colour channel as the luminance. I've blended blue filtered data into the luminance data before to improve contrast for strong reflection nebulae. Many possibilities await, you simply need to experiment to see what works.
I feel certain many others can offer their LRGB tricks and tips. Good first attempt I reckon. Keep at it mate. The deeper down the hole you go, the more you find there is to learn - its a constant state of evolution.
Forgot to add...To further improve your chrominance data quality, try a "stepped" data acquisition. Shoot red filtered data while the target is relatively low (30 to 50 degrees), then shoot green, followed by blue filtered data as the target crosses the meridian. So your data acquisition imaging plan would look some like RRRGGGBBBBBBGGGRRR. In doing so, you minimise the impact of atmospheric extinction common in blue filtered data. This of course can also be circumvented by increasing colour weights based on target elevation at the time. I'm not always rigid in this method, sometimes I shoot based on moon phases. If the moon is not fully illuminated and low in the sky or has not risen, shoot green and blue filtered data. As it rises, switch to red filtered data has it is not as impacted by the blue reflected light from the moon. Once the moon gets too high or bright, switch to Ha. Manual normalisation will for the most part help in balancing the colours when the moon is about. You may have still need to deal with gradients. Cloudless nights are precious so anything you can do to maximise the output and quality of data obtained, can improve the final result.
Jase, Thank you so much. A lot there to work with and on. It looks more and more like the learning curve is going to be enormouse to say the least but at least I won't get bored while waiting for an imaging run to end. Plenty to do and learn.
I take it the 30 second subs are because you did this unguided?
First order of business would be to setup an autoguiding solution so you can get longer subs.
10 minutes is a good length for that chip. Too long and the small well depth mean too much bleed on bright stars. Too short and you won't get enough signal to get past noise. I'd say 10 minutes on most objects and perhaps 4 or 5 minutes on bright globs.
I get no banding at all on my 8300 chip. In fact the darks are quite flawless. Maybe something to take up with the camera manufacturer.
Excellent result Doug! After doing some work with Monte's M81 set I have a respect for the level of processing required. Phew! Lots of work in the LRGB game I can see. For a first-of-many I reckon you've done very well!
I take it the 30 second subs are because you did this unguided?
First order of business would be to setup an autoguiding solution so you can get longer subs.
10 minutes is a good length for that chip. Too long and the small well depth mean too much bleed on bright stars. Too short and you won't get enough signal to get past noise. I'd say 10 minutes on most objects and perhaps 4 or 5 minutes on bright globs.
I get no banding at all on my 8300 chip. In fact the darks are quite flawless. Maybe something to take up with the camera manufacturer.
Greg.
Hi Greg, The 30 sec images were taken because I was eager to see what I was going to get and because I had what looked like a reasonable image on the screen. I think the banding may be an aberation which may be removed by bias files but I have posted a querie on the QHY forum. Time will tell.
Quote:
Originally Posted by Omaroo
Excellent result Doug! After doing some work with Monte's M81 set I have a respect for the level of processing required. Phew! Lots of work in the LRGB game I can see. For a first-of-many I reckon you've done very well!
Thanks Chris, I didn't realise how much diference there was in putting an image together in RGB compared to an OSC camera. Nice to have the likes of Jase here to give us fools a hand.
I think the banding may be an aberation which may be removed by bias files but I have posted a querie on the QHY forum. Time will tell.
Doug,
It may help if you post one of your bias frames or ping me an email offline. Probably a stupid question, but you're not confusing column defects with banding are you? If the bias frame contains considerable banding, you're more than likely dealing with electrical interference. The bias being a zero second exposure, its guarenteed to not be thermal noise related. Check your USB cable is not too close to power etc, perhaps route them via alternate paths to separate/sheild them as much as possible to see if that changes the situation. It could be camera related, but out of the few times of assist people with this problem it has turned out to be interference or dodgy cable. Maybe worth investigating.
It looks like I may have a few problems, The filter wheel has the filters in the wrong positions, don't ask me how I stuffed this up but when I get the filter wheel sorted and a good set of data to work with I will see if the problems are problems or just me and my missguided ways.
I have usually gone for luminance when the object is near the zenith to get maximum clarity in the image.
Given that contrast and detail comes from luminance which has sharp boundaries and that colour changes gradually (hence the practice of some to slightly blur the RGB) wouldn't it better practice to make sure your luminance is taken at the absolutely best angles (just before and after zenith) and then your colour BGR?
One minor problem I have is I use CCDsoft and when doing LRGB colour shoot it names the file with the colour. However you can set the filter to anything so to do BGR it would name them RGB which can get confusing later with mislabelled colour images when you process. Perhaps Maxim allows to you to name the file correctly?
I am assuming you do all the imaging in one night here and not luminance in one or two nights and colour on another.
Greg.
Quote:
Originally Posted by jase
Forgot to add...To further improve your chrominance data quality, try a "stepped" data acquisition. Shoot red filtered data while the target is relatively low (30 to 50 degrees), then shoot green, followed by blue filtered data as the target crosses the meridian. So your data acquisition imaging plan would look some like RRRGGGBBBBBBGGGRRR. In doing so, you minimise the impact of atmospheric extinction common in blue filtered data. This of course can also be circumvented by increasing colour weights based on target elevation at the time. I'm not always rigid in this method, sometimes I shoot based on moon phases. If the moon is not fully illuminated and low in the sky or has not risen, shoot green and blue filtered data. As it rises, switch to red filtered data has it is not as impacted by the blue reflected light from the moon. Once the moon gets too high or bright, switch to Ha. Manual normalisation will for the most part help in balancing the colours when the moon is about. You may have still need to deal with gradients. Cloudless nights are precious so anything you can do to maximise the output and quality of data obtained, can improve the final result.
It looks like I may have a few problems, The filter wheel has the filters in the wrong positions, don't ask me how I stuffed this up but when I get the filter wheel sorted and a good set of data to work with I will see if the problems are problems or just me and my missguided ways.
A common problem Doug. I find it is good practice to get a torch and shine THROUGH the filter (some filters look to be a colour other than what they are - Astrodons for example). Select each filter and a short exposure whilst holding the camera in your hand or on a table and verify the filters are correctly labelled in your software.
LRGB. L is obvious looking at the subs. Blue is usually the least sharp but red and green are not that easy to tell apart. So that leaves RGB = possible combinations!
I have usually gone for luminance when the object is near the zenith to get maximum clarity in the image.
Given that contrast and detail comes from luminance which has sharp boundaries and that colour changes gradually (hence the practice of some to slightly blur the RGB) wouldn't it better practice to make sure your luminance is taken at the absolutely best angles (just before and after zenith) and then your colour BGR?
One minor problem I have is I use CCDsoft and when doing LRGB colour shoot it names the file with the colour. However you can set the filter to anything so to do BGR it would name them RGB which can get confusing later with mislabelled colour images when you process. Perhaps Maxim allows to you to name the file correctly?
I am assuming you do all the imaging in one night here and not luminance in one or two nights and colour on another.
Greg.
Hi Greg,
Yes, in the previous post I only referred to RGB subs, not luminance. You're correct in that luminance should be acquired when the target is highest, thus the stepped approach would see luminance collected in the middle of the run (dividing the blue filtered data (RRGGBBLLLLBBGGRR).
Given there are only so many hours in a night, I do find myself collecting luminance on another night. In fact, the recent Orion's belt wide field 4 panel mosaic image, I collected blue and luminance filtered data one night, then green and red the other. As you're aware, imaging with a monochome chip and colour filters takes time. Clearly, if your motives are to produce an image at the end of every imaging session, then a "one night stand" needs to accomodate all filtered exposures, which at times can lead to "skimping" on the quantity of data collected.
Everyone has different thoughts on this matter and it depends on your goals. I'm personally not concerned about pulling off an all night imaging session with nothing to show for it as I know that final image will benefit once I've collected sufficient data in subsequent nights. I've got a few projects waiting more data, so can go back to them when the timing is right. Its the old quality of quantity spiel...I'd prefer to release a single quality image every 3 months than shotty work once a month.
Can't say I've used CCDSoft extensively, so probably not much help there. I would have thought that if you set up an imaging sequence, the filtered subs would simply have 'R','G' or 'B' ammended to the file name. Just because the sequence is RGB, doesn't mean the each of the subs would be ammended with RGB does it? Doesn't seem logical to me...I'm not certain what you mean by setting the filter to "anything"... I use ACP for data acquistion which in turn uses Maxim, but its really for camera control, guiding etc. ACP is where the smarts are given its an fully fledged automated observatory control program. I've got ACP naming the subs with target name, exposure time, bin and filter, along with UTC timestamp so they look something like this M20-600s-b1-Red-0352.FIT. As the UTC timestamp will be different for each image, no files are overwritten/replaced. Of course you can ammend a number like 01, 02, to each sub, but a timestamp is more useful. If you can make the file names as meaningful as possible it saves time, especially when you're dealing with a large quantity of mosaic panels or huge data sets. Anything to make your life easier and improve the data aquistion efficiency is a good thing.
Looks good to me Doug! Definitely could use more exposure I think, but as a quick LRGB first go, its nice!
Im definitely looking forward to more LRGB imaging from you, and also looking forward to picking your brain a bit when my filters arrive... I've organised a CFW8A for my SBIG through Martin Pugh, So it seems we will both have to thank Martin for the ability to add colour to our images!
Very encouraging Doug with such short subs, I've got my eye on the QHY9, would be most interrested in those vertical lines, please post once you have more info.
Nice first colour image, I have no RGB filters for my QHY9 so I have a lot to learn. As for the bands if you leave the fast download button unchecked they should be gone.
clear skies Ken.
Looks good Doug
you're off to a fin e start!
imagine what the next images will look like with your filters in the right place and the knowledge gained from Jase and Greg!
(Solid gold, I dont own a mono CCD but making notes for the future!!!!)
frank
)...it can be daunting going from one shot colour to a mono camera with filters. Once you've been there however, its difficult to go back.
