Mosaic ? how to ?

[Marko of Oz (Mark)]

Hi people,
just broadly speaking, how do you do a Mosaic? I ask because you can't fit much sky on a dsi chip. eg: see attached quickie of Orion.

I think that's pretty much the core. But to get more, do I just point the scope at say 6 overlapping points(3x2 area or whatever) and then use photoshop or something similar to patch them together?

thanks

Mark

[jase (Jason)]

Hi Mark,
Mosaics are a challenge, but very rewarding. I've done a few with mixed results. The key component to mosaics is planning, planning and some more planning. It takes sometime to determine the best framing for an object and the quantity of frames required. Start with a small project i.e. no more than four frames to begin with. Below is some of the software I use when performing mosaics. I'm progressively learning what it takes to be successful in this field. To do it right takes dedication and patients.

TheSky6 for all planing of frame composition. It has an advanced mosaic feature factoring image overlap for alignment and guide star selection. You can determine how many frames are required based on your imaging train field of view and object size. Very powerful features. Mosaics are very difficult without the right tool to plan your frames.

Once you've got the plan organised. You then need to decide what your image composition will be. Conventional RGB is good for starters, but you may want to try LRGB etc. Keeping in mind the more frames, the more data acquisition will be required to deliver the final image.

I use ACP for frame acquisition. For mosaics to be successful you need to be able to have precise control over where your telescope is pointing. Sounds simple and with goto telescopes, how hard can it be right?... Wrong, with mosaics we are talking about high precision pointing. What I mentioned above about image overlap for alignment is important. If your telescope doesn't point with high precision it will not be possible to guarantee overlap is achieved. The last thing you'll want is for your frames to not align correctly due to miss framing. ACP does a few things to assist the mosaic imager. Firstly, it does plate solving. Plate solving takes ab image and matches it against a known star catalogue that has astrometry calculations to determine where the telescope is pointing. If the telescope isn't point exactly where you've told it to, then the plate solving process will remedy this by informing the telescope to move x/y coordinates. I've explained this in the most basic form, if you want to know more about plate solving, let me know. Secondly, TheSky Mosaic plan can be exported into ACP to automate the image acquisition process. It will simply acquire each frame of the mosaic as instructed via the script (plate solving as it goes to ensure each frame is precisely centered as per TheSky plan). Once you've got the data for each frame, you'd calibrate (flats,darks,bias), register (align) and combine it. So you end up with each individual frame which is noise reduced etc.

Now you're on to the fun part... image registration. Each mosaic frame needs to be precisely aligned before you load it into photoshop. I use software called Registar for this purpose. Very effective tool. Once each image is registered to each other, its a good idea to set the basic background brightness levels using DDP or pixel path. Each frame will certainly not be the same brightness so you'll need to compensate for this.

Then, bring each frame into photoshop. As they are already aligned, they can be pasted as layers and manually moved into position as required. Once you are happy the frames match - in particular the brightness between each frame is a smooth transition (so it looks seamless), the image can be flattened and image processed using your typical processing routines.

Here is one of the many resources available online to get you started - http://www.astromatt.com/Articles/MosaicsPI1.html

Enjoy!

[Marko of Oz (Mark)]

Wow, that's a great explanation and I've bookmarked that article for future reference. But it seems like mission impossible without some sort of goto/guiding capability.(which I don't have at present)

I'll just have to have a crack at it the old fashion way for now and try manually pointing, aquiring, moving on. There will be plenty of hair pulling frustrating nights ahead.

cheers

[avandonk]

Jase there is a calibrate function (F7) under Operations in Registar that will automatically adjust colour and brightness. The frames have to be registered first. Then it is simply a matter of average combining the calibrated frames to get the mosaic.

Bert

[jase (Jason)]

Thanks for the tip Bert. But i've used this Registar function a few times in hope that in contained "miraculous powers" in matching mosaic frames. But it fell short, with all hope rapidly turning to despondency.

What I didn't understand as a part of this Registar process was whether it took a sample of pixels from each image to match or the entire image. It is noted in the manual: "In order for a source image to be successfully calibrated, it must have at least 1000 pixels in common with the reference image. If this is not the case, no calibration is performed and a warning message appears." So does this mean it only takes a sample of 1000 pixels. Is only the background averaged and sampled - what happens if one frame has a bright globular cluster, will this be included in the calculation?

I guess the manual does warn you:
"Note: The Calibrate command is most useful if the source image(s) are similar to the reference image. If the source and reference images are inherently different, or have stars of substantially different sizes, this command can return unexpected results."

I do any brightness matching in photoshop with initial work either using DDP or pixel math - the old "clunky" way, but it works and is extremely flexible. Perhaps you've had more experience with the function and can shed some light on any preprocessing you performed to make it work as documented. If it can simplify the process I'd be happy, but perhaps there are no free lunches!