DSO Mosaic Planning /Live Stack Mosaics

[Fir Chlis]

I normally do Live Stacking of DSOs, but I've just tried out the Lunar Mosaic Planner last night (too much hazy cloud for DSOs) and, with the auto align feature, I was very impressed with the result.

Which got me onto thinking, If I can do mosaics of the Moon when it's too near full to fit in my FoV, is there an easy way to get SharpCap to run mosaics on DSOs larger than my FoV (e.g. something like M31) – a DSO Mosaic Planner. I know that I could use the Sequencer to set up a series of captures, but I'd need to work out the centre point for each of the panels. Or do I use some other package to generate a list of RA/Dec targets and drop this into the SharpCap Sequencer?

And (maybe asking too much) but if SharpCap can /could do this, could I watch each of the panels Live Stack as the sequence progresses? I suppose that a manual way to do this during Live Stacking would be to capture one panel, then use Click to Recentre to walk around the target and capture successive Live Stacks, and subsequently stitch them together outside of SharpCap.

And then, flights of fancy here, I could envisage the Live Stacking image display allowing the user to set up a grid that would cover a large DSO, e.g. a 3x3, 2x3 grid, Sharpcap would live stack into one cell, recentre itself to the next cell, then live stack into that, and so on, with the Live Stacked mosaic gradually building up to form a full picture of the DSO.

Cheers


Geoff

[admin]

Hi Geoff,

great to hear that the moon mosaic tool with auto-align worked nicely for you

Right now you could use another tool to generate the panel centers for the deep sky mosaic and then use the sequencer step 'for each RA/Dec Coordinate in <coordinate list file>' as the repeat block containing all the stacking/capturing you want. That block can read co-ordinates form a text file with one co-ordinate per line and will GOTO each co-ordinate and then run the steps inside the block at that co-ordinate.

I would like to have some sort of deep sky mosaic tool - in order to make that useful though you need some way to identify the areas of the sky to be included, which is actually probably the biggest part of the work for doing that feature. The two parts of that are showing some sort of sky image so you can see what you are selecting and actually making a way to draw/paint/pick/set the area to be imaged. It all becomes rather messy, as to draw the sky image you either need to install a massive amount of deep sky image data with the application or have a network connection to download it... I haven't got a good answer for that yet.

cheers,

Robin

[Fir Chlis]

Hi Robin

Thanks for the reply – I can understand the issues involved with implementing a deep sky mosaic planner – having spent most of my career developing software in a similar genre to SharpCap.

I’m trying to think of the user experience for an EAA observer, watching an image develop in real time. For many, I suspect that (like myself) changing to a scope or camera with a larger FoV isn’t feasible, so the idea of being able to build up a larger DSO from smaller panels in (semi)real time would be attractive.

I’m wondering if something like this would work:

1. Specify a grid of camera FoVs, e.g. 3 wide by 2 high. The Live Stack display shows each cell, initially empty.
2. Specify the RA/Dec of the centre point of, say, the top left cell.
3. The above two values (cell pattern and RA/Dec) could be worked out using, for example, Stellarium, with the sensor reticule.
4. Specify a capture duration per cell, say 5 mins.
5. SharpCap Live Stacks into the first cell for the specified duration, saves the stack, recentres to the calculated centre point of the next cell, then continues to Live Stack each cell in turn until all the captures are complete. The EAA observer would therefore see the whole picture gradually build up.
6. There would likely need to be a (user specifiable?) overlap for each cell, though I don’t think that a very accurate alignment /stitching algorithm would be necessary, as I suspect that can only be done accurately with post-processing. However, I don’t think that an EAA user would be too concerned, provided that the cells were roughly aligned.

Here's a mock-up of what I was thinking, part way through a run. Although not as good as a fully aligned mosaic, I’d be happy with seeing this during a Live Stacking session if I could see big targets appear in real time, and I could always stitch the individual saved stacks together afterwards.

Live Stack Mosaic.png (841.08 KiB) Viewed 584 times

When I get a chance, I can simulate this manually, using Goto Dec/RA to Live Stack individual panels, and then load them into GIMP after each panel has completed. And that may well give me what I want.

Anyway, I’ll leave this for now and maybe it will /maybe it won’t get into a future release.

It’s good to be able to correspond with the developer, rather than a large corporation.

Cheers

Geoff

[admin]

Hi Geoff,

thanks for the suggestions on how it might work and for producing a mockup. I must admit that I hadn't even got as far as thinking of how the results would be displayed - more stuck on how the user might specify the area to include in the mosaic. Fortunately plate solving will let SharpCap calculate the orientation/size/position of the current field of view, and therefore allow the layout of the panels to be planned once the total area to be covered is known...

cheers,

Robin

[Fir Chlis]

Thank you. I've now produced a spreadsheet that will calculate the panel centres for a grid, having specified an anchor position that I obtained by planning using the sensor reticule in Stellarium, and I can copy and paste these from Excel directly into the SharpCap Goto text box. I've tested the procedure with Sky Simulator, loading the individual panels into GIMP while the next panel is stacking, so I'm all set to go on the next clear night.

Hopefully this procedure will give me what I want for now - seeing bigger DSOs in more or less real time.


Cheers


Geoff

[admin]

Hi Geoff,

sounds good - looking forward to seeing the results!

cheers,

Robin

[Fir Chlis]

Hi Robin

The second clear night on the trot allowed me to do a proper test on the procedure.

I went to M81 and set up the histogram stretch to produce a nice image and left it set at this level for the remainder of the test. I’d had a go the previous evening, but stretched each of the panels individually, which ended up rather like a patchwork quilt with different colours in each panel.
I then moved to the first panel, stacked for 5mins, Save as Seen, then copied and pasted the next coordinates and repeated the process for the subsequent 5 panels.

Stack_38frames_304s - Copy.jpg (250.7 KiB) Viewed 398 times

I opened the first panel in GIMP and resized the canvas to 200%/300% so that I could accommodate a 2 across by 3 high grid. I opened subsequent panels as “Open as Layers”, so that I could easily move them for alignment. The pictures below show the mosaic building up.

There are a few things that came up from this.

Alignment. You’ll see that the panels aren’t completely aligned, I suspect because my camera isn’t absolutely aligned with the axis of the OTA. Next time I’ll try the Click to Recentre functionality to move – though I’ll have to do this twice for each panel in order to move across by a full panel width. I think this would be less hassle than updating and then copying and pasting from the spreadsheet. A new reticule could help with this (although the Circle reticule can probably work well enough as a guide), rather like the shape of a St George’s Cross, with a vertical and a horizontal line bisecting the image. If the lines had graduations near their outer end (e.g. 5%, 10%), then it would be easier to pick the Click to Recentre point to maintain the adjacent panels close to orthogonal and to choose a suitable overlap.

Stretch. Subsequent panels followed on immediately after the initial setup on M81 (over the period of an hour) and although each of them is consistent with the others, they don’t match my original M81 image. I’ll have to check this again another night, but maybe just accept that I have to stretch each panel and that I may end up with a patchwork.

Pasting coordinates. Sometimes I couldn’t paste into the Goto text box and had to type in by hand. I’m not sure if this is an artefact of me pasting from my main laptop to the scope laptop via VNC, but it did work some of the time. Speculation, but does SharpCap prevent Goto while a stack is live? I wondered if it might be some ‘safety’ feature to prevent accidental movement while stacking. This becomes redundant if I use Click to Recentre.

Although the image isn’t quite what I wanted, it does demonstrate the process so I’ll try it out again and, once I’ve sorted it out, I can try on something more interesting like a 2x2 panel sized nebula.

Cheers

Geoff

Snap 1.png (149.2 KiB) Viewed 398 times

Snap 2.png (158.48 KiB) Viewed 398 times

Snap 3.png (165.18 KiB) Viewed 398 times

Snap 4.png (168.64 KiB) Viewed 398 times

Snap 5.png (172.13 KiB) Viewed 398 times

[admin]

Hi Geoff,

looks like a decent result for a trial run of that approach. Well done.

From the SharpCap side of things, there is nothing to stop you starting the GOTO procedure while live stacking is active - in fact in SharpCap 4.1 the GOTO box is aware that live stacking is running and will automatically save your stack before starting the move the telescope (and, optionally, start a new one when arriving at the new target). I don't think anything should block you from pasting co-ordinates into the 'GOTO Target at RA/Dec' box if there is valid text on the clipboard. It might be worth firing up notepad and seeing if you can paste into that on the target machine if you encounter a problem - if that has a problem then it is likely VNC related.

To get your camera lined up with the axes of the mount, you should be able to use the plate solving tool in SharpCap - when you run a plate solve, the notification bar shows the results and those include a bit along the lines of 'Up is 13.2 degrees E of N' - that means that the upwards direction in the image is 13.2 degrees east of pointing towards celestial north. In that case you would probably want to rotate the camera 13.2 degrees anticlockwise to make 'up' in the image be in the direction of the NCP (although if you have an odd number of mirrors in the imaging train, the direction will be reversed). Anyway, you can plate solve again to check for improvements and fine-tune.

The brightness of the panels is going to be tricky, as conditions may change during imaging, or the movement of the target across the sky may lead to background brightness changes due to light pollution gradients. I think I would be tempted to paste together the panels before stretching, but also to try to feather the panels into each other so that there are no hard edges.

cheers,

Robin