that would take a great deal of the pain out of the process, but not all. so you're back to batching the solves if you could somehow iteratively provide the hints to the solver then at least that part could be automated. with 100 panels i don't think any one hint could ever suffice. however just the ability to batch-solve would be huge, and although ImageSolver can do that, it needs a hint which lies somewhere reasonably close to the center of the image being solved.
#PTGUI LINUX REGISTRATION#
the registration in MBC is custom and runs in javascript, and is single-threaded, so as you get closer and closer to the final mosaic it's going to become super slow.Īnyway i think Wei-Hao's point is well taken, nothing automatic could really ever do this properly. Oh you totally can, but at some point you'll be working with the entire thing even if it is just 4 panes or something it's still going to be a lot of pressure on the javascript engine in MBC. all are active in the PI forums.Īny reason you couldn't mosaic smaller batches of frames, then mosaic the mosaics? or if andres pozo (who wrote ImageSolver and MBC) or john murphy (who wrote PMM) are interested they might be able to help. Possibly if you got juan interested in this you might get some support. but it requires all the images to be run thru MBC first, and you need to assemble the mosaic frame by frame with PMM, which might get old with such a huge mosaic. There is a new mosaic tool in PI called PhotometricMosaic which does the panel blending using photometry. MBC does give you the option to set the pixel dimensions of the output so that could mitigate memory problems, though then you'd compromise the size of the mosaic. if the script doesn't contain enough GC hints i can see it falling over - since each output frame is the union of all input frames, i can imagine the output frames being absolutely enormous. but the problem is with 100 panes i can see the MBC script easily being overwhelmed, as PJSR of course uses automatic garbage collection for memory management. i think you have to use PI's solver since MBC needs all the distortion parameters that ImageSolver calculates. The MosaicB圜oordinates tool in PI is automatic save for the need to plate-solve each pane, which is what you'd want to script. and in PI this would mean writing Javascript under PJSR which is not impossible of course but there is quite a large learning curve, outside of just trying to learn PI itself. The 100 panel thing is what makes this super hard. Astro subjects don't generally move, and they are all at infinity so parallax isn't an issue either. The big problems with stitching in conventional photography are parallax and motion. The situation is unfortunate because astrophotography ought to be relatively easy to stitch. One possibility that has occurred to me is that I could use PixInsight or (maybe APP or SIRIL if they output it) to get lists of registered stars, then use those as control points for Hugin or PTGui. If PixInsight users tell me that it is great for this then I will try to climb the learning curve. I know however that it has registration and stitching features.
#PTGUI LINUX CODE#
I think the problem is that their code that looks for control points that matching one frame to another don't work well for star fields. But I have had a hard time getting them to work. They should work for gigapixel panoramas. They should work well, and because they are based on the old PT Tools code. Hugin and PTGui are made for conventional photography. SIRIL also has a serious bug with stacking unless the stacks are perfectly aligned.ĭeep Sky Stacker works well for stacking, but hasn't work well for me for mosaics.ĪSTAP works well for me for plate solving, but it also supports stitching and mosaics, but so far it just crashes with my big images.
SIRIL does stacking, but not mosaic stitching. A different algorithm would avoid this, but so far it is what it is. stitching N panels takes time proportional to N^2). It also uses an algorithm that seems to have N^2 complexity (i.e.
#PTGUI LINUX SOFTWARE#
It's been a hard slog finding software that works, so I am asking for your experience.Īstro Pixel Processor (APP) works very well for small mosaics, but it is very memory limited. I am stitching large (100 panel) shots of the Milky Way.
0 kommentar(er)
