Hi Mel,
Looking forward to the update. I'm curious about the upcoming auto-align feature... is it image based similiar to terrain mapping or plate solving techniques? Is there any chance of also adding autofocus routine to the mix? Then automatic lunar mosaic would be fully automatic. Also, is there ANY possibility of creating a similiar app for solar mosaicing? My ES127 and Daystar Quark would thank you! Finally how useful would partial lunar mosaicing be, i.e. only capture a portion of the lunar disk depending on the feature of interest? Just dreamin' out loud
Thanks for your effort and offering this great software for free. I think it would be worth purchasing. I love how software and hardware developers fill the various needs that arise and enable even better capabilities!
Ciao,
Mel
The auto-align feature is based on still images taken with the USB camera. First, there is some image normalization (brightness, contrast, resolution reduction). Then I use the computer vision package OpenCV to identify features in the image. The descriptors of those features are matched with the reference frame (taken at auto-align startup) to measure the image shift for each individual feature detected. The shift should be identical for all features, so I use a clustering algorithm to find outliers caused by misidentifications. All this works well with "synthetic" moon pictures which I created from old images taken through my telescope, so I hope that it will work well on the real moon as well.
I don't need a complete "plate solving" here, because I know that the image is just shifted by some pixels in x and y. Therefore, this part is easy. It only gets difficult by the fact that I have to deal with a blurry moon image, rather than well-defined star images where plate solving techniques are used traditionally. Translating the x,y pixel shift values in RA,DE corrections is straightforward because I know the camera orientation, the telescope's focal length and the sensor's pixel size.
I don't know how to implement a good autofocus for moon images. As far as I know nobody has solved that problem yet. I discussed the issue with Torsten Edelman, the author of FireCapture. After some experiments he gave up on this topic as well. Perhaps I will do some more experiments later when there is no more work on MPM.
I don't think that a similar program would make sense for the sun. The point is that the sun is changing too quickly. At a reasonable resolution you notice a substantial change after some three minutes already. I don't see any possibility to record the video files quickly enough so that everything is finished within this short time.
Restricting the action of MPM on a section of the lunar disk is possible already now. Before starting the video acquisition loop, you only have to "mark all tiles processed", draw a rectangle with the mouse around the tiles you want to record, and "mark those tiles unprocessed". This way MPM will record only those tiles contained in the rectangle. Is that what you dream of?
I guess that you wonder why I'm waiting so long before doing the real tests on the moon. The reason is that after a surgery at the moment I cannot lift heavy weight, such as my telescope. I hope that in August one of my sons will help me setting up the telescope in the garden for the final tests.
All the best,
Rolf
Thanks for the response and all the best to you for a full recovery.
Ciao,
Mel