19 June 2021: Our upload server https://upload.astropixelprocessor.com/ has been migrated successfully to our new office with higher upload and download speeds (nearly 10MByte/sec up/down ) ! We now have 1 general upload user called: upload with password: upload. The users upload1 - upload5 have been disabled.
31 May 2021: APP 1.083-beta2 has been released ! APP 1.083 stable will follow soon afterwards. It includes a completely new Star Reducer Tool, New File Saver Module, Improved Comet registration and much more, check the release notes here!
Artefacts inside Bayer/X-Trans drizzle integration result
Hi, I'm very very happy with APP : )
Recently, I started with narrow band imaging and got very good results using the tutorials.
I experimented with Bayer/X-Trans drizzle and found out, that despite having a limited amount of Frames (17), I got best results with drop size 0.5 and scale 2 compared to all other possibilities, including drop size 1 and scale 1.
But now, with a different set of lights, I get again quite smooth results but get a bit geometrical artefact in the image. This occurs only with scale 2. Why so ? Why exactly in than place ? The frames look all very similar only one being of much less contrast. I'll repeat this w/o this image but this result is not plausible to me.
Also I get a tiny linen like pattern, that's not as bad and is also slightly visible in drop 1 scale 1. But this is not so much a show stopper … but I hadn't this issue before.
All (successful results before and this problematic results) Images were made with the same camera, same dark and bias frames, different flats (Tair 3s 300mm lens vs. Jupiter 21M 200mm lens).
Any suggestions what could be wrong ? Thanks in Advance : )
I have to say I haven't seen those artefacts yet. But what I do know is that for proper drizzling, you do need well dithered and preferably undersampled data. Dithering meaning, 4-5 pixels and enough frames to be able to reconstruct the signal.
Thank you, the data is well dithered (more than 4-5 pixels).
I believe there is enough data because this are good frames with 16 minutes exposure @ ISO400.
In all other areas the result is super smooth. This is a local problem not being present in the lights. And, as the shape or position may change slightly, I do not belive that is originates from the data.
I've asked Mabula to chime in here as I've not seen that before specifically. I'll report when I know more.
Ok, so could you try to increase the droplet size to 0.6? Mabula also mentions 17 frames is not a lot of data for a droplet sizes of 0.5 so more data for that would definitely be better.
This solved the issue for me. Thank you very much : )
The first pic below might be interesting for you. This is the stretched difference between droplet 0.5 and 0.6 results.
It reveales many more artefacts I didn't notice at first glance. Only the top right one drew my attention.
Only the linen like pattern is still there (third pic) but I'm confident it will be gone with more frames.
Yes, that has everything to do with the amount of data. The more you have (well dithered), the better APP can "fill in" the pixels that are required. The regular pattern you see is an indication the algorithm worked well btw. 🙂 But with more data it can get rid of that as well.
You're very welcome. 🙂 Good luck with your processing!