Using DSLR movies for Astro Image stacking application

[sophiecentaur]

TL;DR Summary

Does the compression system used for movies in DSLR cameras interfere with the selection and stacking of individual frames?

I have used my entry-level Zwo colour camera to get movies of planets and, subject to my limited skills, those movies (Avi format) are handled quite happily by Autostakkert and Registax. I have used my DSLR successfully to take multiple long exposure images and they have stacked fine (of course). But my Pentax K-s2 stores movies as .mov files. Despite converting the format (several different apps) to .avi I find that the stacking apps won't load those files. It's not a file size problem.

After a few hours of trying different things, I thought about what those movie files from the K-s2 actually consist of. The compression does its best to take all the information of all the frames and stuff it into as few MB as possible. Mpeg went to a lot of trouble to produce compressed files that are subjectively good to watch. Processing into .avi may not actually result in [Edit: the original ] separate individual frames, which is what the stacking progs want.

What experience do other members have of using DSLRs for planetary imaging? Thing is, it's a lot less trouble to pop a DSLR on an unguided scope and let the software get the best out of the resulting movies. The image sensor size may not be ideal etc. etc. but people do seem to do it. Is it me, the camera or the choice of software? I guess my problem is that people who 'just do it' aren't aware of my problem so they haven't considered it. I just need to find that one member who has solved my problem.

 

[russ_watters]

I have used my entry-level Zwo colour camera to get movies of planets and, subject to my limited skills, those movies (Avi format) are handled quite happily by Autostakkert and Registax. I have used my DSLR successfully to take multiple long exposure images and they have stacked fine (of course). But my Pentax K-s2 stores movies as .mov files. Despite converting the format (several different apps) to .avi I find that the stacking apps won't load those files. It's not a file size problem.

AVI is a generic file wrapper, not an actual file format (compression/decompression scheme). Do the different apps let you choose the codec to use when converting? "Uncompressed" is still a specific codec and will have a name, such as yuv-12:
https://wiki.videolan.org/YUV

That's what my DMK camera uses and it works in Registax.

Mpeg went to a lot of trouble to produce compressed files that are subjectively good to watch. Processing into .avi may not actually result in [Edit: the original ] separate individual frames, which is what the stacking progs want.

The "m" stands for motion and "peg" is as in "jpeg". An Mpeg is just a really fast slideshow of individual jpegs.

Incidentally though, Registax 3 doesn't list mpeg as an acceptable format, but Registax 6 does. But I use 3 because the added features and complexity of 6 tend to not really work very well.

What experience do other members have of using DSLRs for planetary imaging? Thing is, it's a lot less trouble to pop a DSLR on an unguided scope and let the software get the best out of the resulting movies. The image sensor size may not be ideal etc. etc. but people do seem to do it. Is it me, the camera or the choice of software?

I'd say it is probably the firmware/processor on the camera. The problem is generating a useful data output from a DSLR video. DSLR sensors have a huge number of pixels and you can't download them raw at a high bitrate and in realtime. Not even a PC could handle that kind of data processing and transmission rate. So lower resolution and compression is required/baked-in to the camera. And it's probably the compression that's the bigger problem. You don't just lose detail in the compression (and resolution with available formats), but you lose color depth, which hurts you a lot for stacking. JPGs are just 8bit or 256 levels per color. My 10 year old Canon T3i is 14 bits or 16,384 for a raw file.
[edit]
For perspective, a full resolution raw image from my 18 MP camera would be 94MB. Or for a 30fps video, 2.8GB/sec. If it could convert in realtime to HD that would still be 300 MB/sec.

 

[sophiecentaur]

An Mpeg is just a really fast slideshow of individual jpegs.

The M stands for Motion (of course) and the Experts Group were concerned in the optimal coding of Moving Pictures, starting with a very slow frame rate. The temporal subsampling of movies is a severe impairment (wagon wheels going backwards etc.) and a simple sequence of jpeg images cannot deal with that. Mpeg uses temporal filtering with earlier and later images in a sequence to improve on that and Astro Movies are not helped by that. AVI files are very bulky because there is less processing (bigger individual images). I have a feeling that the best way to use the large sensor in a dslr is probably just to take a sequence of stills at as fast a rate as the camera can handle it. Then the stacking / moving software has the most to work with. If a movie is compressed to give a small .mov /mpeg file then damage has already been done before it gets to Registax. Mpeg isbuilt for human viewers and not astro software.

For perspective, a full resolution raw image from my 18 MP camera would be 94MB. Or for a 30fps video, 2.8GB/sec. If it could convert in realtime to HD that would still be 300 MB/sec.

Yes - a sequence of full images, 18MP per frame would be the best to watch and the best to process but camera manufacturers don't need or want that. The same thing is true with audio. CD format is just too data hungry and very few people notice the difference when they listen to their collection of hundreds of compressed sound files, squeezed into their iTunes library.

 

[Ibix]

An Mpeg is just a really fast slideshow of individual jpegs.

You seem to be decribing the M-JPEG format, which is distinct from the MPEG format.