Frame blending / averaging

[Shadowfiend]

Hi everyone!  I'm new 

I have just purchased a second hand EOS 600D from eBay and I'm really looking forward to getting into the awesome features of ML.

I am running business that involves 3D scanning from photographs with Agisoft PhotoScan.  In order to produce flawless detail from photos I will be shooting objects on a turntable with a greenscreen, front lit with LED lights in RAW format.  However I want more ;c)

I had written a Photoshop script that took a number of frames and averaged them together to reduce noise and increase colour depth.  It works in 32bit float for the best possible quality.  For example;

I take 10 shots where nothing moves in frame for every rotation step of the subject.

If I rotate the object 10 times, that's 100 RAW flies to process, which leaves 10 frames to feed into PhotoScan.

The script loads the first frame and successively blends the following 9 frames one by one and then saves the final frame.  The blending amount or weighting is important, so each frame affects the final frame equally.  Because it only loads 2 frames at any one time there are no memory issues, but what I was hoping that it might be possible to the same thing in camera in ML.

I hope all that makes sense and I'm greatly looking forward to your replies.

Many thanks.

[walter_schulz]

Generally speaking:
What can be done in post won't be done in ML.

You may take up the fight why it makes sense to put it into ML. Present convincing use cases.

[dmilligan]

http://www.magiclantern.fm/forum/index.php?topic=14373.0
http://www.magiclantern.fm/forum/index.php?topic=13408.msg129520#msg129520
http://www.magiclantern.fm/forum/index.php?topic=10176.msg98049#msg98049

[Shadowfiend]

Thanks for the replies guys.

walter_schulz, this wasn't really a feature request, but it would seem from reading through the links that dmilligan kindly posted that there already is considerable interest in this kind of functionality with Light Trails Mode, EekoAdd and Magic Shutter looks very interesting.  I understand that if something can be done is post there needs to be justification for replicating this functionality in ML, but it would seem that there are major benefits and artistic freedom that these features would allow.

For me, the feature would mean getting stunning picture quality by averaging any number of frames together and no post processing saving a lot of time.  I'm sure that other people would be interested in this :c)

dmilligan, thanks for the links.

Guys, what are your opinions about this functionality?

Many thanks.