Messages

I shot a small demo movie of Trap Focus on ML2.3rc2 on a 50D I have access to.   Somebody put my prior,  less informative,  demo into the ML Cinema page but this one is better and shows setup for a nice compelling use case.

Could somebody swap this in for the other one? (Thanks MIchael).

Yes, without review it is fast enough now. Like I said, two pics of the same car sometimes.  I may now need to put a half second sleep in there.

[Edit:]

I will look into that, thanks. That seems promising.

Edit: OK, that works, the thing sometimes even triggers twice on the same car moving through at 35 mph. pretty good stuff.

I like having image review on to confirm quality and functionality in a long session.  Is there any reason why ML needs to wait for image review? I am guessing not.

Thanks, but that does not really work for this application.  This will get me the second car but then if there is a third car close enough behind you do not reset fast enough for that. 

I am interested in improving the cycle time in Trap Focus on the 50D.  After the Trap Focus has taken a picture it takes a second or so to reset, it seems.  I am interested in improving this performance.

First, is the cycle time an issue of the 50D, for example if I get a 60D will it improve?  Or is it something in Magic Lantern? Or it is inherent in the Canon focus assessment system independent of camera model?


Here is a short video showing it in action.  At around 50 seconds in the video the green car triggers the shutter but it does not reset fast enough for the black car following it.

Trap focus on vimeo


Thanks in advance for any ideas.

20 metres  was a random number I just used for illustration.

The distance is maybe 6-10 metres. It is the case that my setup is very sensitive in that accidental triggers by things further away or nearer than my desired focus band seem to happen quite a lot.  The focus band I desire is maybe 1-2 metres deep and I frequently have to deal with false positives from one side or another with the F4 lens.

I am working up to just getting that F2.8 lens and wanted to check my sanity on these ideas first, thanks for the input.

Another question is noise in the focus assessment done by the body. Does anybody have measurements on this stuff?

   SD

I am using a 50D with 70-200 F4  with ML 2.3.

I put this arrangement to use with Trap Focus in CONT mode. I leave the thing sitting in out for a couple of hours and come back for my data, all well and good, I am getting  a stream of data but wish to really clamp down on edge cases.

Questions:

Obviously if my focus plane is say 20 metres away, it is not limited to those objects exactly at 20 metres, there is a range, from say 18 to 22 metres that will trigger the focus indicator in the camera and hence the shot.  The accuracy of this range is an issue for me.

If I use a 70-200 F2.8, will the inherently smaller depth of field hence narrow that trigger range?

Would a different camera body help?


   Thanks, SD.

I have the most recent/upcoming version RC1 running on 50D with 1.09 also, works fine.


Note previously I had downgraded the 50D from 1.09  to 1.08 to run the (then) extant version, that worked also.

I am using the trap focus on a 50D with the 2.2 release of ML but would like to use it repeatedly.  I am new to ML so may have missed this somewhere.

Specifically I would like to put the camera on a tripod,  keep the camera on focus half press, and just let pictures be taken as things come into focus.  Right now it only takes one picture as I keep it on focus half press. You have to release and half press again to get another picture.

Thanks,
     SD

EDIT: The Trap Focus menu item for me only has "Off" and "Hold", there is no "Cont."  Is this an interaction between 2.2 and a 50D?

EDIR 2: The 2.3 release has the CONT mode in there on my 50D, hence I am good to go.