Recording Video for More Than 30 Minutes (600D/T3i or 700D/T5i)

Started by mishmash, February 03, 2015, 05:57:48 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.


Greetings... I am new to video and considering purchasing a Canon camera for stills and video. I have narrowed my choices to one of two models: a 600D (T3i), or 700D (T5i). I am satisfied that either will take quality still pictures, but I do have a specific requirement that the camera is able to record video continuously for over 30 minutes (using on-board microphone), and without dropping frames or adding small gaps if/when splitting files (I'll be recording events up to an hour long without break using both 24p and 30p). I'll be sure to get a good card and am willing to format in exFAT, if that helps. For which of those two does Magic Lantern support lengthy and seamless video?


Due to European tax laws at the moment you cannot go above 30 minutes.
T3i and Kiss X4 (550d (T2i)) Tamron 18-200mm, Sigma 28-70mm f2.8 (need firmware upgrade) Olympus 50mm f1.8  Olympus 28mm f2.8 and Olympus 24mm f2.8
Fancier 370 tripod and LCD hinged loupe. DIY Slider and crane.


Thanks the prompt reply. I was afraid that might be the case, and I suppose even ML can't work around certain limitations. I guess I could try to find an uber-cheap dedicated camcorder on the used market, but giving up the chance to try interchangeable lenses is a pity.


Well I think that when decreasing the bitrate from 1.0x to 0.1x you get longer recording times but worse quality. I definitely heard that this is possible but have need tried by myself as I'm more a scenic film maker. ;)


I've just learned something about the T4i (650D): it can record video 'continuously' for 30 minutes by creating multiple video files of either 4GB or 12 minutes in duration (whichever comes first), and it keeps doing this for a full half hour until stopping. And there aren't any gaps or dropped frames; it does so quickly and in synchronicity. You then concatenate the files together in a video editor to make a seamless movie. This behaviour is the default on an unmodified T4i and is a feature specifically mentioned in the camera manual. I'm quite sure the same would apply to a T5i since they are both so similar, but I haven't checked for certain.
So, is there now a way for ML to take advantage of this fact and lift the 30 minute time limit for at least the T4i (and likely T5i)? It seems plausible since the file-system limitation is taken out of the equation... Or am I wearing rosy glasses?


Why does one want to shoot for such a lengthy time frame? I'd doubt it would even fit on an CF card.


Recording a lecture at the college might well require up to 50 or 60 minutes of continuous recording.


There is a feature in ML called movie restart that tries to restart recording after it stops at 30 minutes. It causes about a second worth of dropped frames though. Can be found under movie -> movie tweaks when the camera is in movie mode.


Thanks everyone. Your replies have been helpful, and it looks like I'm getting a T3i! I've thought about a dedicated camcorder for the purpose, but that would limit me to video only, and I enjoy still photography from time to time.


I too am in a position of needing to record 2 hour lectures, hour-long interviews, and 1.5 hour musical recitals and it would be very useful if I could use my 700d to do it.  Is this something that can happen in ML?  is it being worked on?

What exactly causes this limitation?  clearly the 700d spans 4gb files pretty easily, and yet recording still stops at 29.9 minutes...

I'm aware of the European legal reasons that canon would choose to do this, but I'm in America and not governed by those tax laws and thus I feel perfectly justified in my want to circumvent such a trivial limitation. - is there something I'm not getting as far as a possible technical limitation?  I have a 128GB sdxc card that allows for a consistent 40MB/s write speed. (I believe the 1080p/30 h.264 encoding only uses about ~6MB/sec so there shouldn't be any issues there) which means, even if i rounded up to 10 to make the math easy, I should be able to store a good 3.5 hours AT LEAST.

If this is a technical problem, could someone please outline it for me so I can understand it better?

Thank you!


The technical problem is called "reverse engineering".

It's what we have to do to be able to do anything at all on the camera. It is a very slow, tedious, time consuming, unpredictable, and challenging process. The fact that you have to wait 30 minutes just to setup the conditions to begin investigating the problem and try something (and the another 30 minutes each time to try something else), makes it extremely time consuming (and rather boring) to investigate this particular issue. This combined with the fact that this is a volunteer/hobby project, and those of us who work on it, only do it for fun (so we usually only work on things that are interesting), and only have very limited amounts of time we can spend (which is nearly zero these days).


There are lots of issues when using a DSLR to record video. Basically, it isn't the right tool for the job you are trying to do. Video cameras have sensors designed for video and still cameras have sensors best suited for the demands of shooting high resolution single images. Large sensor video cameras that mimic the cinematic look of film and DSLR's have special cooling features because as the sensor heats up it develops more noise. Ok, so much for stating the obvious so short of suggesting to use a "real" video camera for shooting long takes here's a few suggestions.

  • Use more than one camera and stagger their starting record time, then edit the two angles together.

    This will make for a more interesting video and if one camera fails you won't have a total loss on your hands.

    Yeah, your budget is probably already stretched to the max to get your dream DSLR but maybe consider a couple of less expensive cameras? You will probably be shooting H.264 for those long takes and you won't need the bells and whistles of the higher end cameras for that. I have several EOS-M bodies that combined cost less than a mid-range DSLR.

  • Look beyond the Canon cameras that can use Magic Lantern.

    This may sound like blasphemy but ML wasn't made to solve every possible task you may encounter. A GoPro can shoot over 30 minutes, so can a Canon C100 (which is currently discounted to about the price of a 5D3 at B&H). Ok, that's a video camera but there are other cameras, DSLR and mirrorless, that don't have that half-hour limit.

  • Finally, if you really want to solve this time limit issue using Magic Lantern--

    • Learn how to use the special debugging builds like dm-spy-experiments and post your findings in a bitbucket issue.
    • Learn reverse engineering and how ML code works. (Ok, that's a lot harder.)

Here are some final pointers -- Get AC power adapters or tap into heavy duty batteries. Still camera batteries aren't designed for the continuous load of shooting long video takes and might run out of juice before your take is finished possibly corrupt the video file. If you want usable audio don't use the built in mics as you indicated in your post. Either plug in an external mic or record to a dedicated audio recorder and sync in post. With long takes you'll probably discover that the audio drifts so you'll have even more work to do in post.


My 2 Cents worth on this is , HDMI
Why not Capture the HDMI ? it can be clean (just need to toggle off the over lays)
According to the Feature comparison matrix under "Display" CLEAR_OVERLAYS
Much better the h264 by a long shot.
Rent yourself a Atomos Ninja   I know in Canada the rental is cheap Atomos Ninja2 Portable Field Recorder $45.00 for a weekend CND .
The big bonus with that you get straight to 10bit ProRes422HQ etc... which is worth it for the cost.
But be aware that the picture style profile will effect the out come of the image , not like raw where it doesn't matter.
I do this a lot with my 5d2 with my 1st gen atomos ninja & I usually go for a neutral (Semi Flat) picture style or what ever you like .
Now if all is equal with all cam that ML supports , you should get a uncompressed 4:2:2 8bit 1650x1080 60i/24p (24p embedded , need to do a 3-2 pulldown 60i->24p)
At least the t2i & 5d2 do , but the t5i may get you full 1080p never tested one thou you would have test to be sure.
To over come the Liveview timing out ,  do one of 2 things press the "half shutter button" to refesh Liveview .
Not to practical as you my bump the cam or move it , the other way is to make the "half shutter button" Sticky
go in to ML menu and make the button sticky.
You will still need to plug in a audio input e.g. Mic. to the ninja , as the hdmi has no audio on the canon dslr.
hope that helps