Video recording: Is there an interruption between files or not??

Started by abooster, January 09, 2016, 12:34:28 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

abooster

I understand, ML can automatically restart video recording once a file reaches the 4 GB limit
(which is what attracted me to ML in the first place),
but the big question is:
Is there any interruption (in the video) between the individual files??

I tried searching but there seems to be no real information on this subject.
So, if I buy say Canon EOS M (which I gather is supported by ML),
how does it work in terms of video?
Will I be missing seconds (or frames) in the video footage? Or is it completely seamless?
If it's not completely seamless, what are the suggested workarounds when shooting something like an interview in 1080p (where it needs to be seamless)?
Is the only workaround having a second camera?
Or does ML restart the video recording seamlessly so that I won't me missing any frames?

I can't be the first one asking those questions (prior to buying the camera),
but for some reason, I can't find any answers here.
HELP please!

Walter Schulz

Roughly between 0.5 to 1 seconds doing H.264 recording (native Canon movie mode with MOV output).
No gap with RAW/MLV recording.

Remark: Older cams will stop H.264 recording if file size reaches 4 GB or at 29:59 recording time (whatever comes first). Newer cams will not stop after 4 GB. The cam will generate splitt files (<4 GB) without gaps until 29:59 is reached.

abooster

Thank you, Walter, but I'm not much wiser.
The interviews I need recording definitely will go over the 29:59 minute mark.
(typically they will be around 40 minutes or so, but some might go up to 70 minutes)

I'm guessing that RAW recording will probably generate unacceptably large files, correct?
How large will be a RAW file after 1 hour of continuous 1080p recording at 30 fps? (I have no idea)

Is MLV recording the same as RAW? Or is it a smaller file size?

Are there any workarounds to cover the 0.5-1 second gap?
(other than having a second camera)

Regarding newer cameras generating clean 4 GB files up to the 30-minute limit:
Sounds very good, but how can I find out (before buying) whether a particular camera generates clean 4 GB split files up to the 29:59 minute mark?
I mean, if I know for sure that I can get clean, uninterrupted recording in 1080p at 30 fps up to the 30-minute mark, then that would be good enough. I would just set an alarm clock for 29 minutes, pause the interview, restart the recording and then continue the interview (not ideal, but acceptable).
However, in this case, I have to know for sure, before buying the camera that this will actually work that way.
Any ideas where/how I can check that?

Walter Schulz

MLV/RAW: 1920 x 1080 x 14 x 30 / 8 = 103.8 MByte/s (and 5D3 is the only cam coming close)
7200 seconds will give you 730 GByte.

No way bypassing 29:59 limit.

4GB splitting: Find out by reading manuals. I think all cams with DiGiC 5 (or newer) do this.

But if you are about to buy gear why not buying a camcorder/video cam without those limits?

abooster

Well, correct me if I'm wrong, but as far as I know camcorders have very shitty video quality as compared to something like Canon EOS M or Canon EOS 700D.

dmilligan

Something made explicitly for video recording, not as good as something whose primary purpose is not video recording? Really?

abooster

@dmilligan
Yes. Not if you compare the quality you get for the same amount of dollars (as far as I know).
I recall a video with a founder of Magic Lantern saying something to the tune that ML turns a regular cheap DSRL into a camera/camcorder that would normally cost $100,000 or something.

I also noticed that virtually all camcorders don't have lenses that are wide enough for my purpose.
I need at least a 24mm equivalent lens and pretty much all camcorders seem to fail in this department.
(what they call "wide" is not nearly wide enough)

On the other hand, I'm certain that if you shell out $3,500+ for a camcorder, you'll get a wide enough lens and sufficient quality. But that would be a HUGE beast (which I certainly don't want). And I also don't feel like spending $3,500 on a camcorder if I can buy a DSLR for $500-600 that would produce the same or better video quality.
Are you with me?
Or am I missing something here?
I think ML was *primarily* created with people in mind who want to use DSLR for *VIDEO* recording.
I mean, the guy who created ML, created it specifically because he wanted to use DSLR for video.
Yet, you seem to be completely oblivious of that, dmilligan.

So, my reasoning is that:
If the pro who created ML is using DSLR for video (and not camcorders),
then this is the way to go if you want quality for a reasonable price.
Simple, right?

But I might be totally wrong. So, let me know.

abooster

@dmilligan
I was quite surprised that you seem to be a seasoned user in this forum and yet you don't seem to know the basics in regards to DSLR for video vs video camcorder.
Here's an old youtube video that will give you a quick run down in 4 minutes:
https://www.youtube.com/watch?v=1SLn-wY6Vrk
At about 3:30 in that video the guy (who seem to know what he is talking about) explains why the now obsolete Canon 2Ti/EOS 550 does a better job when shooting video than a $20,000 camcorder.

Long story short:
I don't know about you, but to me it seems that if I can buy a DSLR for $500 that will produce better video than any $10,000 camcorder... well, it's a no-brainer to me.

baldavenger

This is the funniest post that I've read in quite some time. Absolute gold.
EOS 5D Mark III | EOS 600D | Canon 24-105mm f4L | Canon 70-200mm f2.8L IS II | Canon 50mm f1.4 | Samyang 14mm T3.1 | Opteka 8mm f3.5

reddeercity

Every camera has a propose , Think of ML enabled camera as a "film camera"
It has it limitations , just like everything in life .
Most clips only needed to be no more then 1-5 mins long for most video/film project's
For that reasons ML is perfect for that type of work , but if you need to do endurance recording
ML will disappoint you , unless you invest in extra gear & hardware (hdmi hard drive recorder) .
so if you had a Real Video Camera you will have no limitations , and better Compression internally.
When I do long interviews/events 15 mins+ I use a Canon HV20 (bought used for $175.00CND) and connect the hdmi (clean & uncompressed) to
my atomos ninja record(bought used for rental store $325.00CND with 256GB SSD) and save it to ProResHQ and I'm really to edit asap
My investment is $500.0 and image is amazing .
right tool for the job.
Don't get me wrong ML can endurance record but you will have to deal with a lot a data
or suffer a second or 2 lost data with h264 ,
unless you get another cam & stagger the start time but then again you investment so up .

abooster

Aha! Excellent tips reddeercity! Thanks a lot!

So, it looks like the only possible (and sensible) workaround options are:
1) Get a second DSLR and stagger the start time to cover the 1 second loss
or
2) Get an HDMI hard drive recorder for continuous shooting in RAW.

Well, OK. Good to know. I think I'm gonna go with #1 then (shooting with the second camera at a 45-degree angle).

dmilligan

I almost don't want to dignify this thread with a response but there's a lot of misinformation that I feel needs addressing.

Quote from: abooster on January 09, 2016, 08:18:57 PM
I think ML was *primarily* created with people in mind who want to use DSLR for *VIDEO* recording.
Quote from: abooster on January 09, 2016, 08:18:57 PM
I mean, the guy who created ML, created it specifically because he wanted to use DSLR for video.
I don't know where you're getting all this nonsense, but here we prefer facts and arguments supported by logic, references, and real evidence:

ML is a basically a continuation of the CHDK project. ML was created and is maintained by people who are mainly embedded software programmers, reverse engineers, and hackers, who also happened to have a secondary interest in photography. The person who started it is named Trammel Hudson (this is the same researcher who developed the thunderstrike exploit) . His website: https://trmm.net/About . From reading that page, I gather that he is more interested in photography than videography. This is also true of many of the developers including myself (a1ex, the current maintainer and main developer, has also expressed that he is more interested in the photography than videography).

Personally, I found and got involved in this project because it is at the intersection of two of my interests: programming/electronics and photography. I'd say a large portion of the contributors to this project are in the same boat. This is a hobby for us, we do it for fun, because we find it interesting, not because we want to make $500 cameras into $10,000 camcorders for our budget feature films (which is why most of the people who aren't contributors to this project are here).

All the videography interest in ML is more or less just a side effect. Raw video was just sort of stumbled upon in the process of reverse engineering these cameras, but this is what caused an explosion in popularity of ML in the videography community. This has, btw, more or less ruined this forum with the endless questions from people who found out about ML from a videography blog and come here and fill up the forum with the same damn questions over and over again, instead of having interesting technical conversations where we figure out more about how the cameras work (like this).

Your particular question or something similar is probably asked once a week, other popular ones include: how to enable raw video module? how to downgrade Canon firmware? why can't I record for more than x seconds? what are all these pink dots in my videos? why is color/white balance off in my raw video? when will you release the next version?

Quote from: abooster on January 09, 2016, 09:04:21 PM
Here's an old youtube video that will give you a quick run down in 4 minutes:
https://www.youtube.com/watch?v=1SLn-wY6Vrk
Thanks, I now understand that even though DSLRs are terrible at everything having to do with videography, they have crap tons of bokeh, so therefore they are superior. This explains everything, I was totally in the dark.

Quote from: abooster on January 09, 2016, 08:18:57 PM
I also noticed that virtually all camcorders don't have lenses that are wide enough for my purpose.
Quote from: abooster on January 09, 2016, 08:18:57 PMBut that would be a HUGE beast (which I certainly don't want)
Quote from: abooster on January 09, 2016, 08:18:57 PM
I also don't feel like spending $3,500 on a camcorder
Is this wide enough, cheap enough, and small enough for you? $200, ~14mm equivalent, weighs 2.6 oz, shoots higher frame rates and higher resolutions than even 5D3.

Quote from: abooster on January 09, 2016, 08:18:57 PM
Yet, you seem to be completely oblivious of that, dmilligan.
Quote from: abooster on January 09, 2016, 09:04:21 PM
you don't seem to know the basics in regards to DSLR for video vs video camcorder.
You're right. I clearly know nothing about ML or DSLR video

axelcine

@abooster - I have worked as a journalist since 1981, produced radio 1994-2004, TV 2004-5 and education ever since. I suggest, that you wrap your interview techniques around what the hardware can do. And never NEVER harrass people with 40-70 mins interviews. That's something a bad cop in a bad movie does. Give the folks a cup of coffee while restarting your cam, let them go to the bath room, have a smoke or whatever. It's got something to do with respect.
EOS RP, 5dIII.113/Batt.grip, 5dIII.123, 700d/Batt.Grip/VF4 viewfinder + a truckload of new and older Canon L, Sigma and Tamron glass

Markus


abooster

dmilligan, my friend, re:
QuoteI think ML was *primarily* created with people in mind who want to use DSLR for *VIDEO* recording.
Quotethe guy who created ML, created it specifically because he wanted to use DSLR for video.
I got that information directly from the guy who created Magic Lantern.

And since you seem to be keen to support your statements by "facts" and "evidence" etc.,
you don't seem to be too keen to support your own claims by facts and evidence, do you?
Your claim that
QuoteML was created and is maintained by people who are mainly embedded software programmers, reverse engineers, and hackers, who also happened to have a secondary interest in photography.
is simply factually WRONG.
ML was created by ONE single person by the name Trammell Hudson.
I know that despite being a total noob here because I checked the corresponding Wikipedia article and more importantly watched a video with the founder himself where he explained the ins and outs of this software and the reasons for creating it in the first place.

So, unless that first paragraph in the Wikipedia article contains factually incorrect information and the video I mentioned was recorded by an evil clone/imposter of Trammell Hudson, it is YOU who are factually incorrect here and that's what's prompted my original reply to you because you must be a seasoned user here (and I am a total noob).

But despite being a seasoned user you had demonstrated cluelessness in that particular point and I had to reply because it was related to my topic. I'm not trying to be rude or something. I simply always tell people straight what I think, be it face to face or online.
And I don't try to sugar-coat things when there's nothing sweet.

QuoteHis website: https://trmm.net/About . From reading that page, I gather that he is more interested in photography than videography.
Aha, that's where you got it from. Fair enough.
Well, my friend, that about page doesn't really contain much information.
(Plus, he might not be doing video anymore.)
But in the video I watched he stated explicitly that he originally created ML to use DSLR (Canon EOS 5D) for shooting video.
Absolutely NO ambiguation there. For shooting video. Not photography.
Everything he originally put into ML was designed for shooting video.
Specifically, tools for tweaking/controlling audio in video recording is what he started with (when creating the first version of ML) and then later realized that he could also add lots of other cool features.

That's why I was quite surprised to see that a seasoned user like you doesn't know these basics and felt the need to correct you.
No more, no less.

Well, I do understand now that over time ML has developed into a tool that is primarily used by photographers. OK.

QuoteThis is a hobby for us, we do it for fun, because we find it interesting, not because we want to make $500 cameras into $10,000 camcorders for our budget feature films
Fair enough. I was just talking about the origins of ML and the guy who got it started.

QuoteYour particular question or something similar is probably asked once a week
I would have much preferred NOT needing to ask that question but instead to find the answer on the website! (FAQ or something)
I tried to search the website - found nothing.
Tried to search the forum - found nothing.
That's where I had to register as a user to ask the question.

So, if you guys feel annoyed by the same questions, it's your own fault.
The most basic FAQ would greatly cut down the number of those questions.
But maybe people running this forum would prefer to force people to register in the forum by not proving FAQ?
(I know some forum owners prefer that)

QuoteIs this wide enough, cheap enough, and small enough for you? $200, ~14mm equivalent
Thank you for the pointer, but I'm not looking for a fisheye effect.
~24mm equivalent is what I'm looking for. (sorry, should have mentioned. camcorders tend to have 29mm which is not acceptable)

QuoteI clearly know nothing about ML or DSLR video
I never said that.
I was merely talking about the "basics" by which I meant the origin of ML and in that respect your statements were/are wrong.


abooster

@axelcine
Thanks for advice but I don't really care what you do or what you have done, mate.
I just know what is time-tested and proven to work regarding the interviews I'm planning.
That's good enough for me.
Most of my interviews will be cut down to 30 minutes (but the cameras will probably be running for about 35-40 minutes) and some of the interviews will in fact be 60 minutes, yes.
Well, I'm guessing you've never done the kind of interviews I'm planning. Nevermind.

Markus

Your personal issues that seeps trough between the lines of your posts are overwhelming. =D

axelcine

@ abooster: I am truly happy with my cams, my ML's and my methods. I was taught by some of Denmarks finest media people and I passsed my exams with pride. And I'm happy with what dmilligan and a lot of other fine people have done for photography and movie making, including Trammell Hudson, Shane Hurlbut and all the contributors to the ML project. If you're going to revolutionize film making, take this discussion up with a few other skilled Danes, Thomas Vinterberg, Lars von Trier (Dogme 95), just to name a few.

I really don't think this thread is going to be very useful to any of us.
EOS RP, 5dIII.113/Batt.grip, 5dIII.123, 700d/Batt.Grip/VF4 viewfinder + a truckload of new and older Canon L, Sigma and Tamron glass

baldavenger

I say we need more advice from someone who seem to know what he is talking about.



EOS 5D Mark III | EOS 600D | Canon 24-105mm f4L | Canon 70-200mm f2.8L IS II | Canon 50mm f1.4 | Samyang 14mm T3.1 | Opteka 8mm f3.5

Walter Schulz


axelcine

Original question was answered in reply 1.

Last: Sorry Walter, your reply came just as I was posting. But I fully agree.
EOS RP, 5dIII.113/Batt.grip, 5dIII.123, 700d/Batt.Grip/VF4 viewfinder + a truckload of new and older Canon L, Sigma and Tamron glass

dmilligan

Quote from: abooster on January 10, 2016, 09:41:49 PM
The most basic FAQ would greatly cut down the number of those questions.
Top of Page -> User Guide -> FAQ
How do I record for more than 12 minutes?
Quote
Lower the `bitrate`_ (CBR 0.4 will let you record continuously for 30 minutes).
Use `Movie restart`_, but you will lose a few seconds when a new file is created.
To record continuously for more than 30 minutes, you need to use a HDMI recorder. Enable the Clear Overlays feature to hide the focus box and the 16:9 bars, and make the half-shutter button sticky to prevent the camera from turning off LiveView after 30 minutes.
Technically, there's no 12 minute limit. There's a 30 minute limit and a 4 GB limit, whichever comes first. With default bitrate settings, the 4 GB limit is reached after around 12 minutes (more or less).

Audionut

And the sticky post in this section.

The sticky to end all stickies. Read this first.
Quote from: Audionut on May 30, 2014, 04:57:54 PM
How do I record continuously for longer then 30mins or greater then 4Gb with H.264?

You can't!