uncompressed 14-bit RAW video recording

[squig]

I'll do an audio sync test on the MK3 over the weekend. What build are you running on the MK2? Do you have another app like FCP that you can drop the files into just to rule out whether it's an AE problem. I'm having some issues with exposure flickering in AE and I just don't trust it.

I found this-

"The Zoom H2 seems to record audio at a very slightly different speed than most video recorders. The real problem is that non-professional audio recorders record at a sample rate of 44.1kHz. The audio quality is just fine -- CDs run at 44.1kHz -- but professional video cameras record audio at 48kHz (broadcast standard). That slight difference is enough to cause about a one- to two-frame (at NTSC video's 29.97fps frame rate) offset every minute or so. However, if you run these recorders at 48kHz, that will sync up with video from most cameras -- provided they are also running at 48kHz."

[1%]

In camera audio is 48k

[mageye]

I'll do an audio sync test on the MK3 over the weekend. What build are you running on the MK2? Do you have another app like FCP that you can drop the files into just to rule out whether it's an AE problem. I'm having some issues with exposure flickering in AE and I just don't trust it.

I found this-

"The Zoom H2 seems to record audio at a very slightly different speed than most video recorders. The real problem is that non-professional audio recorders record at a sample rate of 44.1kHz. The audio quality is just fine -- CDs run at 44.1kHz -- but professional video cameras record audio at 48kHz (broadcast standard). That slight difference is enough to cause about a one- to two-frame (at NTSC video's 29.97fps frame rate) offset every minute or so. However, if you run these recorders at 48kHz, that will sync up with video from most cameras -- provided they are also running at 48kHz."

Thanks for the response

I am thinking that this kind of problem will become quite common unless there is a distinct strategy to deal with it. I have been into audio production for some years now and am reasonably aware of sampling frequencies and what is used for what. I also understand about the mismatch (and assosciated problems) that could happen if the speed of one is not consistent with another.

Anyway since I got the ZoomH2 I have always used the frequency of 48khz 16bit. I chose this setting for two good reasons. One was for quality and the other was for compatibility. As I already said I had no problems synchronising the H264 video with audio from the H2. It's an enigma. I am not trying to fault the RAW video (to me its a dream come true) but I want to be able to work around possible problems. I want to develop a workflow that works and I am sure plenty more people will also want that too.

I think I may try using other programs (such as Premiere and FCP) and see if they produce different results. I am open on this one and if I find a way that works then I will happily tell the world.

Anyway thanks again to have responded so soon and presented some ideas.

[mageye]

Build is May22nd BTW 

[squig]

I was a record producer before I became a filmmaker, when it comes to audio sync I think in milliseconds. What did you have the FPS override set to? I'm assuming the video is perfectly smooth, no dropped frames or cadence issues. How much is the sync off by (time/frames)?

[mageye]

Yes the video is perfectly smooth. No problems with that.

I have the fps override turned OFF. If I go into the fps override and turn it on it just reads 2.502 and will record at that rate. I don't seem to be able to change the fps override (is there some way of changing this that I have maybe missed?).

My belief is that it's recording at 29.976 (that's what it says at the bottom when the RAW video option is selected)

Also I haven't been able to measure properly how many frames its offset yet. I think I need to record something more precise than my fingers tapping! something more definite (maybe a machine or something?). I was thinking about doing the clapper several times within a take that will probably be the most reliable so far for me to measure.

I like the line of enquiry. I have always wondered why I can't seem to change the fps override???

[squig]

Click on the "Q" button to go into "FPS override" then the "set" button and turn the set dial until you get 24 then scroll down to "optimize for" and select "exact FPS", it should show "actual FPS: 24.001". Also set H.264 video recording to 24p in the Canon menu, I'm not sure if that makes a difference but just to be on the safe side.

I just recorded 8 minutes of cookie monster with my Sony PCM D50, I'm transcoding the file to prores in AE and I'll try to sync it up in FCPX. Oh boy oh boy oh boy, mmm cookies!

[ted ramasola]

Yes the video is perfectly smooth. No problems with that.

I have the fps override turned OFF. If I go into the fps override and turn it on it just reads 2.502 and will record at that rate. I don't seem to be able to change the fps override (is there some way of changing this that I have maybe missed?).

My belief is that it's recording at 29.976 (that's what it says at the bottom when the RAW video option is selected)

Also I haven't been able to measure properly how many frames its offset yet. I think I need to record something more precise than my fingers tapping! something more definite (maybe a machine or something?). I was thinking about doing the clapper several times within a take that will probably be the most reliable so far for me to measure.

I like the line of enquiry. I have always wondered why I can't seem to change the fps override???

I too was boggled how I could change the framerate since what appears in the info below the raw  options is 29.97, even when fps was changed, then I realized I had to go to the CANON MENU and change the movie rate there from 30p to 24p, now the info in the raw option reflected 23.976.

[squig]

Ahh right, I never shoot 30p so I hadn't noticed that. I just synced up a minute of cookie monster, no drift. Trying for 10min now.

[mageye]

OMG I feel like such a moron!

I have been wondering how to access those 'Q' menus for some time . I really should read more manuals!

I have now managed to fps override to 24pfs. I don't know if this solves my problem yet as I am yet to test it with the sync setup (clapper board and stuff).

I do feel however, that I this will go a long way into solving my little problem if not completely resolve it. Thanks so much for your help.

I will do some tests in the day (today) and hopefully I will triumph. I will indeed post my results and conclusions.

Good luck with the Cookie mmmmm!!!

[mvejerslev]

Since we're lobbing everything in here, I thought I'd re-iterate the discovery of someone else here: Setting photo quality to small JPG in-camera instead of Raw, frees up buffer, at least on the 5DII. It enabled me to shoot twice the amount of frames at any given size before buffer overrun. It has no effect on the frame size in bytes on the DNGs. Perhaps the devs could look into exploiting this memory allocation?

[ted ramasola]

Since we're lobbing everything in here, I thought I'd re-iterate the discovery of someone else here: Setting photo quality to small JPG in-camera instead of Raw, frees up buffer, at least on the 5DII. It enabled me to shoot twice the amount of frames at any given size before buffer overrun. It has no effect on the frame size in bytes on the DNGs. Perhaps the devs could look into exploiting this memory allocation?

The devs have already noted this since the very first builds of ML raw, and that was recommended to testers of the nightly builds.
I think once every feature is stable , that this will be mentioned in setup.
I should suggest that if performance take so much hit when raw stills is enabled, that ML team should might as well automatically disable raw stills once raw video is engaged.

[Picture and Color]

Notice hot pixels flaring on her zipper below the pendant at ~1:03.

You're right... I'm seeing them too now. Not just in that one clip, but others too. Is this happening with anyone else's footage?

[squig]

You're right... I'm seeing them too now. Not just in that one clip, but others too. Is this happening with anyone else's footage?

http://www.magiclantern.fm/forum/index.php?topic=5614.msg39655;topicseen#msg39655

[aaphotog]

The devs have already noted this since the very first builds of ML raw, and that was recommended to testers of the nightly builds.
I think once every feature is stable , that this will be mentioned in setup.
I should suggest that if performance take so much hit when raw stills is enabled, that ML team should might as well automatically disable raw stills once raw video is engaged.

I do not like the idea of automatically setting this. I  am working with a 5D3 and I've tried setting my image quality to sRaw. Normally(with raw being selected) my buffer doesn't go beyond 1* and I can record until I manually stop recording(1080p raw). When I selected sraw though, I only got one second of recording before it automatically stopped on me.

[kbru]

Recording raw via HDMI to Atomos Ninja w/SSD possible???

[squig]

Recording raw via HDMI to Atomos Ninja w/SSD possible???

No, but a CF card slot to SSD drive option is being explored.

[squig]

Ok I've tested external audio sync on a 12min 24p file in FCPX, it's perfect!

[ted ramasola]

Now for those curious if rolling shutter in Raw video is different as in h264.

Here is a rolling shutter test on 2 cameras, in raw vs h264.

You can draw you own conclusions on the test.

[tihon]

Now for those curious if rolling shutter in Raw video is different as in h264.

Here is a rolling shutter test on 2 cameras, in raw vs h264.

it seems to mark2 in raw provide more jello:)

[budafilms]

Do you know when recording draw if the preset have any influences over the footage? And the WB? Thanks!

[ted ramasola]

Do you know when recording draw if the preset have any influences over the footage? And the WB? Thanks!

You mean picture profiles? like neutral, faithful, cinestyle etc? I think Not when shooting dngs. Also with WB as you can change all that in post.

[dhilung]

Now for those curious if rolling shutter in Raw video is different as in h264.

I think the actual image acquisition pipeline of DIGiC is the same for both cases until the RAW frames are copied over using EDMAC (raw_rec) or encoded in H.264 (normal). This means the rolling shutter effect should be the same.

[Lars Steenhoff]

The rolling shutter you are comparing is between the 5d and the 7d, that's not a valid test, full frame vs crop camera, even when if they had the same sensor.

[bumkicho]

Ok I've tested external audio sync on a 12min 24p file in FCPX, it's perfect!

Great to hear. I am assuming you set up your external audio to 48hz and set fps override to 24fps for this test, right?