Raw video on 5DMK2

[deknk12]

With a SanDisk 16GB Extreme 90mb/s card, I hit the 4gb file size limit at 1880x880 and 1880x720 with global draw off.

Now, if I could only get the DNG out of that file...

Quick sample:

https://vimeo.com/66530384

[JackDaniel412]

With a SanDisk 16GB Extreme 90mb/s card, I hit the 4gb file size limit at 1880x880 and 1880x720 with global draw off.

Now, if I could only get the DNG out of that file...

Quick sample:

https://vimeo.com/66530384

Really? With the same CF I arrive to a maximum of 600 frames at 1880x840...

[cmac]

Just shoot a small - 4-5 frames file at the same resolution, then make a cmd copy (if at windows) from the prompt go to the directory and do > copy file1(4gb) /B + file2(fewframes) /B testfile /B
/B - indicates binary read and write. Then try to process testfile with raw2dng. Hope it helps!

[Kenji]

Thanks to the developers for making this possible.

Im trying a couple of benchmark tests but cannot get happy results.

Just got a Lexar 1000x 64gb and its pretty terrible (1024 partition alignment)

Any ideas? Maybe the 5D mark II cannot handle 64gb that good. Dunno

[haysuess]

Honestly I wouldn't worry about
I just recorded the same thing on the 5D Mark 2
1880x720 24p 306 frames

default format
1024 align
4096 align

all of them were still 306 frames
no increase in performance

But the fact that you can get 1880 and I only get 1280 with half as many frames is really bad. Like the card is half the speed it should be compared to others. I guess I just have a bad one, but the benchmarks suggest otherwise.

[mageye]

I do apologise. More flowers

RAW Test - KomputerBay Professional 32GB 1000x CF UDMA7 - Part II

http://www.youtube.com/watch?feature=player_detailpage&v=RBt9nvER--o

[moomilk]

Anyone tried Transcend 1000x card? They got to be pretty stable quality and capable of up to 120Mb/s write speed.

[Redrocks]

Komputerbay 64GB 1000x: perpetual recording upto 1880x840 using 20 May build from a.d.

[Lukasz]

Hi, wanted to add my Lexar 16gb 400x RAW test put out off 134 frame clips in 1880x720 which I am able to do. Continuosly 1880x540, only on first builds.

[deknk12]

Really? With the same CF I arrive to a maximum of 600 frames at 1880x840...

Sorry I misspoke. I went well beyond 600 at 1880x720 but I'm now thinking I was at 1280x720 when I hit the 4gb mark.

[domisol]

May be this could interest devs. Had an issue with some underexposed and overexposed frames.
5D2 with ML Build 19 Mai by forum member "a.d."
I uploaded the frames here :
http://cembalo.free.fr/ml2
000092.dng is normal exposed frame
000093.dng is under exposed
000099.dng is over exposed

[a1ex]

M mode without any sort of HDR enabled? Changed any setting while recording?

[domisol]

A1ex, yes, M mode, no HDR, no exposure change during shoot.
however, I must admit this occured only once on the series of tests I did yesterday.
thanks

[mvejerslev]

I appended some more test footage (iso tests) and upgraded my Vimeo account so it can display 1080p:

[4135541355]

When 5D2 ml shot raw1880*1080.

CF card choose Toshiba 1066x or Relisa 1000x?

Card fast enough?

Useful please advice, thank you!

And, when shooting raw video, audio synchronous do

Hope ml can shoot.Raw and recording function

[CrewOfOne]

I've finished a (small) complete project, shooting RAW on the Canon 5D2!

Here's my report: http://crewofone.com/2013/canon-5d2-raw-vs-h264-in-real-world-production/

Here's the BTS video.

[cmac]

Test the moire with 1X and 5X modes. For me there is less "bad/false color" moiree on the 5x mode. Still some bayer moire is present, but I think that is something that is hard to pass unless You shoot with foveon or 3CCD setup.

I expected the 5x to require 20mm lens against 100mm at 1x, but in fact it ended with 35mm very close to 100mm - shot from tripod.

I did some successful recovery of files, but sometimes it extract only some frames.
I think that if someone more advanced that could explain which filed in the footer correspond to resolution, speed and frames that could restore a lot of work. I use HxD as a hex edit, but that might not be of importance. I also might be doing something wrong, because I'm just adding the footer to the file. I usually shoot short several frames before each session and then start the real shoot.
With a slower 2GB SanDisk I constantly hit a limit of 670MB - even when I set skip frames and lower the stream to lesss than 20MB/sec (that's with the 19th may build).

[DAK29]

Greetings! First post from me here, just want to open by saying congrats to all devs involved and major thanks!

I have a MARK II 5D and am just beginning to get into shooting. Am wondering a few things.

We have an anamorphic lens. My understanding is unsqueezed this fills the entire frame, correct?
And then the software performs the squeeze which displays it letterboxed on the 4:3 camera screen, right?
But the letterboxed 4:3 display on the camera is not actually the resolution being recorded...
The full frame is still recorded, regardless of if you have an anamorphic lens on it or not, right?

So why should you ever change the recording ratio? Or are people talking about only changing the playback ratio?

I've heard people saying that you can set ML to record 1920x(Insert various cropped vertical sizes here) to do anamorphic.
It would be wonderful if I could record full RAW with anamorphic lens but technically, the full 1920x1080 frame is still being
recorded unsqueezed, right? What is the point in cropping then? Wouldn't you be cropping the full unsqueesed image?

I know that on Blurays that play in 1920x1080 but are 2.35:1 - they open a true 1080p screen but only display 1920x817 right?
But isn't this just for playback? Is there an advantage to using anamorphic which let's us record in 1080p because the
full vertical resolution isn't being used or is this just during playback because the full vertical resolution is used during recording?

Sorry to be so verbose! Tried to explain that best as possible!

I am seriously eyeballing a Komputerbay 1000x 64GB card even though 65megs seems the max transfer rate.
Does anyone have any definitive proof that it is a hardware limitation and that no greater speed will ever occur without
hardware modification? Each build seems to bring different results - sometimes better, sometimes worse.
I keep remembering back in the old days when they got an extra 4k of RAM out of the Commodore 64 by using the cassette buffer...
Software tricks can sometimes do amazing things! - And what about hardware mods?

I've heard HDMI is too slow to take the RAW signal. What about splitting the load out CF and HDMI simultaneously?
(Dump raw binary data partly out hdmi and partly to CF in order to grab the full thing then join together later?)
The way I understand, the CPU is fast enough, it's just the CF interface which is bottle necking it, correct?

Thank you for taking the time to listen!!!

P.S. - I had to use a crappy SD to CF adapter to install ML. I have a 300x CF card but unfortunately, no USB adapter for it!
(I have a USB adaptor for the SD, that's why I had to use it!)

Normally, I would just get the files from the camera off the 300x card by hooking to the camera's USB port. However...
It seems the camera won't let me WRITE to the card from the USB so I can not get the ML files onto my 300x card until I get
a CF USB adapter which I plan to order with the Komputerbay 1000x 64GB later this week. In the meantime I am wondering...
Is there a way to enable write to CF card from camera's USB port?
By the way the horrid SD to CF adapter benches at about 1.7 megs hahha so it is useless!
I can play with ML and look at all the functions but I can not record anything!

[marten]

First, thank you all developers!

Made a little test video with my 5D mkII. Using a Sandisk 60mb/s card I used the 1600x720 resolution. I got anywhere between 10s to 60s before first framedrop. When I got the shortest clips I restarted the camera and then I could get longer clips again.

I'm really missing synced audio in camera. It was a nightmare to sync the short clips throughout the movie. I have not found a consistent way to sync the recorded audio files. Have I missed some post about how to do this?

The blown highlights (actually they are not blown but very close) are intentional and made in post. It was so amazing to be able to work in Lightroom the same way I do with stills.

[ted ramasola]

Nice video, I'm testing audio sync as well.

My tests were at 24P - 2frames delay of video so nudge back audio 2 frames on timeline.

30p - 4 frames delay of video so nudge back audio 4 frames on timeline.

What are your observations?

[a1ex]

So, audio starts a bit too late?

[Yoshiyuki Blade]

For me, the audio loses sync continuously and maybe not even smoothly. It starts off close to sync, but gradually shifts out of sync. I don't know if it's the fault of the program I use (Avisynth with the ffms2 plugin), but audio remained in sync with the same program in the "vanilla" recording mode.

[ted ramasola]

So, audio starts a bit too late?

I guess its the video that starts 2 frames "late." I put it in quotes as I don't know the technical reason, could be it starts same but don't record 1st 2 frames of video, thus the 2 frame "lag".

The only file process I do is the raw2dng.

I then open the dng sequence in AE, conform framerate to 23.976 to correspond to recording framerate.

I did this test with a "clapper" in the head and tail and some short recording of short speech in between.

[marten]

I made my recordings in 25p.

When I do that the audio starts before the video. This is easy fixed in post. But the audio is not running at same speed as movie so its out of sync quite soon.

So I thought the audio is recorded in "24" or "23.97". I slowed down my footage to that and tried to sync. It was much better, but still drifted after some time. In a 90 sec clip it was synced half way, but then got out of sync. I couldn't get it consistently synced over that clip.

Do we know exacly at what speed audio is recorded? My initial test makes me wonder if the speed even varies throughout the clip.

[a1ex]

Audio is recorded at 48 KHz, without any sort of sync. I'm not sure how to do that...