Raw video on 5DMK2

[moet_chandon]

Hello,

one strange thing, when i change the picture quality in original menu it changes the number of frames i can get.

Example:

- select jpg M in original menu. i get 579 frames @ 1880x960/24 (DNG Size 3.122KB)
- select Raw in original menu. i only get 343 frames @ 1880x960/24 (DNG Size 3.122KB)

[ted ramasola]

Hello,

one strange thing, when i change the picture quality in original menu it changes the number of frames i can get.

Example:

- select jpg M in original menu. i get 579 frames @ 1880x960/24 (DNG Size 3.122KB)
- select Raw in original menu. i only get 343 frames @ 1880x960/24 (DNG Size 3.122KB)

and if you totally disable raw and leave just jpeg you will get even higher number of frames. 

[Frostypb88]

Got a chance to fly the MKII on a cinestar 8 tonight. Wanted to see how much shadow could be recovered. I was able to hit the 4gig limit using a 8gb sandisk 60/mbps at 1600x720. Ill post the video later as soon as it uploads. Only issue i ran into is 4GB = 1:23, so not much flight time on it.

[savale]

Since there is a (small) buffer it is probably possible to write a shot to multiple files on the card to create longer shots. If creating and opening the file takes a bit too long it would be nice if this could be done in advance:

i.e. when you press record it creates and opens lets say 10 files. The only thing that needs to be done is to change the write pointer from one file to another during the recording. I might try to create something like that this weekend.

[xvince1]

Don't know if it will fit on the right thread...

As I have seen the test between RAW/H264 and GH2 hack on 150Mbit/s, I was wondering if a good CF can handle the 3x bitrate on the "normal" ML firmware instead of RAW. Personnaly, I use 2.3x bitrate with a komputerbay 64Gb 600x wich is a bloody shit (benching only à write 22 mb/s). So maybe with a good CF, the 3x codec bitrate can be a real good alternative for the Mk2

Waiting for a real good 128Gb or 256GB CF to buy.

So any comment about write speed and format system files on CF (Perso, I have seen some huge difference between cluster size on the komputerbay, but as I can't format on FAT32 in windows (don't know why, I can only NTFS or exFAT, and sadly, the mk2 do not recognize the card in exFAT)

[savale]

Cool stuff going on for the 5d2 in the 14to12bit or 14to10bit topic. This might make higher resolutions and/or longer recording possible

[lokki19]

Hi guys, first time posting.

I'v been reading a lot about the raw for 5dmkii and decided to give it a try and shooting stuff that is not plain landscape, I wanted to see how it behaved with fast movement and close up. This clip was shot in 1280x480 with forced 24fps. I could not get anything bigger do to my card speed (kingston 16GB 266x). I recorded an average of 480 frame before the buffer filled up. There is no color correction apply, everything is ASIS.

I notice that there is still a bit of noise in flat surface, the strait lines across the FOV are a bit jaggy, and there seams to be a slight purple tint. I don't know if it's cause by the built or my sensor how is showing signs of wariness.

I'll do more test with newer built as the come.

[ted ramasola]

For those with a 5DmkII and a VAF filter, I tested that the filter is essential at 1x resolution.

However, its mushed up soft when used at 5X crop mode.

Here is a new chart test at 1880 x 1080 & 1920 x 1078 comparing the DNGs with a still CR2 sraw2 frame and also with a chart at h264.

This is using NB may 22.

[GR Productions]

Hey folks,

I thought to open a specific topic for all discussions related to raw video on the 5dmk2.
 
Thanks
Oli

any chance of this fo the 5d3?

[cmac]

There is written 35mm - but in fact it's 24mm - I made the first test with 35 but the zoom didn't matched - 35mm at 5x is closer to 100mm.

Here is my extract on 5x vs 1x - I think there is much less color moire at 5x mode - shot with 24 mm lens. The normal moire is visible in both shots, and unless we have 3 cmos or foveon it will be there.

VAF filter @ 5X is not good, because it's meaning is to blur the image to a level that the same portion of the image (pixel) is covering several horizontal lines - so the are even in color - kind of bluring them toghether. So at 5x it will do more harm then good.

[tihon]

There is written 35mm - but in fact it's 24mm - I made the first test with 35 but the zoom didn't matched - 35mm at 5x is closer to 100mm.

Here is my extract on 5x vs 1x - I think there is much less color moire at 5x mode - shot with 24 mm lens. The normal moire is visible in both shots, and unless we have 3 cmos or foveon it will be there.

VAF filter @ 5X is not good, because it's meaning is to blur the image to a level that the same portion of the image (pixel) is covering several horizontal lines - so the are even in color - kind of bluring them toghether. So at 5x it will do more harm then good.

Thank you for the test. Now i don`t think that i really need to get vaf.

1920 x1078- i wish it be 1880 x 880 to... The interface speed not gives us a write speed to take this resolution normally. Maybe 14bit to 12 bit will solve that.

[haysuess]

Hello,

one strange thing, when i change the picture quality in original menu it changes the number of frames i can get.

Example:

- select jpg M in original menu. i get 579 frames @ 1880x960/24 (DNG Size 3.122KB)
- select Raw in original menu. i only get 343 frames @ 1880x960/24 (DNG Size 3.122KB)

I'm seeing the same thing. I only got 150-170 frames with RAW selected as my photo capture type. I disabled RAW and chose S JPG only, and got 290 frames! The buffer seemed to have 2-3 extra stars. Very cool!

[domisol]

Hi
I want to report bad recording in 5D2 with X10 MODE.
Running 21 may ML build, by a.d.
First, the problems are different withing even frames (00, 02, 04...) or odd numbered frames (01, 03, 05...).
Situation 1 = 1920x1078 : even frames badly encoded, like noise information. odd frames nearly ok, but with a few bad lines at the top or bottom.
Situation 2 = 1440x720 : even frames OK, odd frames are bad, in the sense that they are just the repetition of the same frame (frame #01).

Also, I want to say that even if recording properly @23,97 fps in X1 mode, it switches to 29,97 in X10 mode.

I'm currently uploading some frames at the following URL :
http://cembalo.free.fr/mlx10
Hope this helps.

Jean-David

[No Def]

Haven't seen this posted, but there are a lot of posts now 

When using a Marshall Monitor connected via HDMI the frame marker is offset and inaccurate.  Seems to be pushed left consistently.

Has this been the experience of anyone else, or addressed already?

[DAK29]

SAVALE - "cool stuff going on for the 5d2 in the 14to12bit or 14to10bit topic. This might make higher resolutions and/or longer recording possible "

Link please? These forums are so damn big it's hard to find stuff! Thanx!

I said in my first post here that I remember back when an extra 4k of RAM was gained for the Commodore 64
by using the cassette buffer and that you can never underestimate the genius which can occur with software
given the proper amount of time and people putting effort into it. Hope this news is good!

[ted ramasola]

Haven't seen this posted, but there are a lot of posts now 

When using a Marshall Monitor connected via HDMI the frame marker is offset and inaccurate.  Seems to be pushed left consistently.

Has this been the experience of anyone else, or addressed already?

same with my SWIT monitor. Not yet addressed but already posted before.

[gandalfk7]

SAVALE - "cool stuff going on for the 5d2 in the 14to12bit or 14to10bit topic. This might make higher resolutions and/or longer recording possible "

Link please? These forums are so damn big it's hard to find stuff! Thanx!

I said in my first post here that I remember back when an extra 4k of RAM was gained for the Commodore 64
by using the cassette buffer and that you can never underestimate the genius which can occur with software
given the proper amount of time and people putting effort into it. Hope this news is good!

http://www.magiclantern.fm/forum/index.php?topic=5601.0
enjoy!

[abaeza]

Haven't seen this posted, but there are a lot of posts now 

When using a Marshall Monitor connected via HDMI the frame marker is offset and inaccurate.  Seems to be pushed left consistently.

Has this been the experience of anyone else, or addressed already?

SAME OVER HERE

[Hazer]

Ok silly question -- how do I keep the LCD from shutting off while filming longer clips?  Not that I really *can* film long clips just yet, but you know, I'm an optimist.

Edit:  Never mind, *both* Canon and ML power saving options need to be switched off.  Carry on...

[ted ramasola]

For those curious to know about the rolling shutter "bendy thingy" in raw dng mode, I made this test with 2 cameras camparing raw video vs h264.

Since I only have one 5D2 and a 7D, I shot them first both at H264 to create a baseline behavior.
If both have similar behavior in h264, then I can use the 7D to do a side by side with the 5D recording raw while the 7D at h264.

You can draw your own conclusions.

[Africashot]

For those curious to know about the rolling shutter "bendy thingy" in raw dng mode, I made this test with 2 cameras camparing raw video vs h264.

Since I only have one 5D2 and a 7D, I shot them first both at H264 to create a baseline behavior.
If both have similar behavior in h264, then I can use the 7D to do a side by side with the 5D recording raw while the 7D at h264.

You can draw your own conclusions.

Thanks for posting this Ted,
IMO rolling shutter is so bad in all of them; it probably will not even make a difference if RAW or not... or I'd have to compare it looking at freeze frames, but heck; if its that bad on h.264 anyway why not just stick to RAW and enjoy all the other benefits!

[ted ramasola]

@Africashot,

Yeah, those are very extreme aggressive whip pans just done to provoke the "rolling shutter effect" for this test. I was doing this since I've read in some post in other forums that some say RS is worse in raw dng shooting. I think its about the same.

[DAK29]

gandalf - Thanks! Sounds promising but no actual module for download yet. Hopefully soon!

It appears that if they can get 12/10 bit working, we will easily be able to do RAW 1920x1080 on MK2D.

This is fucking awesome news!!!

[NitromanX]

Well said Dak29 - that is f*cking awesome news !!!

All we now need is to fix the moire and extend the record times on the 5D2 !

In my work rolling shutter won't be an issue as they are whip pans ...

Is it likely we'll be able to get 25fps too - I'm in PAL land UK ?

Damn you Magic Lantern guys are Gods ! 

[Yomommassis]

Personnaly, I use 2.3x bitrate with a komputerbay 64Gb 600x wich is a bloody shit (benching only à write 22 mb/s). So maybe with a good CF, the 3x codec bitrate can be a real good alternative for the Mk2

Waiting for a real good 128Gb or 256GB CF to buy.

So any comment about write speed and format system files on CF (Perso, I have seen some huge difference between cluster size on the komputerbay, but as I can't format on FAT32 in windows (don't know why, I can only NTFS or exFAT, and sadly, the mk2 do not recognize the card in exFAT)

With my mark 2 I can't manage a 3.0x bitrate speed even with my Komputer Bay 1000x, and if I'm not mistaken it was my card that topped the benchmark chart

Windows cannot format FAT32 past 32GB natively, for this I used a program called FAT32Formatter.exe, but regardless I never noticed any speed increases with any chunk allocations or partition alignment/offset