Magic Lantern enables 14-bit RAW CinemaDNG on 5D

I just saw that he gets strange results after the first file.. What does one file translate to in real time? Lol, sorry didn't see that you posted again. Still wonder how long "the first file" is.ail

There's also something really strange going on in that video...in the upper right section of the grass. The shadow detail is increased and the detail retained outside is much better....but wonder what that effect is. Doesn't look like typical skew...it's just odd.
 
Last edited:
It's interesting. A little over a year ago, the MKIII was announced, and folks were, by and large, underwhelmed. It was better than its predecesor, for sure, but seemingly not a massive improvement. Shortly after, Digital Bolex and then Blackmagic, announced cameras that record 12-bit raw for less than the cost of the MKIII. This video now comes to mind:


So now, here we are a year+ later, and it looks like the MKIII might actually get raw recording capability before Blackmagic ships in quantity, and before Digital Bolex ships at all. I have no point to make. I just find the irony mildly entertaining.
 
Last edited:
Wesley Byram said:
So now, here we are a year+ later, and it looks like the MKIII might actually get raw recording capability before Blackmagic ships in quantity, and before Digital Bolex ships at all. I have no point to make. I just find the irony mildly entertaining.
Click to expand...

...and also I remember around 3 years ago, Philip Bloom posted an April Fools post about a canon DSLR being able to record raw and boy a lot of posters on his blog were dissing him for it! ...funny when you think about it now.

http://philipbloom.net/2010/04/01/canonraw/
 
Blackmagic just weren't prepared for the Russian winter. The M43 BM has an extra stop and a bit of dynamic range and the S35 BM has a global shutter but I'd still rather shoot with the RAW 5D.
 
A bit off topic, but right now seems like a critical time for the "$3000 raw cinema" camera.
1.) BMCC- EF mount is an odd choice for the sensor size, M 4/3 seems like the right mount.
2.) BMCC 4/3 - Bloom said this mount makes perfect sense, beautiful image, but..MAJOR SHIPPING Issues, no one has one.
3.) Bolex - It's anybody's guess when it will come.
4.) BMCPC (4K) - Who knows if that will come on time either, with the M 4/3 still M.I.A.
5.) Canon 5DMK3 - Could Magic Lantern pull magic out of their hat and make this the "one"

Lol, I'm not speaking as a consumer here...or even about that mythical perfect camera. I'm mainly looking at it from the point of view of wondering who will be the 1st company to really nail things. Black Magic seemed to have it down 1 year ago, but now it's not so clear. The EF BMCC is the only choice at the moment, we'll see how it all shakes out.

Lol Wesley...we were thinking the same kind of thing...did not see your post as I was writing this.
 
The camera is just part of the RAW equation. @ 3Mb per frame I figure I'll need at least 20 terabytes to shoot my feature. So with 3 backups and room for proxies, cutting and grading I figure I'll need at least 4x 16 terabyte raid arrays which adds up to $6000. Like anything sexy RAW ain't cheap.
 
Yeah, it seems like a ton of data. That's why I don't mind the BMC pocket cam will initially only shoot Pro Res 4:2:2. That's all I'm ever going to use. Lol, those are big enough files as it is.
 
Wesley Byram said:
So now, here we are a year+ later, and it looks like the MKIII might actually get raw recording capability before Blackmagic ships in quantity, and before Digital Bolex ships at all. I have no point to make. I just find the irony mildly entertaining.
Click to expand...
Several suppliers (including B&H) have BMDCC in stock right now.
 
Improving on the image from old cameras like the 5D also makes the migration to new raw capable cameras viable.
Now you have a B cam or C cam already in your possession when you acquire a new raw capable cam.
 
squig said:
It looks like maybe he had optical flow or something on in AE.
Click to expand...

The way the grass moves does look a little weird. It almost looks like inaccurate motion compensation in the video encoding (Youtube?). Or, and I'm just speculating here, but could it be that what we're seeing in this video is actually the result of raw .DNGs taken at, say, 12fps and then interpolated up to 24p with optical flow, or twixtor, etc. On the other hand, if that's it, I would expect some noticable warping around the dog, which is moving more.
 
Last edited:
Ted Ramasola said:
I'm also wondering what the lens is, if IS was used. The ALL I version also has that but not so pronounced.
Click to expand...

It's hard to glean too much from a Youtube compressed video, but the movement of the grass in the all-I version looks like ordinary parallax to me. There might be a touch of MC distortion from the video compression, but nothing too out of the ordinary. The raw-sourced video just looks...different. Maybe this is the "strange results after the first file" that was mentioned?
 
I've shot RAW with an earlier build in silent pic burst mode and there were no strange motion issues but I haven't tried lv_rec mode yet. There's no motion problem when the dog moves so it seems more like a judder issue.
 
Last edited:
I would wait for samples from somebody else, or for words from a ML developer: I could fake this video real easy, just put a light inside so the DR required is not that huge, shoot it in video mode, once with contrast=-4 and once with contrast=+2. Not saying this is a fake, just that it could be.

Anyway, I'm still optimistic about this RAW thing. I know the camera can write the data fast enough for this resolution, the issue they were fighting for the last couple of weeks is somewhere else (they have to keep the camera from choking itself by doing lots of stuff that are not needed, like debayering the image as if it was a still).
 
Hehe, the only way it could have been faked is if he shot it with another camera, look at the resolution. There's no binary available yet and it's a bit beyond my skill-set to create my own so I can't test it myself till one of the ML devs throws me a bone but 75Mb per second on a 1000x card is totally doable. There's been a lot of activity going on @ ML command over the last few weeks and a lot of breakthroughs. It wouldn't surprise me if there was a stable beta available in just a few weeks time.

I had an incident last week when my MK3 recorded 24p RAW DNGs continuously until I stopped recording but when I checked the files they were just a lot of white noise, but the files were 1.6Mb per frame and that was on a slower 600x card.
 
pulpfiction007 said:
2.) BMCC 4/3 - Bloom said this mount makes perfect sense, beautiful image, but..MAJOR SHIPPING Issues, no one has one.
Click to expand...

Besides that, why would you order one knowing that Blackmagic has the ability to do a powered m4/3 mount now? They need to cancel this product and re-release it with the powered m4/3 mount from their pocket camera.
 
Its starting to look like they are going to get this to work.

"Heh, people got excited without even knowing the big news: g3gg0 just discovered how to use the DMA cropping routines, which just made possible RAW video recording at 1920x1080 at 24fps on 1000x cards." from a1ex
so far so good :2vrolijk_08: in the next month or two maybe this might be a real solution.
 
You must log in or register to reply here.
Back
Top