Canon EOS 5D Mark III vs Blackmagic Design Cinema Camera

[clicstudio]

Man the 5D III was blown away...
I am not a videographer so my view on the subject is purely impartial and for fun...

I am very disappointed that after 150 years of photography, nobody can come up with something close to what the human eye sees. I have been dreaming of a photographic camera that can deliver the kind of dynamic range the Blackmagic camera has.

No matter how expensive the camera, the dynamic range is still limited. HDR is a joke. Who has time to post process 3 photos to create what should be done in camera, and in real time.

I think it is totally possible to have a professional camera with 2, or 3, smaller sensors: One for highlights,  one for shadows, and mid tones with a real time preview and on the fly processing.
I wouldn't care for huge megapixels or frames per second.
I would rather take a 7MP camera with this kind of dynamic range than a huge 40MP camera that still "sees" One light...
No matter the price... a $199 pocket camera or a $7000 1DX, the problem is still the same. You have to choose just one exposure...
I really dream of the day where the output of my camera is the same thing I see thru the view finder.
For example, take a picture of a well lit interior on a bright sunny day or a silhouette against a sunset without using a flash...
Maybe the technology is already out there. Even cheap video cameras can get close, and now seeing this amazing Blackmagic camera makes you realize the original 5D II was used by cinematographers because, at the time, it was cheap in comparison.
Who would use a photo camera to do video if they had a choice? Nobody...
The Blackmagic will dethrone the 5D II and III as the inexpensive cine camera alternative...
As I understand, the workflow can be a pain but then again, videographers spend countless hours editing, so they should be used to it... And with results like these, who cares?

[dirtcastle]

Here's a good article comparing the Blackmagic and 5D3...

http://www.eoshd.com/content/8841/5d-mark-iii-or-blackmagic-cinema-camera

It will be interesting to see how this whole sensor/codec shakeup plays out between Canon, Nikon, Sony, Panasonic, Blackmagic, and anyone else who joins the fray.

While I feel like my 5D3's video capabilities are about to be dated, the bottom line is that you can make great videos with even modest equipment. Imagination still trump specs.

[Marsu42]

No matter how expensive the camera, the dynamic range is still limited. HDR is a joke. Who has time to post process 3 photos to create what should be done in camera, and in real time.

[...]

As I understand, the workflow can be a pain but then again, videographers spend countless hours editing, so they should be used to it... And with results like these, who cares?

Well, at least you can do hdr video with Magic Lantern - and if ml manages to get 1080p/50 or /60 out of the 5d3 at least the current dynamic range would be much higher. And in comparison to merging photo hdr shots the workflow on video is so time-consuming another step (merging & motion-compensating hdr) doesn't make that much of a difference?

[Axilrod]

I want to see how the GH3 does.... Panasonic has deep pockets, is eager, hungry, and talented. Canon cannot hold back, the competition is fierce and getting hotter.
I wish I was 30 years younger, such an exciting time for video and photography.

I've heard nothing but good things thus far, apparently their new ALL-I codec is very solid.  Detail levels on the hacked GH2 were ridiculous, comparable to the C300. 

Here's a short Philip Bloom shot with a pre-production model (I'd recommend downloading it, as Vimeo's compression definitely affects color/sharpness): https://vimeo.com/49420579

[Axilrod]

As I understand, the workflow can be a pain but then again, videographers spend countless hours editing, so they should be used to it... And with results like these, who cares?

It doesn't matter how long editing normally takes, time is money, either way the extra time to process footage is still an inconvenience.  And it's not so much the additional workflow as it is the cost of SSD space, I think 30 min of footage in CinemaDNG is 200GB.  Not to mention you need some serious processing power to even work with the files, it'll bog down even very high-end machines.

It's definitely not a perfect camera (but there isn't really such a thing), but  I'm still interested.  I may end up opting for the MFT mount instead of the EF.  I'd probably end up shooting in 1080p ProRes most of the time, but true 1080p is still going to look much better than the 5D3.

[hutjeflut]

makes you wonder how much better dslr's cn do when the software on the camera gets a totalo redo.
the sensor is obviously better yet the immages dont lie (unless they are falslky edited on purpose to favor the ciname cam.)

so with better software or better raw data  management the 5d2 should be a LOT better then the cinemal cam.
if the photo's are that much better thn the ciname fram captures than the movie in theory should be able to be be much better also as its nothing more then stitched togethet photo's after all.

all this movie shows to me is that either he wants to discredit the DSLR filmers or that the DSLR video software needs a LOT of improvement and when it does it wil be better as the sensor simply is better.
if not i want this sensor in a dslr as its mucgh sharper with the same lenses
somthing is just off here either canon firmware or video editing of the 5d as it simply doesnt match the theory.

[risc32]

I'm not a video guy, but i've been taking little clips of my kids in the backyard, and at their sports, and they look pretty darn good. I must be missing something, as the 5d's video quality in these videos looks pretty bad.  When shown in this fashion every little thing is noticeable(as should be done) but that seems to be a bridge to far. I mean, that 5d footage REALLY sucks compared to the BM.

[dhofmann]

When the 5D3 downscales the raw image data from 22MP to 2MP for video, I wonder if it's throwing away information that could be used to increase the dynamic range. Removing 91% of the pixels creates bigger "virtual pixels" and that ought to add at least a few stops of dynamic range.

So if the 5D3's processor is fast enough, maybe we'll see a new firmware that improves the dynamic range of video.

[lourenco]

The DR of the 5D Mark III does not seem that great. DXOMark shows it only DR of 11.7 at best. It just gets worse at the higher iso. I was considering of upgrading to the 5d Mark III. I am not sure now after seeing this video. I am now waiting for Canon to come out with DSLR camera with 1080p at 60 fps under $3k. 60fps with MagicLatern HDR video would allow someone to cheat a little bit to increase the DR. I am hoping that Canon is going to come out with something better next year.

If I was just into doing video I would consider the Blackmagic.  I currently need something to do both video and photography. 

[Policar]

The DR of the 5D Mark III does not seem that great. DXOMark shows it only DR of 11.7 at best. It just gets worse at the higher iso. I was considering of upgrading to the 5d Mark III. I am not sure now after seeing this video. I am now waiting for Canon to come out with DSLR camera with 1080p at 60 fps under $3k. 60fps with MagicLatern HDR video would allow someone to cheat a little bit to increase the DR. I am hoping that Canon is going to come out with something better next year.

If I was just into doing video I would consider the Blackmagic.  I currently need something to do both video and photography.

These lighting conditions were meant to stress the cameras, and they really don't get much more punishing, and the 5D would have done much better were highlight tone priority set to on although that would also make the dark areas noisier (if it was, then these tests reflect really poorly...), but it's clear that the BMC camera does very well. It's the softness that really gets to me, though. If Canon can fix that a bit I can surely live with the poor DR. The DR is also much better for stills, particularly when shooting raw.

I think the issue is that Canon's sensors have a ton of read noise and the technology to prevent it is proprietary to Sony sensors. It's introduced between the sensor and the ADC and it's not terrible in stills, but since the sensor bins on-sensor, in effect the read noise is 4X more present (or scaled up 4X larger) for video. And when using highlight tone priority it really gets noticeable, but that can help the highlight detail a lot.

[LetTheRightLensIn]

The little GH cams have crop modes, but can Canon give their DSLR crop mode video? Nope. Marketing would slap the engineers silly for daring to try to take anywhere near close to full advantage of the hardware. The 5D3 sensor actually has an ASP-C section that could be read exactly the same way as the C300 in 2x2 blocking since it is the same 8MP in the APS-C crop portion.

[LetTheRightLensIn]

Problem with both GH3 and Sony a99 is that to me they both still look kinda video-y. At least, despite the small sensor, the BMCC does have a nice cinematic quality to the footage that I've seen so far.

The GH2 was a touch on the video-looking side, true. While the 5D3 kind of soft, the GH2 is sort of almost a touch too video-crisp.

[NormanBates]

The DR of the 5D Mark III does not seem that great. DXOMark shows it only DR of 11.7 at best.

I love DxOMark tests, but they're only relevant for stills. When shooting video, you can't record RAW, you can only record debayered, processed, compressed H.264. The curve that Canon (and Nikon, and Sony) apply when debayering kills a lot of the DR captured by the sensor.

On the Canons, in video you actually get around 11 stops, if you don't mind running some NR to clean the shadows. For Nikons and Sony the difference is much bigger: the D800 has a DR advantage over the 5D3 of a bit less than one stop; the NEX-5N is on par with the Canons.

If the BlackMagic is actually recording 14 stops of DR, as they claim, that's nearly 3 stops more than the best DSLRs.

Check my measurements of the D800, T2i, GH2, NEX-5N, at the end of this:
http://www.vimeo.com/similaar/shootout2012

[lourenco]

Check my measurements of the D800, T2i, GH2, NEX-5N, at the end of this:
http://www.vimeo.com/similaar/shootout2012

It is interesting to see how bad the GH2 did. I was somewhat suprised that the D800 did not do any better. The 550D results look good.

The guys at MagicLatern are currently working on increasing the bit rate I believe on the 600d. They are currently limited to the max limit of the SD card of 20MB/s or 160mbits/s.  It would be very interesting to see what happens if they can apply that to the 5d Mark III where the CF card is much faster. It is not fast enough for raw though. 

[facedodge]

https://vimeo.com/38989476#

This test shows the 5D3 can reveal more detail with post sharpening. The video says he did color grading but didn't mention sharpening.