BIG NEWS - Hands on with CONTINUOUS raw recording on Canon 5D Mark III

[Danyyyel]

People that are contesting Dxomark today are trully baseless. It has proven during the last few years that it was the most reliable measurement for sensor performance in the dslr photography world. It total sensor score is somewhat subjective but the rest is scientific and has the methodology. For example it test at 8 megapixel, so for example the D800 is 14.5 stop at 8 megapixel but about 13.9 at pixel level. 

 

Most of the time, the people that are contesting it are Canon fanboys because it is hard for them to see that Canon sensor tech has not evolved a lot the last 4 years. In any case I am not saying that the 5d3 raw hack is not incredible. At 11.7 to 12 stop of DR (perhaps a little lower at pixel level) it is still very good and much more usable than in a compressed 8 bit format. Another of note is that I did not see any real scientific test of the DR of the BMC, so for now I take the 13 stop with a little grain of salt.

[Julian]

I did a dynamic range test with the D800 and 5D III in raw (photo mode), by lifting the shadows from an underexposed image.

http://www.digifotopro.nl/content/canon-5d-mark-iii-vs-nikon-d800-dynamisch-bereik

 

The difference is definitely there. You can lift the D800 4 stops without much penalty. That is the extra dynamic range of the sensor paying off.

 

 

 

Also in the highlights.

 

 

Not measured in numbers, but you can see the difference. In line with the DXO results.

[Andrew Reid]

Re-uploaded, totally new video

 

 

Some wide shots in there as well (24mm)

 

I'm blown away by this thing.

[Andrew Reid]

Around 2:10 the grass detail looks very strange on the top half, I'm wondering which kind of interpolation is computed.

 

You are looking at H.264 compression. It is good but not perfect. Maybe I should upload a 2GB ProRes file for people to truly get an idea of the quality.

[jgharding]

So it's basically safety: if you get it really wrong with an even higher dynamic range sensor and have to push four stops, the Nikon is better.

 

I suppose it's best not to get it four stops wrong!

[ScreensPro]

I did a dynamic range test with the D800 and 5D III in raw (photo mode), by lifting the shadows from an underexposed image.

http://www.digifotopro.nl/content/canon-5d-mark-iii-vs-nikon-d800-dynamisch-bereik

 

The properly exposed shot only has a few stops of DR to deal with, so you are just testing the very extremes of what you can recover from each sensor at the non native ISO.

 

Like i've said before though.... anyone who has ever shot a still on a Canon know's the raw image is superb.... We are past the point of looking at numbers on a graph.

 

Every hurdle that gets knocked down, another is put straight up..... Do people really think that 12 stops is not enough? Do most people ever know when their scene has more than 12 stops?

 

If you can't get a good image from a Canon raw file, you have far more worries than 3 stops of extra DR.

[ScreensPro]

Loving the extended video... great colour and detail. Certainly up there with the BMCC and RED

 

Keep up the good work

[Danyyyel]

This was on my 5DII.... if that isn't enough DR, i don't know what is. (That is very bright cloud on the way up a mountain, it would ruin the shadows in h264... The tree stumps would be almost black).

 

 

I am sorry if I sound rude but d you know what Dynamic range is. To show such a gloomy low contarst scene to show dynamic range is really arkward .

[Danyyyel]

Nobody I think is saying that about 12 stop (11.7 by DXO) of DR is not good. But having now a D800 and compared to the 12 stop DR D300 before it, I can say that it reaches the treshold of near human vision. That is why I think the Alexa is so much hailed for its DR, because this is about really near to what I can see. (I know the different theories about the human eye DR). There is by DXOmark nearly 3 stop of DR difference and I will believe it beyond any people guess, experience or whatever. Again that does not mean that the Canon is not able to do supperb image.

 

I sincerely hope that we see a DXOmark type of rating in the video world soon. What Dxomark has done is to pacify the dslr world. With it scientific method it standardise sensor performance ranking and except certain Fanboys everybody now has a standard to rate what is in the camera. The result is that the endless my camera is better than yours at least in image quality has subsided and people have gone more into the creative talking.

[ScreensPro]

 

I am sorry if I sound rude but d you know what Dynamic range is. To show such a gloomy low contarst scene to show dynamic range is really arkward .

 

I could ask you the same question..... That bright cloud light (you can even see the sun trying to burn through) meant i had to shoot the scene very dark, or else all the top detail would be completely lost (you can see it starting to go, even after protecting the highlights).... and the forest floor was really not light at all, yet i managed to pull out plenty of detail....

 

The fact that you think that shot didn't require much DR is, partly,  a testament to the camera.

[Andrew Reid]

One request would be: APSC/S35 crop mode. I have some old cinema glass that wasn't designed for 135 frame. Is this possible?

 

You can enter any resolution in the Magic Lantern raw module code. But I have not tried larger than 3.5K. The card isn't fast enough at the moment. 2560 x 720 is my limit. That is smaller than Super 35.

 

If we can get a sensor window that large, who knows what will be possible in the future with this hack!

[Edward Zaee]

The solution for higher resolutions <3K will be a CF - SSD interface. I won't be surprised if we see something like that in the near future.

[Andrew Reid]

The 50 pin CF 1.8" SSDs I've seen are all slow for SSDs... 80MB/s or so. Anyone found faster?

[Germy1979]

The 50 pin CF 1.8" SSDs I've seen are all slow for SSDs... 80MB/s or so. Anyone found faster?

The old Red CF cards were "turbo-fied" by Lexar specifically for the Red cams weren't they? I've never used one though so I'm not sure whether you can use one in another camera.

[Edward Zaee]

Did ML say this would reach the 7d? would be great if it did...but the 7d usually gets left behind in such things.

7D version is already under development. Just wait a little longer  ;)

[Edward Zaee]

The old Red CF cards were "turbo-fied" by Lexar specifically for the Red cams weren't they? I've never used one though so I'm not sure whether you can use one in another camera.

 

Are you talking about this? http://www.red.com/store/products/red-station-compact-flash

[Andrew Reid]

7D version is already under development. Just wait a little longer  ;)

 

Are you sure? Hmm. I've not seen any evidence they are working on raw with the 7D.

 

6D, 600D, yes.

 

BTW the Red CF cards are interesting but unlikely to be faster than 1000x.

 

Red Raw to those CFs was compressed. The 5D is doing uncompressed raw.

[Edward Zaee]

Are you sure? Hmm. I've not seen any evidence they are working on raw with the 7D.

 

6D, 600D, yes.

 

BTW the Red CF cards are interesting but unlikely to be faster than 1000x.

 

Red Raw to those CFs was compressed. The 5D is doing uncompressed raw.

 

g3gg0 from the ML forums: 7D: working on it, but it seems the interesting part is being handled on master digic that is harder to access (only via FIR)

 

http://www.magiclantern.fm/forum/index.php?topic=5405.msg34778#msg34778

[Andrew Reid]

Ah good news, that is a new development. 7D would be interesting with those faster processors.

[HurtinMinorKey]

This a huge upgrade for the 5D3. Still not seeing anything on the Alexa level, or anything close.  Point it at some faces! we need to see skin tones!