Japanese website mono-logue released this short 30-second video comparing footage from the Canon 5D Mark II and the new 5D Mark III captured at ISO 12,800. The difference in noise levels is remarkable (be sure to watch it full screen and in HD).
Here’s a closer look:
(via mono-logue via EOSHD)
Ooooooh yes. :-) Now that’s what we like to see.
Wow! There’s no comparison. The 5D III is much better than the old 5D II. Im sure HDSLR users will be happy. :)
Why does the 5D II seem to apear a lot sharper? Maybe he needs to calibrate his lens or learn to focus better! OOF would make the noise seem less I believe.
how is it different when used for still photography at low iso’s?
I also believe what Casey Myers says. Not a fair test. Why is the Mk II sharper? The Mark III almost looks as it has luminance noise reduction applied or is out of focus making it softer.
I had a go with the 5D III today. Not much to say, but my initial thoughts. http://www.audioguy.co.uk/6390 I’ll add more at a later date.
The 5D MKIII is not shown to be that impressive because the editor of this ran noise reduction filters on the footage and therefore degraded the overall DPI/resolution of the video. I’d like to see raw comparisons of footage in the same light, same time, and more movement.
High ISO performance was the deal maker for me on the 5D II, so I can’t wait to see how the 5D III does. But, I will most likely stay with the 5D II. Maybe get a Nokia 808 LOL.
Video on MKIII seems to be A LOT smoothed, just like high-noise pictures from my iPhone after going through Photoshop Express on 100% denoise. Not that I am complaining, I don’t even do video o my MKII, just saying…
Who cares, I just hope that MKII will decrease it’s price.
The level of detail in the background tree falls!
I’m only speculating, but I would say they just increased the noise reduction in the incamera filters. The tradeoff being not as sharp video. And no, OOF will be just as noisy as if it were in focus.
Anyone know anything about the auto-focusing in the MK III video?
but at web size (where most of my stuff is) the 5diii is much better. However: I might still buy a d800
What we really need to understand is what is the sensitivity of the sensor. All you’re looking at is the tricks played by the post-processing chip on board. If the 5D M2 was 800 ISO, and the 5D M3 is 160ISO then you have one more stop of sensitivity and that’s it, nothing worth switching to the M3… for $1.5K more. I can give the exact same image with the M2 after a good denoising. Takes 5 seconds.
What we really need in the M3, was a real sound capture input (some kind of XLR), sound controls on the body of the camera, and a true downsampling of the picture to 1080p not, just a selection of every three lines of pixels. That sucks, and brings the quality down. Canon didn’t put a real effort to release a new camera, but just a firmware update, with a slightly faster chip. That’s too bad.
Yeah, it looks like the smeary noise reduction you see on jpeg output. Still, it’s not bad.
The reason the Mk. III looks “blurrier” is because of the much less aliasing. That is a great thing. Adding sharpness in post is easy. The bigger deal between the two is the noise/blocking of the Mk. II. It’s really night and day.
Yes we should see RAW image first to evaluate the sensor, is it goes from DXO mark low light 1850 to 2000 and it has just much better NR on board, there is no point. But if the sensor can handle iso 2500 with low SNR it rocks. 5d2 has very high dark area or readout noise, hopefully they have fixed at least this, because they claim sensor no board D/A circuity like we have seen three years in Sony/Nikon sensors. This on board sensor from the other hand causes low iso highlight area noise in Sony design at least, 5DII looks very clean at iso 100-200 studio light.
Ok for the noise, but the video shows too that the details are sacrified for a smoothing process.
Less aliasing? that thing doesn´t exist. Sharp or not sharp –> more or less washing textures, more or less denoise. Don´t be freaked, there is nothing great with that.
Adding sharpness is never easy because it ALWAYS degrades quality. You can’t magically add detail where there is none. You can accentuate contrasts but that’s always artificial boosting and can not replace sharp source material.
Zacuto did a series of comparisons of various hd-dslr’s along with digital cinema and film cameras last year and went to great lengths get as close to a side by side equal comparison as you can.
It was really well done and I hope to see them do it again with the new Canon models and the Nikon models and possible some of the mirror-less models from Sony and Olympus.
otherwise even if the 5D Mark III is processed different in the above footage low-light shooting still seems to be a order of magnitude better and brings Canon to parity with Nikon for iso capabilities so if you have a Canon lens setup already it is all win win I’d say.
So basically, the 5DIII does in-camera denoising whereas the 5DII leaves it up to the user to do in post. I’m impressed with the processing power of the camera, not impressed with this solution.
canon has to do much more propaganda to convince the average joe, who would be interested to dish out an extra 1500$ on the MKIII rather than the MKII. in my amateur opinion 5dii is still much more of a bargain.
You can use selective noise reduction in post processing for 5D Mark II.
Maybe the Mk III video is really from the D800 LOL.
I was just reviewing the video and I’d say the test is a fail. The Mk II looks better to me, even with the noise. I would be more convinced that the test is flawed rather than the Mk III is not up to par.
I agree, lets evaluate it on a RAW image rather than a flipping youtube clip!!
Yes and no. With video you can add some perceived sharpness (unsharp mask), but you can’t take it away once it’s already there. It’s why many turn down the in-camera sharpening on the current crop of Canon cameras. Too much in-camera sharpening can take away from the filmic look.
Hi peeps, here’s some HOT INFO for you – the 5DIII looks blurred because the camera was designed to stop Moire and Aliasing. The only reason 5DII looks sharper is that is uses aliasing to create sharper edges which creates the illusion of being sharper. The 5DIII simply has this feature removed. Secondly the 5DII actually only captures at 720p then upscales it to 1080p. The output files say ’1080′ and they look crisp, but it’s not really true 1080 and it’s not really that sharp naturally. The 5DIII is better allround, as it is recording closer to real 1080 and does not have moire and aliasing problems like the 5D & 7D… just add some sharpening in post.
Agree, the 5DII looks better to me.