Been saying it for months...

[shadowblade]

1Dx2 vs D5 vs A7r2 sensor

Been saying it for months - now the measurements are out. The 80D-style on-sensor ADC has made a huge difference in the 1Dx2.

1Dx2 vs D5 sensor DR:
- 1 1/3 stops ahead at base ISO
- Stays a long way ahead up to and including ISO 800
- More-or-less equal at ISO 1600 and 3200
- D5 ahead by around half a stop from ISO 6400-51200
- Equal after that

vs A7r2 sensor:
- A7r2 slightly ahead at ISO 100-400 (and almost two stops better than D5), then more or less follows the 1Dx2 above that, except it jumps up earlier at high ISO to beat the D5 again from ISO 51200 onwards

With regards to image quality, Canon seems to be back in the game.

[Paul2660]

To me, it's  really not fair to compare the 1dx MKII or D5 to the Sony A7RII, 42MP, vs 21? I would totally expect the less dense chips of both cameras to have a better high ISO.  Each camera has it's own area that it will excel in.

DxOMark, on this site, lovers and haters, the Canon and Nikon are tied at 88.  If you feel that ISO 51000 is going to work looks like the Canon will be the winner!!.  Personally, I don't think the average photographer will be able to see much difference between the two as both have been very well tuned.

Can't wait for Dpreview to get their full review up, so the side by side comparisons they offer can show. 

I am sure both will turn out to be excellent cameras, neither are on my list, just don't need the frame rate. 

What interests me more, will be if Canon takes this tech into the 5D MKIV or 6D MKII, that would be a big positive for sure. 

Not sure why it always has to be such a battle, but it always works out that way.

Paul C

[Zorki5]

What interests me more, will be if Canon takes this tech into the 5D MKIV or 6D MKII

Or into a decent mirrorless (not the current Mx jokes) -- then I will have a look.

[shadowblade]

To me, it's  really not fair to compare the 1dx MKII or D5 to the Sony A7RII, 42MP, vs 21? I would totally expect the less dense chips of both cameras to have a better high ISO.  Each camera has it's own area that it will excel in.

Why? BSI technology takes the electronics out of the light-facing area, so the total illuminated area is the same regardless of density, The A7r2 actually has a better SNR than the 12MP A7s2.

What interests me more, will be if Canon takes this tech into the 5D MKIV or 6D MKII, that would be a big positive for sure.

Especially the 5Ds2. High-resolution, low-speed studio/landscape cameras practically live at low ISO. Although the 5Ds does a pretty good job at wildlife, too, due to its pixel density combined with a good AF system.

I'm at a loss as to why they even bothered introducing the 5Ds last year, prior to the introduction to the new technology. It would have done a lot better than it has had it addressed one of the major considerations of those who actually use high-resolution bodies.

Not sure why it always has to be such a battle, but it always works out that way.

Some people have been saying for months that the D5 sensor would beat the 1Dx2 sensor at both high and low ISOs, despite data showing a marked plateau at low ISO that hasn't been present in any other recent Nikon or Sony sensor. Meanwhile, Canon had showed off their on-sensor ADC with the 80D and given every indication it would be used in the 1Dx2 too. It just wouldn't have made sense for the new Canon sensor to underperform the Nikon at low ISO (high ISOs being a tie).

[Paul2660]

All points well taken.  Good one on the Back illumination on the Sony, as I for one was not impressed with the A7rII, mainly due to:

1.  Heat
2.  really terrible long exposures
3.  Not so great higher ISOs

I also agree that if the next Canon 50MP comes out with this on chip ACD, then a lot of folks will be looking for sure.  But for me the real game changer was Pentax and K1.  Really an impressive landscape camera with pixel shift (it's not a gimmick it really works very well).

The real question to me is when both Canon and Nikon move to mirrorless and how they do it.  I realize that with the vast number of lenses out there for their DSLR's it won't be as easy as what Hasselblad just did. But it has to happen sometime.

Paul C

[shadowblade]

All points well taken.  Good one on the Back illumination on the Sony, as I for one was not impressed with the A7rII, mainly due to:

1.  Heat
2.  really terrible long exposures

Both of those really boil down to heat - I guess a larger heat sink would have been nice. The A9 may do better, simply by virtue of being larger.

Still, I had no problems with some recent 15-20 minute shots.

Are you running with the updated firmware for lossless RAW?

3.  Not so great higher ISOs

It's still no worse than any other full-frame camera, including the D5 (which specialises in high ISO).

I also agree that if the next Canon 50MP comes out with this on chip ACD, then a lot of folks will be looking for sure.  But for me the real game changer was Pentax and K1.  Really an impressive landscape camera with pixel shift (it's not a gimmick it really works very well).

The problem is lens selection, or lack thereof. It would be much more impressive on a Sony E-mount. They already have the mechanics for it (also used for IBIS) - just not the firmware control.

The real question to me is when both Canon and Nikon move to mirrorless and how they do it.  I realize that with the vast number of lenses out there for their DSLR's it won't be as easy as what Hasselblad just did. But it has to happen sometime.

The A9 may provide some clues.

The issue isn't with the lenses - they're fine as they are, and mirrorless bodies use PDAF for the high-speed part of their AF anyway, with CDAF and AI-based methods (e.g. eye detection) being secondary. It also isn't an issue with AF speed and accuracy - given the same processor power, mirrorless PDAF is just as fast as SLR PDAF.

The issue is viewfinder lag for action photography. It's not insurmountable, and there's no reason it can't be reduced to an imperceptible few milliseconds. But this takes battery and processor power. Hopefully, the weight saved by not having a mirror mechanism and prism box can go towards a much more powerful battery to make this happen.

[Paulo Bizarro]

The issue is viewfinder lag for action photography. It's not insurmountable, and there's no reason it can't be reduced to an imperceptible few milliseconds. But this takes battery and processor power. Hopefully, the weight saved by not having a mirror mechanism and prism box can go towards a much more powerful battery to make this happen.

I have been wondering about this obstacle regarding the use of EVF for action photography (say sports in general). Don't you think that in the near future, and given the progress in video for some cameras, it will became feasible just to shoot video and extract the important frames from it? Given the not so high requirements from news agencies and print media, I think that it would be a feasible option.

[shadowblade]

I have been wondering about this obstacle regarding the use of EVF for action photography (say sports in general). Don't you think that in the near future, and given the progress in video for some cameras, it will became feasible just to shoot video and extract the important frames from it? Given the not so high requirements from news agencies and print media, I think that it would be a feasible option.

Yes... eventually. To shoot high-quality video and action stills simultaneously, you'd need to shoot at 1000-2000fps and 32-40MP resolution, with a buffer of at least 40 frames. That's probably a decade away. Yes, 1000fps sounds huge at the moment, but, in terms of data bandwidth, it's only 40 times more than a 25fps 8k video camera (some of which are coming out now), and, using fully electronic shutters, moving parts aren't an issue.

Video requires motion-blurred frames (shot at around 1/40-1/50 for 25fps video) to appear smooth, while action shots need fast shutter speeds to stop action (typically 1/1000 or faster). A video made from still photos will look choppy, while stills taken from a video clip will be blurry. To get around this, you'd need to shoot at the shutter speed needed for stills (1/1000 or so), then add 40 consecutive frames together to create a motion-blurred frame for 25fps video (or 20 frames for 50fps video, and so on). Or just run the video at 1000fps, as the case may be...

Of course, we'll be able to shoot 8k video and action stills separately well before then - for that, you just need an 8k video camera with a fast AF module that can focus and track while shooting. Basically, a 32-40MP mirrorless camera. Set it to 1/40 or so to shoot video, or 1/1000 or faster when you want to shoot action stills. Not too far off. So, pretty soon, action cameras and video cameras will be the same thing.

Even with simultaneous video/stills capability, you'd still want lagfree EVFs and fast AF while shooting - they're basically separate issues. Although improving pattern-recognition-based AF methods (e.g. face detection, eye detection and now Nikon's 3D tracking) will likely make AF a non-issue by that stage, particularly with a mirrorless sensor that makes use of the full resolution of the sensor for AF, rather than the low-resolution AF sensor currently used on SLRs.

[BernardLanguillier]

This is really good news, it means that the successor of the 5Ds will finally be back on my list of credible candidates!

Cheers,
Bernard

[shadowblade]

This is really good news, it means that the successor of the 5Ds will finally be back on my list of credible candidates!

Cheers,
Bernard

I'll reserve judgment until the A9/A9r (or whatever Sony wants to call it) comes out. And it actually depends more on the AF than the sensor.

The GM-series lenses are just incredible, but unusable on anything other than a Sony. For a pure landscape kit, a few GM zooms on a compact Sony body will be pretty hard to beat, either for single shots or for stitching.

But, for wildlife and action photography, Canon long lenses on a fast-focusing, high-pixel-density, full-frame sensor would be ideal. Or Nikon lenses on a D810 successor. The current 5Ds already works pretty well for this (provided you're willing to sacrifice frame rate for resolution; if you need frame rate, the 1Dx works better). Whether Sony can match their AF capability with their next-generation, full-featured mirrorless body is still in question. No doubt they'll get there at some stage, but we don't know when.

Of course, the next generation of Canon/Nikon zooms (and primes) will be very interesting. The current ones (I'm talking about the core 14-200mm range) are sharp - the zooms often as sharp as the primes - but not 100MP sharp like Sony's latest offerings. If they're going to bring out sensors in the 70-80MP or higher range, they'll need lenses to take advantage of it.

[Paulo Bizarro]

Yes... eventually. To shoot high-quality video and action stills simultaneously,

I was not thinking about it in terms of being simultaneous; rather, just shoot video and after extract still frames from it to distribute to news agencies?

[shadowblade]

I was not thinking about it in terms of being simultaneous; rather, just shoot video and after extract still frames from it to distribute to news agencies?

It'll get there eventually. First separate shooting, then simultaneous, as available bandwidth increases (eventually, it may be separate at 16k, simultaneous at 8k or something like that).

It wasn't so long ago that 720p digital video was a big enough deal. 8k video at the same frame rate requires 33 times as much bandwidth. That's all happened in just over a decade.