Canon appears to have confirmed IBIS is on the roadmap

Grimbald

I'm New Here
Sep 30, 2017
17
6
Is that because you are concerned about increased noise at low light or file size?
For both of those reason. A "5dsr like" camera will be great for landscapes because of its resolution and the possibilty to crop in, but many landscape shooters like to take milkyway and aurora pictures as well. And most likely, you just won't carry 2 bodies up a mountain, so Canon hopefully comes up with a good balanced solution of not too many MP just because they can and it looks good on paper. They way I know Canon, they still will make some things imperfect with this body, and my bet would be FPS, beacuse a) it's hard to do that with a file size that large and b) they want a clear distinguishable difference between their 1dx line (which I assume to come sometimes sometimes later as a mirrorless version).
 

tpatana

EOS 6D MK II
Nov 1, 2012
1,223
14
I don't care too much about IBIS although it'd be nice. More important features to replace my bodies:

5D4:
-40Mpix+
-Awesome DR at low ISO
-Good generic all-around, wifi etc.
-Dual-slot would be nice but not mandatory
-optional grip
-$3500 or less

1DX:
-20-30Mpix
-DPAF tracking while shooting moving indoor sports in fairly dim gyms, close or better compared to 1DX
-deep buffer
-dual slot mandatory
-all fancy stuff like wifi etc
-1DX style body, i.e. gripped
-$5500 or less

Both of them should have AF point spot metering too. I don't get why Canon has to be hard-a$$ about keeping that for 1D-line only.
 

BurningPlatform

EOS T7i
Mar 4, 2014
63
13
Gosh, I hope not. Anything you can do by physically moving the sensor you can do by digital manipulation, but with fewer moving parts, more reliably, smaller, ultimately cheaper. If the current digital stabilization isn't quite there, fine, do something physical, but I'd much rather see a full-digital solution. The R actually does digital stabilization in movies. I don't know what's preventing them from doing that in stills as well.
Quite substantial development in digital stabilization would be needed, though. The current digital IS systems stabilize movement only between subsequent video frames. You just can not digitally eliminate motion blur during a single exposure. You just can't. The only way to do this would be to divide the exposure to multiple sub-exposures, and to those sub-images use automatic alignment, perspective correction and stacking. Quite possibly possible in a few years, as that is what some mobile phones already do with long exposures.

And with longer focal lengths the movement of the image circle on the sensor plane (hand held) is too large to be compensated digitally anyway (same with IBIS, long lenses will always have lens IS)..
 

3kramd5

EOS 5D MK IV
Mar 2, 2012
2,801
239
Quite substantial development in digital stabilization would be needed, though. The current digital IS systems stabilize movement only between subsequent video frames. You just can not digitally eliminate motion blur during a single exposure. You just can't. The only way to do this would be to divide the exposure to multiple sub-exposures, and to those sub-images use automatic alignment, perspective correction and stacking.
That’s not the only way. You could run deconvolve algorithms. However they’d be no better, and probably quite a bit worse, than if done on a dedicated computer.

But I agree, the ability to keep a subject roughly in the same part of the frame from one shot to the next is completely different than stabilizing an individual frame. The digital IS canon uses won’t reduce blue induced by camera shake in a given frame.
 

SwissFrank

EOS 80D
Dec 9, 2018
106
47
But I agree, the ability to keep a subject roughly in the same part of the frame from one shot to the next is completely different than stabilizing an individual frame.
Canon has a patent for a sensor that can be read 1000 times a second. You could certainly compose a single frame of 1/500, 1/333, 1/250, 1/200 etc. out of a series of such 2, 3, 4, 5, etc. individual frames, and subsequently stacked, with the objects at the focus points especially aligned. (Such a method could even in effect stitch panoramas: it'd be so camera-shake-proof that you could literally turn 360 degrees in a four-second exposure and get 360 panorama with any lens.)

While a desktop PC would have much more power than an in-camera PC-on-a-chip, the camera also is going to have custom circuitry that could be orders of magnitude faster in this operation.

Granted, this method would look stroboscopic if some parts are moving far more than 1 pixel per subframe, but 1) some might like that effect, and 2) ALL the digital processing the camera already has similar downsides and can be turned off at will. For instance fixing vignetting may cause noise. Fixing distortion and chromatic aberration may blur pixels up to half-a-pixel, and so on. If you don't like the look of "making a still from a digitally-stabilized movie," just turn it off.