Can the IBIS in the Canon R5 be used to create stereo3D or higher resolution

[Michael Hastings]

Wondering what is maximum horizontal movement of the sensor with the IBIS in the Canon R5. Could that be be used to create stereo3D by doubling the frame rate of capture of the desired end frame rate and bouncing the sensor from one horizontal limit to the other between frames - i.e. 60 frames to get Stereo3D at 30 frames per second or would the slight time difference be too much visually?

Or could higher resolution be obtained by shifting a single pixel width between frames and interpolating I.e. "temporal pixel offset"?

[Zack Birlew]

I've had 3D on the brain recently too! Sensor movement would be interesting but I wouldn't think IBIS would be able to move fast enough or far enough to create a full fledged 3D image. Much as how there were hesitations in the initial iteration of HDRx, the speed of the IBIS may not be enough for motion work but if it could work, it would most likely be for stills only unless some sort of motion interpolation could be done in software. Of course, I don't know the intricate technical aspects of the R5, I'm just guessing until somebody who might know chimes in. As for the higher resolution aspect, I've only seen that done on Sony cameras, certain digital Hasselblads, and some cell phones to get things like "300 megapixel photos" which is cool but most likely not the same as a real multi-hundred megapixel photo. Nonetheless, if Canon or the Magic Lantern guys could get the R5 to do it, that would make for much larger photographs at the least!

[Michael Hastings]

I've had 3D on the brain recently too! Sensor movement would be interesting but I wouldn't think IBIS would be able to move fast enough or far enough to create a full fledged 3D image. Much as how there were hesitations in the initial iteration of HDRx, the speed of the IBIS may not be enough for motion work but if it could work, it would most likely be for stills only unless some sort of motion interpolation could be done in software. Of course, I don't know the intricate technical aspects of the R5, I'm just guessing until somebody who might know chimes in. As for the higher resolution aspect, I've only seen that done on Sony cameras, certain digital Hasselblads, and some cell phones to get things like "300 megapixel photos" which is cool but most likely not the same as a real multi-hundred megapixel photo. Nonetheless, if Canon or the Magic Lantern guys could get the R5 to do it, that would make for much larger photographs at the least!

One of the reasons I actually like the hydrogen camera is that the sensors are so close together (about 13mm) which makes for better closeup shots than most other 3D cameras with much wider interaxial. It looks like from some of the videos online that other ibis sensors can move about 3 to 3.5mm in either direction so potentially a 6-7mm IA. Using the old 30x rule would make about 7" minimum distance for good 3d separation and obviously the 3d effect would be limited for distances more than a couple feet.

It does seem like the idea of increased resolution would be pretty easy to implement - especially for locked down landscape or similar subjects. The sensor would only need to move a couple microns and it seems like the method could be used with 2 frames with a single pixel shift or multiple frames with a single pixel shift for each frame. Interpolation could be done in camera or in post software. Seems like the current level of processor speed could probably handle this internally possibly with a small lag.

[Jacek Zakowicz]

I'm pretty sure you got it backwards- the images for right and left eyes have to be at a different angle onto the same frame. By moving the sensor you only look at different parts of the same image. In the conventional stereo that would not do anything...

[Karim D. Ghantous]

I'm pretty sure you got it backwards- the images for right and left eyes have to be at a different angle onto the same frame. By moving the sensor you only look at different parts of the same image. In the conventional stereo that would not do anything...

That's true. Moving the sensor is not changing perspective (i.e. moving the optical axis to a new point). You could argue that using OIS + IBIS can produce a stereo image, though. Although I wouldn't claim that without seeing evidence.

[Michael Hastings]

I'm pretty sure you got it backwards- the images for right and left eyes have to be at a different angle onto the same frame. By moving the sensor you only look at different parts of the same image. In the conventional stereo that would not do anything...

I see, good thinking, you're saying to create the 3d you would need to move the lens with the sensor.

What about the single pixel move for greater resolution?