Sony a7RIII Pixel Shift sample

Have questions about the equipment used for macro- or micro- photography? Post those questions in this forum.

Moderators: rjlittlefield, ChrisR, Chris S., Pau

RobertOToole
Posts: 2627
Joined: Thu Jan 17, 2013 9:34 pm
Location: United States
Contact:

Sony a7RIII Pixel Shift sample

Post by RobertOToole »

Not sure if this hasn't been posted already?

Sony posted a full size sample of an image using the Sony pixel shift mode in the a7RIII. Nice detail.

http://support.d-imaging.sony.co.jp/app ... 2x5304.jpg

Image


This new feature takes advantage of the advanced in-body image stabilization system control capability available in the CX62000, precisely shifting the sensor in 1-pixel increments to capture four pixel-shifted images containing a total of approximately 169.6 million pixels* that are later composited using a computer to achieve overwhelming subjective resolution in a single image.

This is not the first body to have this tech, Olympus and Pentax has featured it for time. Not sure of the differences but Sony uses a 1 pixel shift, Olympus shifts 4 pixels I believe.

Sony also posted a video on the technology with samples.

https://www.youtube.com/watch?v=6-811lVi0Qc

Anyone here have the a7RIII?

Robert

rjlittlefield
Site Admin
Posts: 23564
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

Sony's description strikes me as shameless hype. With full pixel shift, the image they end up acquiring has no more resolution than a single shot does. The advantage is simply better color resolution, with true RGB at every pixel instead of Bayer interpolation. The only sense in which this is "approximately 169.6 million pixels" is because that's roughly the number you get if you multiply 7952x5304 pixels by 4 exposures.

The sample image is even labeled as 7952x5304, and indeed that's what it comes down as.

This is very different from the true resolution increases that some Olympus cameras get from their sub-pixel shifts.

Have I missed something here?

--Rik

RobertOToole
Posts: 2627
Joined: Thu Jan 17, 2013 9:34 pm
Location: United States
Contact:

Post by RobertOToole »

rjlittlefield wrote:Sony's description strikes me as shameless hype. With full pixel shift, the image they end up acquiring has no more resolution than a single shot does. The advantage is simply better color resolution, with true RGB at every pixel instead of Bayer interpolation. The only sense in which this is "approximately 169.6 million pixels" is because that's roughly the number you get if you multiply 7952x5304 pixels by 4 exposures.

The sample image is even labeled as 7952x5304, and indeed that's what it comes down as.

This is very different from the true resolution increases that some Olympus cameras get from their sub-pixel shifts.

Have I missed something here?

--Rik
Agreed its hype!

As far as I know it eliminates all the bayer issues and increases detail without moire and jagged edges. Sony did a poor job illustrating the differences.

DPreview did a better job with real world comparisons here, the details really much better:

https://www.dpreview.com/news/564575561 ... landscapes


I was really hoping someone on the forum with a a7RIII had tried it. Or better someone with an Olympus and a Sony :-)

Robert

Lou Jost
Posts: 5948
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

The Olympus cameras shift by 1/2 pixel (and makes 8 sub-images per high-res image), and sub-pixel shifting is the only thing that can add sub-pixel resolution.

Shifting by an integer number of pixels does increase resolution, though, because a Bayer filter on a 7952 x5304 pixel sensor reduces the nominal resolution by up to 75% (depending on the subject--worst case is a pure blue or red subject). Sony's shifting algorithm brings the sensor resolution up to its nominal value.

Edit- I see Robert made the same point while I was typing.

Lou Jost
Posts: 5948
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

I don't think the PEN F sample in the DPR tests is representative. I wonder what lens they used?

rjlittlefield
Site Admin
Posts: 23564
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

RobertOToole wrote:Sony did a poor job illustrating the differences.

DPreview did a better job with real world comparisons here, the details really much better:

https://www.dpreview.com/news/564575561 ... landscapes
Agreed, the DPreview article makes an excellent case.

I'm glad to see they mentioned the problem with movement in the scene. Even in their starting illustration, the DPreview image shows some interesting block artifacts in the sky near the building tops. I presume this was caused by moving clouds.

--Rik

Lou Jost
Posts: 5948
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

Relating this to a theme on another thread, for a given size sensor, one could expect the same degree of improvement by using a monochrome sensor and RGB filters, like astronomic photographers do, instead of a Bayered sensor of the same size.

Post Reply Previous topicNext topic