Sony A7Riii Pixel Shift Examples

Sony a7R III Pixel Shift lifts a veil off your landscapes

DPReview published a test report on pixel-shift results for the new Sony A7Rii. It's been slow on this forum, so I thought it was time to post something new.

The samples speak for themselves. On the keynote image, a city landscape, the resolution appears to be (subjectively) double that of a conventional image from that camera and lens. The effect is most pronounced in saturated colors, which are subject to artifacts of interpolation of the Bayer filter array. There is some evidence of fringing, if not CA, near the corners. This may be an artifact of parallax between the sensor and the Bayer filter. Then too, the extraordinary resolution might be revealing defects in the lens itself.

Pixel shifting merges four image, spaced 1 second apart, moving the sensor 1 pixel at a time, to cover the four pixel array, RGGB, which repeats throughout the filter. Because of the delay, it is susceptible to ghosting when something in the subject moves. That's nothing new for those of us who make bracketed HDRs, stitched panoramas, or ordinary photos on a windy day. Besides choice of a subject and timing, there are software tools which reduce ghosting.

Do you need the equivalent of nearly 170 MP to make a good photo? Perhaps not. I've seen very good 16x24" prints made with a 12 MP camera. The real difference is the sense of texture you get from details too small to see at a normal viewing distance. There is also some sense of satisfaction in seeing whether blinds are drawn in a building a mile away, when the building is barely visible at normal magnification.

I've pre-ordered an a7R III, but I'm keeping my Canon 5DS-R for those few times that I need that extra layer of clarity. I'm more interested in its high DR, 10-fps and AF system, almost as good as the a9. For nature shooting, the a9 will be my prime body and I'll have a 700mm super-telephoto rig on the a7R III. For travel, I think the a7R III will be the lead body.

I'm on the fence about walkabout lenses. The Canon's work as well as they do on the Canon bodies, but eye-detect and face-detect can be addictive when shooting people. I may need to do some comparisons between the 5DS-R and the a7R III to make final decisions.

I've made 72"x36" prints of my 5DS-R images and they'll stand up to photographers putting their eyes right to the image. ("Normal" viewers NEVER do that).

Thanks, Ed. That was quite revealing. I don't doubt that the standard image would have a bit more detail with sharpening, but it would never have more than about 80% of the resolution of the pixel-shifted image. The advantages are obviously more than just for resolution.

The defects in the corners are likely to do with the lens - the Bayer filter is fixed to the sensor. Notice that the A7r II has better corner performance than the A7r III when the latter is shooting normally. I'd say that the cover glass is a bit different.

I've never been impressed by Canon digital cameras. The 5DSR does offer a genuine resolution increase but it's not superior to the A7r series, including the original. Then again there is little difference between 36Mpx and 50Mpx. But I'm not picking on Canon - all DSLRs are redundant.

...

I've never been impressed by Canon digital cameras. The 5DSR does offer a genuine resolution increase but it's not superior to the A7r series, including the original. Then again there is little difference between 36Mpx and 50Mpx. But I'm not picking on Canon - all DSLRs are redundant.

I'm not sure how you know that, but I'll see for myself.