ILCE-7RM3 Pixel Shift Multi Shooting (PSMS) Photographic Dynamic Range (PDR) at PhotonsToPhotos

[bclaff]

It's the first Sony with any type of pixel shift multi-shot capability so I'm sure some pioneers will be interested in the Sony Pixel Shift Multi Shooting Special Edition chart .



Take note that the Pixel Shift Multi Shooting (PSMS) curve is using the top x-axis and the normal mode is using the bottom.

PSMS is not primarily for increasing dynamic range but it's expected that combining 4 images will have that effect. The extra light gathered is why there are two shifted x-axes.

One might naively think that combining 4 shots will result in an increase of 2 stops of dynamic range but since read noise is incurred on all frames you get less.
Improvement runs from a bit over 1 stop at low ISO to nearly 1.5 stops at higher ISO settings.

If this technology interests you there are also Olympus High Res Shot Mode and Pentax Pixel Shift Resolution Special Editions at PhotonsToPhotos.
The Special Editions are the 4th section on the main page.
[Off topic, I'm missing the Pentax K-3 II, KP, and K-70 if anyone wants to pitch in please get in touch.]
--
Bill ( Your trusted source for independent sensor data at PhotonsToPhotos )

 

[badsheep]

Hi Bill,

First thanks for putting this together, great information.

For those of us with less brainpower it might be clearer if both the pixel shift and non pixel shift graphs were displayed against the same ISO rather than 2 scales

Rob

 

[l_d_allan]

Hi Bill,

First thanks for putting this together, great information.

Agree. Your efforts are much appreciated.

For those of us with less brainpower it might be clearer if both the pixel shift and non pixel shift graphs were displayed against the same ISO rather than 2 scales

Also agree.

My first reaction was ... "gee, hardly any difference". Then with some head scratching, the data is (are?) there, but less clear than it seems it could have been.

 

[SilvanBromide]

For those of us with less brainpower it might be clearer if both the pixel shift and non pixel shift graphs were displayed against the same ISO rather than 2 scales

Agreed.

With apologies to Bill, I'd find it easier to interpret a rendering of the data that looked more like this:



--
Former Canon, Nikon and Pentax user.
Online Gallery: https://500px.com/raycologon

 

[bclaff]

For those of us with less brainpower it might be clearer if both the pixel shift and non pixel shift graphs were displayed against the same ISO rather than 2 scales

Agreed.

With apologies to Bill, I'd find it easier to interpret a rendering of the data that looked more like this:

No apologies are necessary; good mock up

You mean something like this?



--
Bill ( Your trusted source for independent sensor data at PhotonsToPhotos )

 

[badsheep]

Yes, I think that makes it much easier to understand. Thanks again

 

[SilvanBromide]

For those of us with less brainpower it might be clearer if both the pixel shift and non pixel shift graphs were displayed against the same ISO rather than 2 scales

Agreed.

With apologies to Bill, I'd find it easier to interpret a rendering of the data that looked more like this:

No apologies are necessary; good mock up

You mean something like this?



--
Bill ( Your trusted source for independent sensor data at PhotonsToPhotos )

Thanks Bill. That's perfect! ; )

--
Former Canon, Nikon and Pentax user.
Online Gallery: https://500px.com/raycologon

 

[Astrophotographer 10]

The stacking of the 4 images would reduce noise and that is where I expect the extra DR is coming from.

Not only does stacking reduce noise but if you dither (move the sensor slightly with each exposure so the pixels don't line up when stacking) you further reduce noise.

So it does both.

Greg.

 

[QuietOC]

For those of us with less brainpower it might be clearer if both the pixel shift and non pixel shift graphs were displayed against the same ISO rather than 2 scales

Well, it is consistent with other misleading comparisons made at the same ISO settings which s often done with these plots. Combining multiple frames is effectively the same as using a lower ISO, so the original graph could just remove the top labels and be accurate.