[beholder3]

I think, "dynamic pixel shift" is a less dramatic wording for "super resolution". And pretty much using 99-100% of all logic and causality of this. It obviosuly also shares all others aspects and limitations.

If you want to see how it works in the old fashioned, slow, fully manual, painful, work-on-my-PC way, look here:

A Practical Guide to Creating Superresolution Photos with Photoshop

The genius of Pentax new offering is to implement this all into the little camera itself. Push a button - done.
And that is a big step for the mass of photographers who do not want to spend ages on a PC.

[Adam]

I think, "dynamic pixel shift" is a less dramatic wording for "super resolution". And pretty much using 99-100% of all logic and causality of this. It obviosuly also shares all others aspects and limitations.

If you want to see how it works in the old fashioned, slow, fully manual, painful, work-on-my-PC way, look here:

A Practical Guide to Creating Superresolution Photos with Photoshop

The genius of Pentax new offering is to implement this all into the little camera itself. Push a button - done.
And that is a big step for the mass of photographers who do not want to spend ages on a PC.

Pixel shift is Pentax's term for in-camera super resolution. It came out in 2015 along with the K-3 II and has been included on all models launched since then.

Dynamic pixel shift is new this year in the K-1 II, and essentially allows you to use super resolution (pixel shift) hand-held. The results will be somewhere between a standard photo and a pixel shift photo shot on a tripod.

[jbinpg]

It also allows those togs who would like to use flash with Dynamic Pixel Shift to do so. Limited use case but at least it is possible now.

[stevebrot]

I think, "dynamic pixel shift" is a less dramatic wording for "super resolution". And pretty much using 99-100% of all logic and causality of this. It obviosuly also shares all others aspects and limitations.

You are correct, though the Pentax implementation has an advantage in that it provides automatic alignment.

Edit: Assuming "super resolution" to mean the technique of combining multiple exposure taken from slightly different camera position to increase spatial resolution. A good discussion and tutorial are available in the seminal PetaPixel article:

https://petapixel.com/2015/02/21/a-practical-guide-to-creating-superresoluti...ith-photoshop/


Steve

---

Last edited by stevebrot; 04-13-2018 at 11:07 AM.

[CharLac]

I would imagine, like HDR, things need to be static for this to work at all....i.e. things cannot be blowing in the wind

[stevebrot]

It also allows those togs who would like to use flash with Dynamic Pixel Shift to do so. Limited use case but at least it is possible now.

I have seen a couple of comments on this site suggesting possible flash support for dPS. Do you have a reference or is this an assumption?


Steve

[photoptimist]

You are correct, though the Pentax implementation has an advantage in that it provides automatic alignment.


Steve

Exactly!

Moreover, Pentax dynamic pixel shift does have one big performance advantage over any Photoshop or post-process PC version of this process. The Pentax version can use SR sensor data to more accurately align the images to each other. The post-process version is forced to estimate the alignment from the image data itself. Image noise, Bayer filter moire, subject movement, lighting changes, etc. all add noise and error to the post-process version of superresolution. The Pentax version has independently measured data on the camera's position and pointing directions that it can use.

Thus, it's an improved version of superresolution.

[stevebrot]

Thus, it's an improved version of superresolution.

When the cameras are more common "in the wild", it would be tempting to directly compare dPS with classic "super resolution" to see which approach delivers more.


Steve

[Medex]

Exactly!

Moreover, Pentax dynamic pixel shift does have one big performance advantage over any Photoshop or post-process PC version of this process. The Pentax version can use SR sensor data to more accurately align the images to each other. The post-process version is forced to estimate the alignment from the image data itself. Image noise, Bayer filter moire, subject movement, lighting changes, etc. all add noise and error to the post-process version of superresolution. The Pentax version has independently measured data on the camera's position and pointing directions that it can use.

Thus, it's an improved version of superresolution.

Why you think that dPS is improved version of superresolution? Short test in dpreview shows that dPS in terms of resolution is not better than 1 picture made in electronic shutter mode. Normal PS gives better resolution (so, it could be named "super resolution).
And another thing - does pixel shift process (shifting the sensor by 1 pixel) really occurs during 4 shutter actuations?

[biz-engineer]

And another thing - does pixel shift process (shifting the sensor by 1 pixel) really occurs during 4 shutter actuations?

Yes. That's amazing but that is true. Evidence is the nature of frame stacking artifact in moving parts of an image. I don't know how they manage to move the sensor by one pixel, but they do it...

Short test in dpreview shows that dPS in terms of resolution is not better than 1 picture made in electronic shutter mode

There is one application of multiframe alignment and stacking that I'm aware of , and it is used by Sony, if I remember correctly. Shooting after sunset without a tripod. Now with the K1 Mk II that will also be possible to get clean images by increasing ISO, and shooting an hhps sequence without a tripod. I see it as a great application of hand held pixel shift, and much more beneficial that trying to get a tiny bit more sharpness on top of 36Mpixel which is plenty already.

[monochrome]

Why you think that dPS is improved version of superresolution? Short test in dpreview shows that dPS in terms of resolution is not better than 1 picture made in electronic shutter mode. Normal PS gives better resolution (so, it could be named "super resolution).
And another thing - does pixel shift process (shifting the sensor by 1 pixel) really occurs during 4 shutter actuations?

Rishi openly admits in comments his analysis was flawed due to use of incorrect PP software to evaluate final images. Same thing happened with original PixelShift

[LensBeginner]

I've always though of pixel-shift as a special case of superresolution, just like multi-frame averaging is another one.
To use the special term to describe the broader sense is tantamount to false advertising, IM(not H in this case)O.

I was among the first to state that it coulndn't possibly be what the name tried to suggest it was, so it didn't deceive me (nor many others), but still...

Rishi openly admits in comments his analysis was flawed due to use of incorrect PP software to evaluate final images. Same thing happened with original PixelShift

A DPR editor wrong about Pentax... how strange!
A DPR editor admitting he's wrong about Pentax... how quaint!

[stevebrot]

Rishi openly admits in comments his analysis was flawed due to use of incorrect PP software to evaluate final images. Same thing happened with original PixelShift

Poor Rishi...he has done this before and probably hates Pentax product assignments.


Steve

[stevebrot]

And another thing - does pixel shift process (shifting the sensor by 1 pixel) really occurs during 4 shutter actuations?

That is the claim. Other brands claim similar for their pixel-shift implementations so the claim is probably credible, either that or universally doubtful.


Steve

[gaweidert]

Here are a few shots I just took using it. No post processing applied. The first is some crocuses. In the second the anemometer on my weather station is rotating in a light breeze. The second image is heavily cropped. Both shot at ISO 3200 on a dreary day.