[biz-engineer]

Noted from experience and the comprehensive PF review of the K1, in camera K1 JPEG conversion is lossy, to the point that when using in camera jpeg conversion the output images are no any better than images from a K3 (that's what I get). In order to gain from the 36Mp FF sensor, one must develop raws with PDCU or other third party software.

Now... where is the pixel shift implementation business logic?
Pentax pixel shift is a specific implementation that other mainstream camera manufacturers don't offer, and therefore pixel shift processing is not natively implemented in third party photo processing software (LR, DXO , whatever).
And the raw files that contains pixel shifted frames aren't boiled down to one raw file but 4 raw files, that's nice if you use PDCU or in camera JPEG conversion.
So, pixel shifted raws are best converted with one of the two options: PDCU or in camera RAW to JPEG.
Now, from market adoption standpoint, I don't think PDCU is more used than industry defacto such as PS, LR and DXO. I think, most Pentaxian don't use PDCU.
In camera K1 raw to jpeg lossy compression eats part of the details gained by the pixel shift process. So basically, the only reasonable option is to use PDCU... great !

Why didn't Ricoh implement pixel shift with in camera raw to raw conversion? i.e convert 4 pixel shifted frames (with MC, or without MC) into a single raw file that can be converted into a JPEG in the same way as it is for any normal raw file.

Well, could still be done by firmware update. Oupps, am I asking for more? Please beat me :-)

---

Last edited by biz-engineer; 06-04-2016 at 01:25 AM.

Adobe Camera RAW and Lightroom do implement the PS but not with MC on. The way to do it is to crank the sharpness up. There is an instruction somewhere on the forum... An article.

[Adam]

Why didn't Ricoh implement pixel shift with in camera raw to raw conversion? i.e convert 4 pixel shifted frames (with MC, or without MC) into a single raw file that can be converted into a JPEG in the same way as it is for any normal raw file.

Because it would stop being a RAW file, which goes against the whole idea of RAW

The real issue is that right now, Photoshop (ACR) does not fully support the K-1's PSR files. Once updates are released, you will be able to use the sharpness slider to "activate" PSR as described in the article below.

Other software developers should (and possibly will) follow suit, though it's entirely in their hands how they decide to implement their interpolation. Having studied interpolation at a technical level, I can tell you that it's not easy to come up with a good general-purpose algorithm for it. Pentax deserves a pat on the back for having MC and pretty good processing to begin with

There is an instruction somewhere on the forum... An article.

Here it is:

Pentax K-3 II Review - Processing Pixel Shifted RAWs | PentaxForums.com Reviews

[biz-engineer]

Here it is: Pentax K-3 II Review - Processing Pixel Shifted RAWs | PentaxForums.com Reviews

Thanks Adam.

[Gimbal]

Why didn't Ricoh implement pixel shift with in camera raw to raw conversion? i.e convert 4 pixel shifted frames (with MC, or without MC) into a single raw file that can be converted into a JPEG in the same way as it is for any normal raw file.

Well, could still be done by firmware update. Oupps, am I asking for more? Please beat me :-)

It would be a very special raw file as it would already contain all information for every pixel, just as a BMP file, or tiff. And wouldn't really need any converter or developer at all. Don't know how a raw converter would react to that. Perhaps this info could be included into the "raw" file, but then I guess it would be form of tiff file.

A 16 bit compressed (loss less) tiff as an option to jpeg, now that would be nice.

[Adam]

It would be a very special raw file as it would already contain all information for every pixel, just as a BMP file, or tiff. And wouldn't really need any converter or developer at all. Don't know how a raw converter would react to that. Perhaps this info could be included into the "raw" file, but then I guess it would be form of tiff file.

A 16 bit compressed (loss less) tiff as an option to jpeg, now that would be nice.

A 16 bit compressed (loss less) tiff as an option to jpeg, now that would be nice.

You can develop PSR RAWs into TIFFs in-camera. The interpolation problem would still remain, however, so the image quality wouldn't be much better. There's no workaround to leaving 4 separate RAWs in the PSR RAW file IMO.

[Rondec]

I think you are wrong. I think most Pentaxians who are using pixel shift are now using DCU software due to the fact that it does include the MC implementation. That certainly is how my work flow goes RAW File to DCU followed by developing the 16 bit TIFF file in Lightroom. At this point, this is Adobe's issue and the question is whether they would put forth the effort to actually include MC implementation in ACR. I frankly doubt it, unless the number of Pentaxians grows astronomically, but they could do it.

[Adam]

At this point, this is Adobe's issue and the question is whether they would put forth the effort to actually include MC implementation in ACR. I frankly doubt it, unless the number of Pentaxians grows astronomically, but they could do it.

I agree that they probably won't bother with MC.

At the same time I think MC is an over-hyped feature. All it does is detect areas in which the 4 images are misaligned. If users want to go through the trouble of using ACR to develop their PSR photos, they should have the time to create a new layer for the non-PS image and simply erase parts with artifacts.

[Gimbal]

You can develop PSR RAWs into TIFFs in-camera. The interpolation problem would still remain, however, so the image quality wouldn't be much better. There's no workaround to leaving 4 separate RAWs in the PSR RAW file IMO.

What interpolation problem would still remain? Motion correction?
If it's a 16 tiff (is it?) you could manipulate it as much as a raw file without having to go through DCU, Which of course you can't do with a jpeg.

[Rondec]

I agree that they probably won't bother with MC.

At the same time I think MC is an over-hyped feature. All it does is detect areas in which the 4 images are misaligned. If users want to go through the trouble of using ACR to develop their PSR photos, they should have the time to create a new layer for the non-PS image and simply erase parts with artifacts.

I guess. I think it provides some automation to a process that otherwise would take quite awhile to manage.

[biz-engineer]

You can develop PSR RAWs into TIFFs in-camera. The interpolation problem would still remain, however, so the image quality wouldn't be much better. There's no workaround to leaving 4 separate RAWs in the PSR RAW file IMO.

YES! I had this idea and just tried it before you post. Result: works great ! I think that's the best route to develop PS images, no loss of detail, no need of software able to deal with the PS MC feature.

[Adam]

What interpolation problem would still remain? Motion correction?

The fact that the camera has already decided how to merge the 4 images and thereafter applied its image processing functions to the result, hence interfering with the sensor data that you'd otherwise encounter in a raw file.

I will test out in-camera TIFF development on a photo I took earlier and let you know if the result is any better than the JPEG.

[Gimbal]

It's getting interesting.
Are the tiff 8 or 16bit per channel? Remember that the sensor delivers 14 bit of data, so if they are using only 8 bits in the tiff they would be wasting a lot of info.

[Adam]

Are the tiff 8 or 16bit per channel? Remember that the sensor delivers 14 bit of data, so if they are using only 8 bits in the tiff they would be wasting a lot of info.

It's 8 bits. Side-by-side the TIFF and JPEG look absolutely identical on my screen, even after I tried pushing the shadows.

The only benefit would thus be the lack of compression, which is negligible anyway. I'd rather have a 20Mb JPEG than a 100Mb TIFF

Note that unlike the 645Z, the K-1 cannot natively shoot 16-bit TIFF files, as only RAW and JPEG are selectable.

[rechmbrs]

This is potentially a good idea but I feel we need to learn how to process PSR data on a computer where software changes and options can be evaluated. Coding then would proceed in firmware. I believe that the camera manufacturers are remiss in placing something like MC in firmware until it is robust enough to handle problems encountered in everyday usage. Ricoh/Pentax has egg on their face by the weakness of MC. Another case of marketing winning over quality.

PSR is too good an idea for marketing to mess it up. Why didn't Ricoh/Pentax ask K3-II owners to test the MC algorithm in a beta test type environment?

RONC

K1 + bunch of other hardware and software.