[Andr3w]

Digital Camera Utility 5 allows you to export a pixel shifted DNG with Pixel Shift, Pixel Shift MC or no PS. Can you also extract a non-pixel shift version of the pixel shifted DNG file in Silkypix 7?

[Adam]

My guess would be that silkypix won't know what to do the file, so it will be opened as a standard image. That's what ACR did before the updates were released.

[Kunzite]

Silkypix supports Pixel Shift Resolution. It's not perfect, but it does.
However, in Silkypix 6 I wasn't able to control the process; if there's some setting it's well hidden.

[tduell]

Digital Camera Utility 5 allows you to export a pixel shifted DNG with Pixel Shift, Pixel Shift MC or no PS. Can you also extract a non-pixel shift version of the pixel shifted DNG file in Silkypix 7?

I doubt it, but you can with dcrawps.

Cheers,
Terry

[OttoE]

Shutter shake ist much lower when using Pixel-Shift-Modus. With long lenses on a tripod I often used the Pixel-Shift-Modus, but develop only a normal picture in Digital Camera Utility 5.

[MJSfoto1956]

quick answer: Yes.
I just tested a K-3ii pixel shift image with DCU-5. There are four export settings under Image/PixelShiftResolution:

• Camera Setting
• Off
• On (motion detection off)
• On (motion detection on)

I set DCU-5 PixelShiftResolution to "on", then exported the image. Then I set it to "off" and exported a 2nd image. Layered the two images in Photoshop with layers setting to "difference". The attached screenshot shows that the two exported images are clearly NOT the same, suggesting that one can indeed export a single image from the set of four embedded into the 128mb pixel shift PEF (presumably the first?) .

Michael

---------- Post added 06-19-16 at 08:51 PM ----------

I should also point out that all four settings provide slight differences that require an extreme curve setting to visualize (see the screenshot above)

M

[jbinpg]

To expand on Terry's post about dcrawps, the following is a little primer on dcrawps.

Here is how I split a pixel-shifted DNG. This would be for those who would like to layer images and paint over pixels themselves:

$ dcrawps -s all -T -4 -w K1A_2688.DNG (-s all = split all constituent layers into separate images; -T = write out in tiff format; -4 = write out in linear 16-bit, i.e. no gamma curve, no brightness changes; -w = use camera WB).

And I get K1A_2688_0.tiff (bayer filtered), K1A_2688_1.tiff (R data), K1A_2688_2.tiff (G data), K1A_2688_3.tiff (B data)

Here is an example of how dcrawps is used for motion correction of a pixel shifted image. It has two abilities: motion detection and motion correction.

Motion detection -

dcrawps -X -G 1 -T -4 -w K1A_2685.DNG

The -X option highlights motion-detected pixels in magenta.

The -G option sets the threshold (in percent) for detecting differences between the two green channel pixels (a lower threshold will have more pixels highlighted).

The -T option produces a tiff file.

The -4 option produces a linear 16-bit tiff (no gamma curve applied, no brightening done).

The -w option uses the white balance settings from the camera.

Here is an image resulting from the above command. Prepare to be shocked. A 1% difference can be exceeded just by the effects of air motion:



This was taken in the morning when the mountain breezes were just getting started with slight swaying of the lupines and rustling of the aspen leaves.

Here is the command setting a 20% green channel threshold to be exceeded:

dcrawps -X -G 20 -T -4 -w K1A_2685.DNG

And the resulting image:



That is more like it. Only the vegetation is showing some movement. So replacing the marked pixels should result in a motion corrected image.

Motion repair -

dcrawps motion correction works by comparing the bayer filtered green channel to the 3rd G data image. If any pixel differences in the G channel exceed the threshold you gave in the command, those pixels are replaced by the ones from the bayer filtered image. However, the syntax is different than what one would intuitively think. You need to think about what percentage of the pixels you do not want replaced for the repair command. So to keep 80% of the image pixels (i.e., replace 20% of the pixels), the command would be:

dcrawps -G 80 -T -4 -w K1A_2685.DNG

And you get this:



You will notice that there is still a tiny amount of movement visible in the lupines. Setting the repair threshold to 70 takes care of that.

dcrawps -G 70 -T -4 -w K1A_2685.DNG



There is always some "slop" with the motion correction using dcrawps and trial and error is the best way to get the best results. Note that since I asked for a linear tiff, the image will be flat as a board and will need some tender loving care in post to bring it back to life.

I am nowhere saying that this method is superior to the motion correction achieved by DCU or any of the other programs out there that are able to decipher Pentax's pixel shifted images. It is just one more tool in the arsenal. Personally, I think the motion detection is worth the price of admission (free).

- Jack

[tduell]

A very useful post Jack, that should help all and sundry get to grips with this very useful software.

Cheers,
Terry

[jbinpg]

I forgot to post where dcrawps can be found:

https://github.com/tomtor/dcrawps

Both linux source code and compiled windows binaries can be found there.

Jack

[tduell]

$ dcrawps -s all -T -4 -w K1A_2688.DNG (-s all = split all constituent layers into separate images; -T = write out in tiff format; -4 = write out in linear 16-bit, i.e. no gamma curve, no brightness changes; -w = use camera WB).

And I get K1A_2688_0.tiff (bayer filtered), K1A_2688_1.tiff (R data), K1A_2688_2.tiff (G data), K1A_2688_3.tiff (B data)

Just a follow up on this.
Are you sure that the above are the 4 output images using "-s all"?
I was under the impression that the PS RAW simply contained the 4 images from the 4 sensor shifts, i.e. R, G, B, G, and that the dcrawps "-s all" just provided these as separate images, but I haven't really looked at the details of dcrawps since the update to support the K-1.
Do you have a reference to where the behaviour you described is explained?

Cheers,
Terry

[jbinpg]

Sorry, you are right, Terry. All 4 images are complete. But the author did tell me that the green channel comparison is only done between the 1st and 3rd images since they are the only frames where the green pixels are in the same position and able to be compared.

[tduell]

Sorry, you are right, Terry. All 4 images are complete. But the author did tell me that the green channel comparison is only done between the 1st and 3rd images since they are the only frames where the green pixels are in the same position and able to be compared.

OK, thanks for clarifying.

Cheers,
Terry

[Adam]

Here is an image resulting from the above command. Prepare to be shocked. A 1% difference can be exceeded just by the effects of air motion:

Great post. It's nice that this tool has become available!

I think this also shows how much bigger of a benefit can be obtained with PSR when you're shooting in a perfectly controlled (i.e. indoor) environment.

[rawr]

Daily reminder that Pentax Digital Camera Utility 5 is not Silkypix Pro 7.

[Jyrkira]

My guess would be that silkypix won't know what to do the file, so it will be opened as a standard image. That's what ACR did before the updates were released.

Silkypix Pro 7 does great job with pixel shift images. It even supports motion correction. What comes to original question I don't know if a single exposure can be extracted from pixel shifted raw file. I must try it when I have time to do it.