Question about PixelShift

I would appreciate any insights into the following:

The K1 allows 2 variants of PS, with and without Motion Correction (MC). I am assuming that the RAW image is the same for both variants and it only when the RAW is converted to JPG that the correction takes place.

The latest version of Adobe Camera Raw 'supports' K1 PixelShift RAW images, but it isn't clear to me what 'support' means in this context.

Does support just mean that ACR will open the files and process just 1 of the 4 images into TIFF/JPG or does ACR combine the 4 images into 1?

If the latter, does it implement Motion Correction?. In DCU5 this is a selectable option for conversion.

Thanks

davek

My suspicion is that for this very specific and model limited feature we'd have to use PDCU to carry out the initial PS options and basic RAW output adjustments. Then just save a master TIFF to use in another editor if needed. This is surely likely to lead to best results given that the software is dedicated to those tasks specifically for the K1 files.

If you send the K-1 to me I'll test it for you and post back in about a week with the answer

Always happy to oblige

--

I'm not completely sure on this, but the raw file image I tried in pixel shift mode loaded OK into Elements 11, the only oddity was the file size of the dng was massive, 160mb compared to a normal one's 40mb. So it seems the camera does any necessary processing (it takes a moment or two to combine the four images) and presumably the camera jpeg is ready processed. So one can see right away if its worked OK or if need to try again. It seems much the same system as the in camera's HDR process, you don't need any dedicated software for that.
What I will say is that the out of camera jpegs from the K-1 are the best I've seen so far, seems post processing makes them worse!

Understandable suspicions, but in the first release of ACR that supported PixelShift, Adobe wrote:

"Please note the following limitations specific to camera support for the Pentax K-3 II:
Pixel shift Resolution (PSR) mode was added in this release. It is recommended to use a tripod and use a 2-second timer to minimize motion artifacts when in PSR mode."

This implies to me that Adobe does combine the 4 images, as this would only matter if the images are combined?.

Hard facts are surprisingly difficult to obtain in this 'Information Age'. A single sentence from Adobe would resolve all ambiguity, but I've had no response to my query.

davek

Like your style.....

davek

I experimented with a pixel shift image in RAW with movement control, and the four images were automatically combined in camera to produce a single RAW file of almost 200 Mb! I guess that you then process this image as normal. Compared to a normal exposure of the same view there was additional sharpness where the shot was in focus, but I hadn't used the timer or remote cable as is recommended so it could have been better.

I think the 2 second timer rule is correct, I had some blurry and some not last night
2 second timer
link
no timer
link

Obviously the 4 PS images are combined in-camera when shooting JPGs (and possibly to produce the embedded JPG in a RAW file) but I suspect that the large (160+Mb) RAW files are effectively just the 4 separate RAW images in one file, which allows combination in post. Thus PDCU can derive MC and non-MC images from the same RAW file.

It's what post-processing is done by ACR that I'd like to know....

davek

I would be very surprised if a third party software product implemented such a camera-specific feature ....

If one only got one of the four images that make up the raw file surely it would be either all red, all green or all blue. But it isn't. The camera processes the data and combines it to produce the jpeg so presumably the raw file is combined too. (Presumed being the operative word!) Maybe PDCU can separate the layers back out. I'm in no rush to install it to find out given its previous sluggish performance that led me to uninstall it last time.

Perhaps someone with a K-3ii has experience of how to process it?

But the K3II pixel shift doesn't have the motion correction feature, is that right?

Right... I've tried pixel shift with the K3 II on various bits of water... Essentially it looks like a multiple exposure on the moving water bits, which I suppose in many ways it is...

Best

Well, from what I understand, MC just involves scanning all 4 images and, where there is evidence of motion between images, using pixels from just 1 image. ACR could do it, my question is, does it.

What may illuminate this somewhat would be to see if the latest version of PDCU offers MC to PS images from a K3II. Can somone point me at a K3II PS image to try this. The K3II obviously can't do it in-camera without new firmware.

davek

I've just converted my pixel-shift test image using Adobe's latest DNG converter (v.9.5.1.)
This is a 100% crop (actual pixel crop)




So it is clearly each image merged but no other processing done on it - seems ACR doesn't do it.

This is the in camera jpeg version at actual pixel crop.



and this is how the older Adobe Camera Raw 7.3 handled the dng file




Not sure what to make of them all!

In case you're wondering, this was from near the edge of the frame and wasn't the point of focus but it does show how the image was processed.

Clearly, any use of pixel-shift requires the use of a very rigid tripod, not hand holding as i've seen suggested elsewhere. And any failed images can be rescued in ACW 7.3 which looks better than the camera's own jpeg version.