Forgot Password
Pentax Camera Forums Home
 

Reply
Show Printable Version Search this Thread
05-16-2016, 10:39 AM   #1
Veteran Member
microlight's Avatar

Join Date: Sep 2011
Location: Hampshire, UK
Posts: 319
Pixel shift motion correction in K-3II?

I was experimenting with pixel shift using my new K50/f1.2 using the K-3II. Developing the photos in the most recent DCU5 v5.5.1, I happened to notice in the EXIF table - 'Pixel shift resolution - On (Motion Correction On)'! Checking back on other PS shots right back to when I got the K-3II in November last year, they all say the same, and the PS icon in DCU5 shows as 'camera setting' - but you can also set it to off or on manually.

So does this mean that the K-3II has always had motion correction in pixel shift mode as per the new K-1, but it only needed DCU5 to catch up with the release of the K-1, in order to be able to display it?!? If so then this makes another strong case for using DCU5 for pixel shift development.

05-16-2016, 11:19 AM   #2
Administrator
Site Webmaster
Adam's Avatar

Join Date: Sep 2006
Location: Arizona
Photos: Gallery | Albums
Posts: 41,822
QuoteOriginally posted by microlight Quote
I was experimenting with pixel shift using my new K50/f1.2 using the K-3II. Developing the photos in the most recent DCU5 v5.5.1, I happened to notice in the EXIF table - 'Pixel shift resolution - On (Motion Correction On)'! Checking back on other PS shots right back to when I got the K-3II in November last year, they all say the same, and the PS icon in DCU5 shows as 'camera setting' - but you can also set it to off or on manually.

So does this mean that the K-3II has always had motion correction in pixel shift mode as per the new K-1, but it only needed DCU5 to catch up with the release of the K-1, in order to be able to display it?!? If so then this makes another strong case for using DCU5 for pixel shift development.
Well, the motion correction feature is essentially a JPEG development option, so I don't see why it would require files from a certain camera. It's great the PDCU lets it run on K-3 II files!

Adam
PentaxForums.com Webmaster (Site Usage Guide | Site Help | My Photography)



PentaxForums.com's high server and development costs are user-supported. You can help cover those costs by donating. Or, buy your photo gear from our affiliates, Adorama, B&H Photo, or Topaz Labs, and get FREE Marketplace access - click here to see how! Trusted Pentax retailers:

05-16-2016, 12:34 PM   #3
Pentaxian
CarlJF's Avatar

Join Date: Sep 2014
Location: Quebec City
Posts: 699
It looks like although the motion correction isn't available in-body, it is supported by DCU. Which is great!

That said, have you try to take a picture in PS mode with some motion in it and develop the DNG with DCU ? Just to confirm the motion correction actually does something on the K3ii DNG...
05-16-2016, 02:17 PM   #4
Pentaxian
Tjompen1968's Avatar

Join Date: May 2012
Location: Norrköping, Sweden
Photos: Gallery | Albums
Posts: 1,311
QuoteOriginally posted by CarlJF Quote
It looks like although the motion correction isn't available in-body, it is supported by DCU. Which is great!

That said, have you try to take a picture in PS mode with some motion in it and develop the DNG with DCU ? Just to confirm the motion correction actually does something on the K3ii DNG...
I took a file from the internet taken with the K-3II and ran it through DCU and used both with and with MC. Big difference between the shots. The MC option works fine with K-3II files.

05-16-2016, 02:25 PM   #5
Veteran Member
microlight's Avatar

Join Date: Sep 2011
Location: Hampshire, UK
Posts: 319
Original Poster
This is a 100% crop of a pixel shift photo taken on my K-3II in April. All I have done is open the DNG in DCU5 v5.5.1, and then take a screen capture with motion correction turned on and off. The output is clear - motion correction works in DCU5 with K-3II files! Excellent news.

On:


Off:
05-18-2016, 10:51 PM   #6
Forum Member




Join Date: Mar 2013
Posts: 79
Pixel Shift Motion Correction is a new feature in version 5.5.1
A good feature for K-3II users like me!
Thank you!

---------- Post added 05-19-16 at 01:21 AM ----------

Image with NO correction. Developed in DCU5.4.1. Resized and cropped in Photoshop CS6)



Image with Pixel Shift Motion Correction ON in latest version of DCU (5.5.1). Resized and cropped in Photoshop CS6



It really works!

Last edited by Amoon; 05-19-2016 at 06:28 AM.
05-19-2016, 08:46 AM   #7
New Member




Join Date: Jan 2014
Posts: 19
This makes sense in that the Pixel Shifted RAW files for both the K-3 II and K-1 simply consist of the 4 shots acquired. The RAW converter then is responsible for merging the 4 files together to make your RGB image. It's really nice that Pentax didn't artificially limit motion correction to the K-1. Of course, motion correction for in-body JPEGs is only available for the K-1.
05-19-2016, 08:40 PM   #8
Site Supporter




Join Date: May 2016
Photos: Gallery
Posts: 203
In-camera PixelShift

I tried the same as what's been posted in this thread, and yes, DCU 5.5.1 corrects for motion and the settings work. What I'm unclear about are two things a) in DCU, "camera setting" shows the same image as with motion correction explicitly on; b) in-camera raw files and jpeg are displayed with no motion artifacts. Does the K3II correct motion but less than the K1? I checked static areas of my test shots and they look more detailed with pixel shift than without, so it's not disabled in-camera if motion is detected. Is it just that DCU was updated to match what the K3II was doing anyway, or does it add better raw processing than what is available in-camera?


Last edited by aaacb; 05-19-2016 at 08:59 PM. Reason: clarification
05-20-2016, 07:30 AM   #9
Pentaxian
Not a Number's Avatar

Join Date: Mar 2012
Location: Venice, CA
Posts: 4,105
From what I can make out from an interview published on DC Watch translated from Japanese via Google is that the K3ii always had motion correction but cannot be turned off in the camera. And even though both the K-1 and the K3ii are using the same basic algorithms motion correction works better on the K-1.

I did say the easiest way to test if this was true or not in an earlier thread was to just process some pixel shift RAWs in DCU 5.5.1 with MC off. Good to see someone finally did this.

https://www.pentaxforums.com/forums/190-pentax-k-1/317115-dc-watch-interview-k-1-designers.html

QuoteQuote:
The --K-1 of realistic-resolution system, but joined a feature called motion correction, please tell us about this process.

Uehara: In K-1, but may be used as they are of the on / off motion compensation, in fact, have also moving object correction is incorporated in the K-3 II realistic-resolution system, always moving object correction it turned on.

--K-3 II faster-moving, such as automobile subject, but I have noticed is that in which they are processed not in multiple, motion correction of the K-1 had thought it became more powerful.

It can be photographed K-1 realistic Resolution System with motion compensation on, not is stronger in a moving scene than K-3 II that?

Uehara: only sentence in which the number of pixels is increased, but those of the K-1 might fine movement has increased more discrimination can case impression of the subject, the basically the same motion correction algorithm K-3 II also K-1 is adopted it has.

- Such as the scene in which the cloud is flowing, or is that slow the movement of the subject is that the still weak?

Uehara: clouds of flow Toka trees of buzz is a difficult subject for detecting a good movement, we recognize that it just weak realistic-resolution system.

- So, K-1 of a moving object correction off, Will what circumstances has been prepared taking into consideration?

Uehara: because motion compensation by comparing the image of the four shots, which part is moving, or not moving, for it has to determine, what is really at rest and there is a very fine picture I also I'm certain it would determine that the wrong move in.

And a discriminated area moving, replaced with images taken first sheet, since the super-resolution processing is not performed, it becomes the same described as normal shooting. For this reason, such as the still life shooting in the studio, consider the case, such as to shoot a subject that does not work absolutely, I was to be able to shoot with the moving object correction off.

By the way, if you shoot with a realistic-resolution system, so much noise of attitude is less than the normal shooting, the sentence, the person who was taken at a fast shutter speed to increase the sensitivity, with less subject blur I think that a clear image is obtained.

- When you shoot at realistic-resolution system, the line is very thin, I think you better also tonality?

Uehara: normal in the sensor of the Bayer arrangement, because it does not only obtained RGB any one color of information in a single pixel, but has created the color and brightness information that is not enough to estimate the pixel interpolation, realistic Resolution When taken with systems, since all RGB color and luminance information in one pixel is obtained, it is not necessary to interpolation processing, it leads to a final sharpness by going to correct the color reproduction of fine place you.

- In the normal shooting and realistic-resolution system, I think I'm naturally also different over how the sharpness?

Uehara: different. In the case of the realistic-resolution system, since the sputtering and the two points are separated, undershoot, when the sharpness processing, such as put the overshoot, because that would impair the dense detail depiction, large edge portion as much as possible to, you do not put the border, we are emphasizing process so as to reverse to highlight the texture of the parts of the face.

- By using a lens with a magnification chromatic aberration, what happens when you shoot in the realistic-resolution system?

Uehara: If a lens corresponds to a lens correction, but it is output by correcting the chromatic aberration of magnification, more magnification chromatic aberration and it is non-compliant lenses It stands out and clearly (laughs).
05-20-2016, 08:36 AM   #10
Loyal Site Supporter




Join Date: Apr 2015
Photos: Albums
Posts: 505
I think the newer PDCU does just that: it uses the in camera algorithm to remedy motion problems, only for RAW not JPEG. I also think that it may apply some more sharpening by default, although the sliders don't indicate that. Or maybe it's just better at PS processing. I've tried some K-1's and K-3II's and I'm not sure that the MC is better, it's perhaps that the image is better (you really notice if your lens isn't expensive, er sharp, enough, and I image a FF just produces a better image than a crop for this purpose.

The odd thing is that on some images I can't seem to turn MC off entirely in PDCU; not sure if that's cuz it's saving settings, etc, since I don't really have a good handle on how it works but sometimes the MC icon just doesn't work on my Mac version.
05-20-2016, 12:07 PM   #11
Site Supporter
stevebrot's Avatar

Join Date: Mar 2007
Location: Vancouver (USA)
Photos: Gallery | Albums
Posts: 26,849
I don't own a pixel shift body , but I am interested in the technology and how well the pixel shift motion correction works. Below are three crops taken from one of Imaging Resource's K-3II test image's DNG.

Web resolution image from Imaging Resource (PS On, no MC) (Full-size version with exif available HERE)



The three images below are full resolution crops taken from TIFF output processed from the Imaging Resource DNG using PDCU v5.5.1.

No Pixel Shift



Pixel Shift On, No motion correction




Pixel Shift On, motion correction applied




Hmmmmm...it appears that motion correction for a K-3II image may not work too well or perhaps it is not supported at all. But wait, comparing the full-frame pixel shifted without motion correction to the Imaging Resources JPEG reveals an interesting difference. Notice the upper left and the tree leaves at upper right.

Full frame pixel shift, no motion correction (PDCU v5.5.1)




Apparently the motion correction is doing something and whatever it did was available when ImagingResources created their (in-camera) JPEG.
The in-camera JPEG (taken July 13, 2015) is almost identical to the motion corrected DNG conversion done using PDCU v5.5.1. The plot thickens!



Steve

Last edited by stevebrot; 05-20-2016 at 12:25 PM.
05-20-2016, 12:44 PM   #12
Site Supporter




Join Date: May 2016
Photos: Gallery
Posts: 203
Thanks for confirming that the k3ii has compensation built in. In my experience it works well, but not needing to develop pixel shift raw files only in camera is definitely great! I wonder how much the sensor stabilization has to do with compensation.
05-20-2016, 03:56 PM - 1 Like   #13
Site Supporter
stevebrot's Avatar

Join Date: Mar 2007
Location: Vancouver (USA)
Photos: Gallery | Albums
Posts: 26,849
QuoteOriginally posted by aaacb Quote
Thanks for confirming that the k3ii has compensation built in. In my experience it works well, but not needing to develop pixel shift raw files only in camera is definitely great! I wonder how much the sensor stabilization has to do with compensation.
SR is turned off in pixel shift mode. That is why hand-holding is not recommended.

BTW...Welcome to the Pentax Forums!


Steve
05-20-2016, 06:04 PM - 1 Like   #14
Pentaxian
Not a Number's Avatar

Join Date: Mar 2012
Location: Venice, CA
Posts: 4,105
Well the impression I got from the DC Watch interview is that motion correction (MC) has always been a feature of the K3ii. Landscape shots with leaves or grass wouldn't be a good examples because Ricoh was saying (so I believe from the translation) motion is difficult to detect in detail like leaves or grass. Easily identified moving objects such as automobiles or ice skaters are no problems. But people weren't taking Pixel Shift shots with these elements.

The question I have is if the MC PS RAWs are encoded differently than non-MC PS RAW files. This can be easily tested on the K-1 by turning off MC but turning it on in DCU 5.5.1 for developing the RAW and comparing to developing with MC turn off (in DCU). If there's no difference between the output then the RAW has to be shot with MC on in the camera. I've requested people try this but haven't checked for any follow up.
05-21-2016, 08:17 AM   #15
Loyal Site Supporter




Join Date: Apr 2015
Photos: Albums
Posts: 505
QuoteOriginally posted by Not a Number Quote
Well the impression I got from the DC Watch interview is that motion correction (MC) has always been a feature of the K3ii. Landscape shots with leaves or grass wouldn't be a good examples because Ricoh was saying (so I believe from the translation) motion is difficult to detect in detail like leaves or grass. Easily identified moving objects such as automobiles or ice skaters are no problems. But people weren't taking Pixel Shift shots with these elements.

The question I have is if the MC PS RAWs are encoded differently than non-MC PS RAW files. This can be easily tested on the K-1 by turning off MC but turning it on in DCU 5.5.1 for developing the RAW and comparing to developing with MC turn off (in DCU). If there's no difference between the output then the RAW has to be shot with MC on in the camera. I've requested people try this but haven't checked for any follow up.
I assume they mean difficult to detect visually. It's not that tough to detect it with software. You can do it down to the pixel level.

My understanding is that MC is a software thing. So even if you have it "on" in the camera it does nothing viz a viz the RAW files, although it might affect the embedded JPEG preview. I could find settings for other corrections in metadata like diffraction, etc but nothing for motion in exiftool. But PDCU DOES show MC settings in the metadata. Confusingly, it shows MC ON for K-3II files.

I think the camera just records the four frames and it's down to the processor to correct the motion. That's different than actually recording different data depending on whether it's on or off. And it's also possible that the processor in the camera does a better/worse job correcting motion than the software; I wouldn't assume the algorithm is the same. PS is sort of the opposite of Astrotracer: it's shifting to gather info about stationary objects rather than shifting to track moving objects.
Reply

Bookmarks
  • Submit Thread to Facebook Facebook
  • Submit Thread to Twitter Twitter
  • Submit Thread to Digg Digg
Tags - Make this thread easier to find by adding keywords to it!
correction, cs6, dslr, image, k-1, k-3, k-3ii, k3, motion, pentax k-3, pixel, pixel shift motion, ps, shift, shift motion correction
Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
Pixel Shift Motion Correction loco Pentax K-1 15 06-30-2016 12:31 AM
Pixel Shift in LR CC? enoeske Pentax K-1 58 05-24-2016 11:14 AM
What's "pixel shift motion correction function"? Fenwoodian Pentax K-1 36 02-27-2016 09:55 AM
K-1 and pixel shift Gimbal Pentax Full Frame 31 02-15-2016 10:14 PM
Pixel shift in Lightroom and ACR! 08amczb Digital Processing, Software, and Printing 14 08-02-2015 10:30 PM



All times are GMT -7. The time now is 05:53 AM. | See also: NikonForums.com, CanonForums.com part of our network of photo forums!
  • Red (Default)
  • Green
  • Gray
  • Dark
  • Dark Yellow
  • Dark Blue
  • Old Red
  • Old Green
  • Old Gray
  • Dial-Up Style
Hello! It's great to see you back on the forum! Have you considered joining the community?
register
Creating a FREE ACCOUNT takes under a minute, removes ads, and lets you post! [Dismiss]
Top