Forgot Password
Pentax Camera Forums Home
 

Reply
Show Printable Version Search this Thread
08-13-2017, 10:37 PM   #1
Site Supporter
Site Supporter
BruceBanner's Avatar

Join Date: Dec 2015
Location: Blue Mountains, NSW, Australia
Posts: 905
Hand held Pixel-Shifting...

So... I own a K-1 and am aware of the Pixel Shift feature and use it a fair bit myself, however up until now I have used it solely with motion correction off and with the K-1 mounted on a tripod (obviously with IS off and more often than not a 2 sec timer as well).

Of late I have seen a couple of pictures on flickr with a few different people taking hand held pixel shifted images, either landscape or perhaps of someone sitting on a bench. I know from my own use when snapping insects, even a 1-2 second exposure can mean the entire shot takes 4-8 seconds to complete so you're hoping the insect won't move during that time, so I noticed from the EXIF of some of these hand held pixel shifted shots the shutter speed is increased to 1/1600 for example and the ISO upped to 400 to compensate for correct exposure.

It got me thinking... is there a trade off when using pixel shifting vs the end results of a 'regular normal' shot whereby shutter speed and ISO perhaps are lower?
I had a quick go at hand held pixel shifting in my garden earlier, and it felt like that even when i raise shutter speed substantially the 4 shots still take the same time...
Example; I could set the shutter speed to 1/1600 or 1/250, and the complete process (not counting the actual 'stitching and processing image part at the end')
still felt the same.

I know from my own experience of using the HDR feature hand held, the results are less than stella. I much prefer bracketing and processing off camera, and one step further.. to use a tripod in that process as well.

I just have serious doubts about hand holding the K-1 for pixel shifting... prove me wrong!

No.. but in all seriousness, has there been a more indepth comparison of hand held pixel shifting vs regular shooting and whereby the trade off occurs? If you start having to go above ISO 800 for example, is the extra noise vs the extra sharpness not worth it? Is this all subjective?

Cheers,

Bruce

08-13-2017, 11:05 PM   #2
Forum Member




Join Date: Sep 2016
Photos: Gallery
Posts: 68
It is an interesting idea which I have never thought of doing. I think I will try. The light here is generally bright, so I could get good speeds for landscape shots.
08-14-2017, 12:00 AM - 1 Like   #3
Pentaxian




Join Date: Feb 2015
Photos: Albums
Posts: 3,756
QuoteOriginally posted by BruceBanner Quote
is there a trade off when using pixel shifting vs the end results of a 'regular normal' shot whereby shutter speed and ISO perhaps are lower?
The interval between each of the 4 pixel shifted frames is approx. 1/4th of a second. That means, even if the shutter speed and ISO are increased, motion blur is still very likely to happens because of the time delay between frames.
Now, for non moving subjects, I've found that PS yield the best results with lowest possible ISO and shutter speeds set at 1/4th of a sec. because a faster than 1/4th sec. wouldn't make the 4 frames taken much faster, and longer than 1/4th sec. would likely make PS more prone to motion artifacts. That said, you can still use PS in RAW, with OVF and take only one frame when converting to JPEG, it's like having an all electronic shutter while still using an OVF, which means, depending on shutter speed, sharper picture thanks to lack for mirror induced vibration and lack of shutter shock. However, all electronic shutter means that subject motion should still be rather limited not to get any rolling shutter effect. You can try different things and see how results compare.
08-14-2017, 12:30 AM   #4
Pentaxian
microlight's Avatar

Join Date: Sep 2011
Location: Hampshire, UK
Posts: 430
I tried hand-holding some pixel shift pictures with the K-3II, and got no improvement over a standard shot, whereas on a tripod with a two-second delay, there is a visible difference. This might be because PS relies on the frames being lined up at pixel level which would be very difficult to achieve hand-held between the separate frames, regardless of how fast the shutter speed is for the individual frames.

For PS, pressing the shutter release blanks out the viewfinder for what seems like an eternity when you have the camera up to your eye, so there's no visual reference to confirm that you haven't moved slightly. Then, if you have moved, with motion correction off the output will be a little blurry, negating the effect of PS; and with motion correction on, the camera will select a reference frame and likely ignore the rest, so you probably end up with a standard non-PS shot.

Don't get me wrong - it would be great to be able to hand-hold PS pictures, but I haven't been able to do it effectively so far. Even with propping my elbows on a bridge parapet, I can't guarantee that I haven't moved just a fraction between the frames.

(BTW, the K-3II doesn't have motion correction control in camera - it's always on - but you can select on or off in DCU5.)

08-14-2017, 01:32 AM   #5
Site Supporter
Site Supporter
BruceBanner's Avatar

Join Date: Dec 2015
Location: Blue Mountains, NSW, Australia
Posts: 905
Original Poster
QuoteOriginally posted by biz-engineer Quote
The interval between each of the 4 pixel shifted frames is approx. 1/4th of a second. That means, even if the shutter speed and ISO are increased, motion blur is still very likely to happens because of the time delay between frames.
Now, for non moving subjects, I've found that PS yield the best results with lowest possible ISO and shutter speeds set at 1/4th of a sec. because a faster than 1/4th sec. wouldn't make the 4 frames taken much faster, and longer than 1/4th sec. would likely make PS more prone to motion artifacts. That said, you can still use PS in RAW, with OVF and take only one frame when converting to JPEG, it's like having an all electronic shutter while still using an OVF, which means, depending on shutter speed, sharper picture thanks to lack for mirror induced vibration and lack of shutter shock. However, all electronic shutter means that subject motion should still be rather limited not to get any rolling shutter effect. You can try different things and see how results compare.
Ah yes, so that explains how I felt that 1/1600 vs 1/250th felt absolutely no different in overall length of time start to finish of PSing.
If what you're saying is true, then the above example I gave with a man sitting on a bench in the park needed not be 1/1600th with 400ISO pixel shifted, a higher quality could have been achieved with ISO 100 and a lower shutter speed (the focal length looked wide so perhaps even 1/100th of a second should be sufficient with a stationary subject etc). Basically what I am getting at is in this particular case there wasn't an advantage to succeeding with a Pixel Shifted image trying 1/1600 ISO 400 over/vs say 1/100 ISO 100 (with all things being equal in terms of exposure, I'm kinda making these figures up to make the point)?

What is OVF might I ask? I only use LR and Photoshop (I usually bypass Camera Raw and head straight into Photoshop), what you're saying is that with OVF you can select one of the 4 frames used as a DNG and then play with that single frame and edit and make JPG?

Because...

QuoteOriginally posted by microlight Quote
I tried hand-holding some pixel shift pictures with the K-3II, and got no improvement over a standard shot, whereas on a tripod with a two-second delay, there is a visible difference. This might be because PS relies on the frames being lined up at pixel level which would be very difficult to achieve hand-held between the separate frames, regardless of how fast the shutter speed is for the individual frames.

For PS, pressing the shutter release blanks out the viewfinder for what seems like an eternity when you have the camera up to your eye, so there's no visual reference to confirm that you haven't moved slightly. Then, if you have moved, with motion correction off the output will be a little blurry, negating the effect of PS; and with motion correction on, the camera will select a reference frame and likely ignore the rest, so you probably end up with a standard non-PS shot.

Don't get me wrong - it would be great to be able to hand-hold PS pictures, but I haven't been able to do it effectively so far. Even with propping my elbows on a bridge parapet, I can't guarantee that I haven't moved just a fraction between the frames.

(BTW, the K-3II doesn't have motion correction control in camera - it's always on - but you can select on or off in DCU5.)
... if what microlight is saying is correct (to my understanding), then with Motion Correction On, and when hand holding the camera, there is a strong likely hood that only one frame is used and 3 discarded due to too much movement, especially with longer focal lenses. Because when chimping my hand held garden shots today I thought to myself "yeh.. they came back ok actually" but of course you're not going to see the benefits from PS on the LCD screen, nor did i pay strict attention to the file size at the time to confirm it actually was 150mb and therefore 4 images stitched successfully! I shall verify!

Is it obvious if the PS shot failed and only 1 frame used (ie lower fle size). Are these handheld Pixel Shifted images I've seen actually in essence likely to be a single frame after all? Because I can see a difference when Pixel Shifted has been used some of the time but not always...
08-14-2017, 04:08 AM   #6
Loyal Site Supporter
Loyal Site Supporter




Join Date: Jan 2007
Location: Conroe, TX USA
Posts: 328
OVF is optical view finder.

I'd like to see comparison of ACR (LR or PS) to Rawtherapee with Pixel Shift "movement correction" turned on. ACR has no correction at all. The correction in RT is highly superior to Pentax program version.

There have been a few hand held PixelShift examples pass by but none showed as being close to the tripod comparisons.

RONC
08-14-2017, 05:02 AM - 2 Likes   #7
Pentaxian
dcshooter's Avatar

Join Date: Jan 2011
Location: Washington DC
Photos: Gallery | Albums
Posts: 4,699
It doesn't cover all scenarios, but for static subjects where light is low, and tripods are not an option, I have had more success stacking in Photoshop than I have in using the built-in pixel shift mode.

Here, for example, is an image of a Picasso I took at the National Gallery of Art over the weekend:


Since no flashes or tripods were allowed in the museum, I ended up taking these photos at ISO1600 1/60 w/ a 35mm lens to get a reasonably sharp image w/o a lot of shake. I also boosted about .8 stop in post, so effectively close to ISO3200 for each exposure.

For the purposes of this exercise, I haven't or sharpened or straightened the edges of the composite above, so as not to introduce any additional artifacting, hence the pincushioning/slight misalignment.

This imageis composed of three shots, loaded into a stack and auto-aligned under File->Scripts. Then the top two layers are reduced to 33% opacity each, leaving each pixel in the resultant image an average of the three values from each photo at each site. While this doesn't have the single sensor photosite exactitude of the in-camera pixel shift, you can take as many photos as you want, giving a more and more precise value for each pixel with each one. Simply use opacity values determined by the number of shots, so for 4, 25%, for 5, 20%, and so on. The bottom layer always stays at 100%. This approach is very similar to the one whereby astrophotographers use a series of shots to achieve lower noise/better sharpness of distant objects, and a simpler version of what PhotoAcute (which does not support newer Pentax cameras) accomplishes.

Here's a 1:1 side-by-side of one of the original shots with the final composite:




The resultant composite can be used as a smart object or have adjustment layers applied to manipulate it however you want.
As you can see, with only three images, there is already a very substantial improvement in noise, comparable to what you'd see with in-camera PS. While the initial impression is a slight softening, the resulting image also responds much better to sharpening, curves adjustments, etc.

On the K-1, with its 4.4FPS full-resolution burst rate, this approach also potentially gives a slight improvement over the exposure intervals used in pixel shift. On something like the K-3II, with a much higher burst rate of 8.3FPS, you might see even more of an advantage. The tradeoff, of course, is added mirror slap/vibration between exposures. Whether the effect on image quality is balanced out by the image ghosting caused by the electronic shutter used during Pixel Shift is a matter of pure speculation without real-world testing.

Last edited by dcshooter; 08-14-2017 at 09:45 AM.
08-14-2017, 05:12 AM   #8
Site Supporter
Site Supporter
dakight's Avatar

Join Date: Sep 2014
Location: Phoenix, AZ
Photos: Albums
Posts: 1,184
The pixel shift algorhythm depends heavily on the camera and subject being absolutely still. As I understand it the motion correction is there to compensate for small amounts of movement within an otherwise motionless subject. One example might be leaves on a tree blowing in a slight breeze. The pixels are compared and some formula is used to determine what the final output should be. Using pixel shift on a handheld camera will likely give worse results that the usual single exposure.

08-14-2017, 05:54 AM   #9
Site Supporter
Site Supporter




Join Date: Mar 2009
Location: Gladys, Virginia
Photos: Gallery
Posts: 16,552
There is no way that hand held pixel shift would give any benefit over just a single shot, even with motion control feature. All the motion control does is masks out any places where there is movement between the frames, replacing it with the first image. In Raw Therapee, you can see how much motion there is and how much masking has to be done. I would guarantee that you would basically have to mask the whole image, leaving you with a single frame.

My experience with motion control in camera is that it still will produce artifacts if you have much movement (Raw Therapee is better). Anyway, there is probably a strong placebo effect here, where people believe pixel shift improves their image, even when it probably is exactly the same as a single image.
08-14-2017, 05:56 AM   #10
Pentaxian
dcshooter's Avatar

Join Date: Jan 2011
Location: Washington DC
Photos: Gallery | Albums
Posts: 4,699
QuoteOriginally posted by Rondec Quote
There is no way that hand held pixel shift would give any benefit over just a single shot, even with motion control feature. All the motion control does is masks out any places where there is movement between the frames, replacing it with the first image. In Raw Therapee, you can see how much motion there is and how much masking has to be done. I would guarantee that you would basically have to mask the whole image, leaving you with a single frame.

My experience with motion control in camera is that it still will produce artifacts if you have much movement (Raw Therapee is better). Anyway, there is probably a strong placebo effect here, where people believe pixel shift improves their image, even when it probably is exactly the same as a single image.
Yep. It's designed to keep areas with movement in a still frame sharp at the expense of image compositing in those areas, NOT compensate for shake in the whole frame.
08-14-2017, 08:37 AM - 1 Like   #11
Pentaxian
microlight's Avatar

Join Date: Sep 2011
Location: Hampshire, UK
Posts: 430
Brian's multi-exposure portrait process is a cool variant of the 'superresolution' method described on petapixel.com (search for 'A Practical Guide to Creating Superresolution Photos with Photoshop'), which actually relies on camera movement to simulate the sensor moving in between exposures. I haven't tried it (although have been meaning to), but have tried a similar stacking process where you can simulate the use of a ND filter for longer exposures in bright conditions, say for creating wave blur on a sea or river.

Last edited by microlight; 08-14-2017 at 08:46 AM.
08-14-2017, 08:46 AM - 2 Likes   #12
Site Supporter
Site Supporter




Join Date: Mar 2009
Location: Gladys, Virginia
Photos: Gallery
Posts: 16,552
QuoteOriginally posted by microlight Quote
Brian's multi-exposure portrait process is a cool variant of the 'superresolution' method described on petapixel.com (search for 'A Practical Guide to Creating Superresolution Photos with Photoshop'), which actually relies on camera movement to simulate the sensor moving in between exposures. I haven't tried it (although have been meaning to), but have tried a similar stacking process where you can simulate using a ND filter for longer exposures in bright conditions, say for creating wave blur on a sea or river.
I do wonder how useful that sort of "super resolution" would be.

The whole point of pixel shift seems to be to give a cleaner (less noise) file with better color data and less moire. When it works, it works well, but it doesn't give extra resolution -- which is fine for me, as having a high quality 36 megapixel file is good enough for me.

08-14-2017, 09:57 AM   #13
Loyal Site Supporter
Loyal Site Supporter




Join Date: Jan 2007
Location: Conroe, TX USA
Posts: 328
Unless you record true data no computer will give something real. Pixel shift and the Olympus algorithm use real data.

Pixel Shift doesn't increase the resolution passed the sensor sampling but it sure fills the spatial response closer to what can be done. Pixel Shift fills in 2/3 of the color information. A whole 66% without an interpolation!!! Even with the so called movement problem and a minimal compensation the data is better. Too bad the eye usually looks for anomalies rather than the loss of noise etc. What is truly hidden under the interpolation?

RONC

Last edited by rechmbrs; 08-14-2017 at 10:07 AM.
08-14-2017, 11:00 AM   #14
Administrator
Site Webmaster
Adam's Avatar

Join Date: Sep 2006
Location: Arizona
Photos: Gallery | Albums
Posts: 44,132
QuoteOriginally posted by BruceBanner Quote
Example; I could set the shutter speed to 1/1600 or 1/250, and the complete process (not counting the actual 'stitching and processing image part at the end') still felt the same.
There is about a 1/4s delay between the shots regardless of shutter speed, so there's not much that can be done about it.

Also with MC on, hand-held pixel shifting should essentially result in a single conventional image (it will cancel out the merge with the remaining 3 frames). Not something that is intended, nor should it (theoretically or practically) carry any benefit given the current implementation. As others have pointed out, manual stacking should prove to be a superior alternative.

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



PentaxForums.com server and development costs are user-supported. You can help cover these 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:

08-14-2017, 11:41 AM   #15
Pentaxian
dcshooter's Avatar

Join Date: Jan 2011
Location: Washington DC
Photos: Gallery | Albums
Posts: 4,699
QuoteOriginally posted by rechmbrs Quote
Unless you record true data no computer will give something real. Pixel shift and the Olympus algorithm use real data.

Pixel Shift doesn't increase the resolution passed the sensor sampling but it sure fills the spatial response closer to what can be done. Pixel Shift fills in 2/3 of the color information. A whole 66% without an interpolation!!! Even with the so called movement problem and a minimal compensation the data is better. Too bad the eye usually looks for anomalies rather than the loss of noise etc. What is truly hidden under the interpolation?

RONC
The bolded statement is absolutely meaningless if it's meant to prove the superiority of Pixel shift over other "superresolution" techniques.

Using multiple exposures and averaging pixel values over multiple aligned images gives you data just as "real" as using pixel shift. Or do you believe sampling error magically disappears just because the same photosite is used for the R, G, and B channels? If that were the case, Foveon sensors would be perfectly noise-free at any ISO value. And they are terrible at high ISO.

Increasing the number of sample events, regardless of whether they have been RGB interpolated prior to averaging as with the Photoshop composting method described above, will still give you resulting values that are closer to the "real" color value for each pixel, i.e. higher quality data. It's simple statistics.

The way the PS algorithm works, if motion is detected, for the exported image, it simply throws out 3 of the exposures for a given pixel and goes with the first one. For a hand-held series of exposures (i.e. the subject at hand), this can literally mean the entire image has its PS information discarded, leaving you with a giant file that is entirely interpolated, no better in quality than a non-PS image.
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!
camera, chance, color, dng, hand, image, iso, k-1, line, mc, method, photography, pixel, pixels, process, resolution, results, scene, shift, shutter, trade, vs
Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
hand-held shutter speed enyaw Troubleshooting and Beginner Help 20 04-23-2017 06:17 PM
Lake Union Fireworks - hand held Btparques Monthly Photo Contests 5 09-11-2016 04:01 AM
Pixel Shifting and Lightroom 6.3 RAW godavari Pentax DSLR Discussion 16 03-22-2016 10:51 AM
Pixel shifting tuggie76 Pentax K-3 17 01-27-2016 02:56 PM
Pixel Shifting & Virtual Tours RockvilleBob Digital Processing, Software, and Printing 8 01-26-2016 03:03 AM



All times are GMT -7. The time now is 06:26 PM. | 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