Web Analytics RiceHigh's Pentax Blog: My Pixel Shift Resolution Test

Sunday, May 22, 2016

My Pixel Shift Resolution Test

Below is the 200% centre crops without PSR and with. In-camera JPEG pics with firmware V1.10. Lens used is FA43 Limited at f/11, exposure 30s, ISO 100, shot with 2 sec self-timer enabled plus cable shutter release. Mirror lock up was disabled despite user manual said that it could be engaged. :-(


(Click to Enlarge)

For the resolution difference but the artifacts, I blog, you decide. :-)

Comments (16)

Loading... Logging you in...
  • Logged in as
There is a noticiable difference in resolution indeed!
About the artifacts it seems that is related to moving objects (water surface, boats). I don't know...but maybe it can be expected for oscillatory pattern movements with certain frequencies, like the water surface. As the sensor will see 4 different images taken with a (very short) time interval it could be related to the frequency of the movement with respect with the frequency of the 4 shoots....I suspect that its something like the "wagon-wheel effect"... (some times the wheel seems to be rotating clokwise, sometimes counter-clockwise, sometimes it appear to be stoped)
The software may solve some linear "predictable movement" but it appears that some movements may trigger some strange effects in the final blended image as your image show very well...
2 replies · active 460 weeks ago
Shooting the night scene (in long exposure) in PSR mode is not instantaneous, a 30s exposure will require 120s. I did not even check if dark frame subtraction was in action, for even longer time required.
You are right...every shot lasts 30 seconds. Anyway for images that show this kind of effects the final result regarding resolution could justify some post processing effort in photoshop even if blending the PSR image with the regular long exposure image is needed. The resolution improvement in the PSR image is notorious....I can read some words "Midland Concrete" in the place the regular long exposure shows some kind of texture in the building
Zos Xavius's avatar

Zos Xavius · 461 weeks ago

Are these out of camera jpegs? Using silkypix is supposed to work ok with movement. You can also take the two frames and just composite them. Easy. Supposedly lightroom still struggles with pixel shift.
3 replies · active 461 weeks ago
There is a mode with motion compensation that I tried but it failed to make a clear picture but will retry.
Zos Xavius's avatar

Zos Xavius · 461 weeks ago

your setup might have moved. you should be able to engage the mirror lockup if you use the timer. at least that is what I read. congrats on the K-1.
The user manual says it can, but in practice it can't, the option is disabled (dimmed).
honest opnion's avatar

honest opnion · 461 weeks ago

Top picture looks better, IMO. Bottom makes shadows overexposed. Look at water in front of boat. Looks terrible in bottom.
1 reply · active 461 weeks ago
Bottom pic has obviously increased resolution that the original pic lacks.
Btw, the starburst of my FA43 is beautiful, which I love. :-)
pentaxk1user's avatar

pentaxk1user · 460 weeks ago

Looks good, but I'm not having much luck with the pixel shift resolution myself. What am I doing wrong? The pixel shifted images are always less sharp. I used 2seconds shutterspeed, 2s delay, and a tripod. Is 2second shutter speed too slow if there is any possibility of wind?

Thanks,
Jan Pelcman's avatar

Jan Pelcman · 460 weeks ago

Did you use SilkyPix to process these images?
1 reply · active 460 weeks ago
No, in camera JPEGs only.
Use SilkyPix with MC, does what it says on the tin and works fine. For significant movement i've had no problems with this approach where I sometimes may get some small artifacting with SOOC jpeg. For slow moving and shorter shutter speeds SOOC jpeg with MC pixel shift works fine. For anything else i go to SilkyPix.
Speaking of shift, there's something Pentax could do to crush forever all competition :
- Sensor shift for architectural photography
- Sensor tilt for increased depth of field or funny effects

Honestly, as they already have 5 axis sensor moves, horizon ajustements and so on, this would be the logical move to make. It wouldn't take much on the sensor level to get the moves needed for some good Scheimpflug effects, to cross the floor plane...

And now Pentax could compete with my Toyo field :)
1 reply · active 408 weeks ago
The only question is lens coverage but then it is only a problem for shift, not for tilt...
So what do you think? Should we ask this next?

This would transform a great camera into a legendary camera :D

Post a new comment

Comments by