Pixel Shifted Images

**Rondec** · 05-04-2019, 07:30 AM - **2 Likes**

Tree Before Dawn (DFA 24-70/pixel shift)

**Rondec** · 05-08-2019, 02:49 AM - **3 Likes**

It is strawberry season again (DFA 100 macro/pixel shift)

**Rondec** · 05-17-2019, 09:15 AM - **2 Likes**

Cattle in a Field (DA *55/pixel shift)

**BruceBanner** · 05-17-2019, 05:50 PM - **1 Like**

Originally posted by Rondec

Cattle in a Field (DA *55/pixel shift)

How do you handle the motion correction for your work? I'm imagining with a scene like this that the PS stuff wouldn't have been pushed through for a lot of the cattle and grass? I use RawTherapee for my motion correction and show the Motion Mask, and it gives me an idea of how much of the is not going to be pixelshifted etc.

This shot for example;

23mm/f8, DA 20-40 on the K-1 (ff mode)

a lot of the grass was ok, even some of the tree, but quite a few of the leaves were moving and thus a good 50% of the tree is not really pixelshifted.

**Rondec** · 05-18-2019, 05:31 AM

Originally posted by BruceBanner

How do you handle the motion correction for your work? I'm imagining with a scene like this that the PS stuff wouldn't have been pushed through for a lot of the cattle and grass? I use RawTherapee for my motion correction and show the Motion Mask, and it gives me an idea of how much of the is not going to be pixelshifted etc.

This shot for example;

23mm/f8, DA 20-40 on the K-1 (ff mode)

a lot of the grass was ok, even some of the tree, but quite a few of the leaves were moving and thus a good 50% of the tree is not really pixelshifted.

I use Raw Therapee too. In the above image, probably 75 percent was pixel shifted, based on the mask, and 25 percent wasn't. As you say, if the majority of the image is masked, you might just as well skip the pixel shifting process.

**BruceBanner** · 05-18-2019, 03:00 PM

Originally posted by Rondec

I use Raw Therapee too. In the above image, probably 75 percent was pixel shifted, based on the mask, and 25 percent wasn't. As you say, if the majority of the image is masked, you might just as well skip the pixel shifting process.

That's pretty good for a scene like that to get 75%, I would have thought the cow/s would sway or move just a tad and the grass etc, just goes to show, cattle me like stone! haha.

Sometimes when I review my shots and see motion on things that really shouldn't have had any motion at all I wonder what's going, tripod not tight enough, a bit of wind etc, the clunking of the shutter even though it's being handled via ES?

**Rondec** · 05-19-2019, 05:42 AM - **6 Likes**

Broken Down Shed at Sunrise (DFA 24-70mm/pixel shift)

house · 05-19-2019, 01:20 PM - **2 Likes**

Depending on the scene it can be useful to turn off "fill holes" in the pixelshift settings. By default (quite reasonably I think) the algorithm fills holes in the mask exempting more area from pixelshift. Try in for your scenes to see what you think. It's under raw tab > demosaicing > Motion correction: custom > uncheck "fill holes in motion mask"

By switching to custom there's lots of knobs to turn and switches to flick if you really want to squeeze the last shift out of those pixels. I think the defaults are generally quite good but its worth fiddling around a bit.

**BruceBanner** · 05-19-2019, 08:05 PM

Originally posted by house

Depending on the scene it can be useful to turn off "fill holes" in the pixelshift settings. By default (quite reasonably I think) the algorithm fills holes in the mask exempting more area from pixelshift. Try in for your scenes to see what you think. It's under raw tab > demosaicing > Motion correction: custom > uncheck "fill holes in motion mask"

By switching to custom there's lots of knobs to turn and switches to flick if you really want to squeeze the last shift out of those pixels. I think the defaults are generally quite good but its worth fiddling around a bit.

Thanks house I'll check that out.

**tduell** · 05-19-2019, 09:12 PM

This may be of interest.
You can pass your pixelshifted RAW to dcrawps and ask it to mark PS artefacts in the output...you can choose the % threshold.
dcrawps looks at the differences in the same pixels in the two green channels.
The switches of interest are as follows...

-G 1-99. 1 detects the lightest artefacts (1% change in green level). 99 detects only the largest artefacts (99% change). Default is 100, artefact detection/repair disabled.
-X Highlight artefacts in the output. If -G is not set, it uses 20%.

Cheers,
Terry

**BruceBanner** · 05-20-2019, 12:40 AM - **8 Likes**

The Kid's Kupcake

(100% made by my youngest).

Image taken with a Lensbaby Velvet 56, composite of 3 varying apertures, pixelshifted, processed using RNI.

Gimbal · 05-20-2019, 03:38 AM - **1 Like**

Originally posted by BruceBanner

Sometimes when I review my shots and see motion on things that really shouldn't have had any motion at all I wonder what's going, tripod not tight enough, a bit of wind etc, the clunking of the shutter even though it's being handled via ES?

Details in the scene doesn’t necessarily have to move to trigger the motion correction algorithms. If the light changes even the slightest during the exposure you will also get “motion” detected due to variances in the images. Luckily (as already mentioned ) you can tweak the settings quite a lot in RT.

biz-engineer · 05-20-2019, 04:03 AM - **1 Like**

Originally posted by BruceBanner

Sometimes when I review my shots and see motion on things that really shouldn't have had any motion at all I wonder what's going, tripod not tight enough, a bit of wind etc, the clunking of the shutter even though it's being handled via ES?

I don't know how concrete is the floor where you install your tripod, but when I take a step back from the camera I could notice camera motion with my 150450@450 zoomed in live view, my own weight slightly changed the floor. So when I take pixel shifted shots of a still subject, I focus then go on the other side of the room and use the IR remote to trigger the PS shot, then I could observe that PS worked as expected with or without MC.

MarkJerling · 05-20-2019, 04:09 AM

Originally posted by BruceBanner

The Kid's Kupcake

(100% made by my youngest).

Image taken with a Lensbaby Velvet 56, composite of 3 varying apertures, pixelshifted, processed using RNI.

Great shot! It looks so real I want to grab it right off the screen!

**BruceBanner** · 05-20-2019, 04:19 AM - **1 Like**

Originally posted by Gimbal

Details in the scene doesn’t necessarily have to move to trigger the motion correction algorithms. If the light changes even the slightest during the exposure you will also get “motion” detected due to variances in the images. Luckily (as already mentioned ) you can tweak the settings quite a lot in RT.

Originally posted by biz-engineer

I don't know how concrete is the floor where you install your tripod, but when I take a step back from the camera I could notice camera motion with my 150450@450 zoomed in live view, my own weight slightly changed the floor. So when I take pixel shifted shots of a still subject, I focus then go on the other side of the room and use the IR remote to trigger the PS shot, then I could observe that PS worked as expected with or without MC.

Good points to both of you, I suspect it could be one or the other, or a combination of the two! In times gone by I did tell the kids to "FREEZE!" if they came into the room I was shooting a longer exposed PS shot hahaha, I am more the wiser now and prioritise a lot of these shots when they're in school

Originally posted by MarkJerling

Great shot! It looks so real I want to grab it right off the screen!

Thanks, yes this one came out really good I think, quite chuffed with it I am (and of course my 9yr old kids piping and decoration efforts!).

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Pixel Shifted Snoqualie Falls using 24-70mm Pentax	AwesomeIan	Pentax K-1 & K-1 II	20	07-01-2016 09:32 AM
How to display pixel shift images.	TroutHunterJohn	Pentax K-3 & K-3 II	8	11-10-2015 07:58 AM
Sensor shifted composite shot of 40MP resolution	MichalW	Pentax DSLR Discussion	23	01-17-2015 03:44 AM
Shifted from Canon to Pentax	ruwanal62	Welcomes and Introductions	5	03-15-2014 11:06 AM
Image count on K200D has shifted to a high number	Stratario	Pentax DSLR Discussion	6	06-29-2008 05:52 PM

05-04-2019, 07:30 AM - 2 Likes	#1006
Rondec Loyal Site Supporter Join Date: Mar 2009 Location: Gladys, Virginia Photos: Gallery Posts: 27,653 Original Poster	Tree Before Dawn (DFA 24-70/pixel shift)

05-08-2019, 02:49 AM - 3 Likes	#1007
Rondec Loyal Site Supporter Join Date: Mar 2009 Location: Gladys, Virginia Photos: Gallery Posts: 27,653 Original Poster	It is strawberry season again (DFA 100 macro/pixel shift)

05-17-2019, 09:15 AM - 2 Likes	#1008
Rondec Loyal Site Supporter Join Date: Mar 2009 Location: Gladys, Virginia Photos: Gallery Posts: 27,653 Original Poster	Cattle in a Field (DA *55/pixel shift)

05-17-2019, 05:50 PM - 1 Like	#1009
BruceBanner Site Supporter Join Date: Dec 2015 Posts: 5,404	Originally posted by Rondec Cattle in a Field (DA *55/pixel shift) How do you handle the motion correction for your work? I'm imagining with a scene like this that the PS stuff wouldn't have been pushed through for a lot of the cattle and grass? I use RawTherapee for my motion correction and show the Motion Mask, and it gives me an idea of how much of the is not going to be pixelshifted etc. This shot for example; 23mm/f8, DA 20-40 on the K-1 (ff mode) a lot of the grass was ok, even some of the tree, but quite a few of the leaves were moving and thus a good 50% of the tree is not really pixelshifted.

05-18-2019, 05:31 AM	#1010
Rondec Loyal Site Supporter Join Date: Mar 2009 Location: Gladys, Virginia Photos: Gallery Posts: 27,653 Original Poster	Originally posted by BruceBanner How do you handle the motion correction for your work? I'm imagining with a scene like this that the PS stuff wouldn't have been pushed through for a lot of the cattle and grass? I use RawTherapee for my motion correction and show the Motion Mask, and it gives me an idea of how much of the is not going to be pixelshifted etc. This shot for example; 23mm/f8, DA 20-40 on the K-1 (ff mode) a lot of the grass was ok, even some of the tree, but quite a few of the leaves were moving and thus a good 50% of the tree is not really pixelshifted. I use Raw Therapee too. In the above image, probably 75 percent was pixel shifted, based on the mask, and 25 percent wasn't. As you say, if the majority of the image is masked, you might just as well skip the pixel shifting process.

05-18-2019, 03:00 PM	#1011
BruceBanner Site Supporter Join Date: Dec 2015 Posts: 5,404	Originally posted by Rondec I use Raw Therapee too. In the above image, probably 75 percent was pixel shifted, based on the mask, and 25 percent wasn't. As you say, if the majority of the image is masked, you might just as well skip the pixel shifting process. That's pretty good for a scene like that to get 75%, I would have thought the cow/s would sway or move just a tad and the grass etc, just goes to show, cattle me like stone! haha. Sometimes when I review my shots and see motion on things that really shouldn't have had any motion at all I wonder what's going, tripod not tight enough, a bit of wind etc, the clunking of the shutter even though it's being handled via ES?

05-19-2019, 05:42 AM - 6 Likes	#1012
Rondec Loyal Site Supporter Join Date: Mar 2009 Location: Gladys, Virginia Photos: Gallery Posts: 27,653 Original Poster	Broken Down Shed at Sunrise (DFA 24-70mm/pixel shift)

05-19-2019, 01:20 PM - 2 Likes	#1013
house Pentaxian Join Date: May 2015 Photos: Albums Posts: 3,306	Depending on the scene it can be useful to turn off "fill holes" in the pixelshift settings. By default (quite reasonably I think) the algorithm fills holes in the mask exempting more area from pixelshift. Try in for your scenes to see what you think. It's under raw tab > demosaicing > Motion correction: custom > uncheck "fill holes in motion mask" By switching to custom there's lots of knobs to turn and switches to flick if you really want to squeeze the last shift out of those pixels. I think the defaults are generally quite good but its worth fiddling around a bit.

05-19-2019, 08:05 PM	#1014
BruceBanner Site Supporter Join Date: Dec 2015 Posts: 5,404	Originally posted by house Depending on the scene it can be useful to turn off "fill holes" in the pixelshift settings. By default (quite reasonably I think) the algorithm fills holes in the mask exempting more area from pixelshift. Try in for your scenes to see what you think. It's under raw tab > demosaicing > Motion correction: custom > uncheck "fill holes in motion mask" By switching to custom there's lots of knobs to turn and switches to flick if you really want to squeeze the last shift out of those pixels. I think the defaults are generally quite good but its worth fiddling around a bit. Thanks house I'll check that out.

05-19-2019, 09:12 PM	#1015
tduell Loyal Site Supporter Join Date: May 2015 Location: Melbourne Posts: 1,514	This may be of interest. You can pass your pixelshifted RAW to dcrawps and ask it to mark PS artefacts in the output...you can choose the % threshold. dcrawps looks at the differences in the same pixels in the two green channels. The switches of interest are as follows... -G 1-99. 1 detects the lightest artefacts (1% change in green level). 99 detects only the largest artefacts (99% change). Default is 100, artefact detection/repair disabled. -X Highlight artefacts in the output. If -G is not set, it uses 20%. Cheers, Terry

05-20-2019, 12:40 AM - 8 Likes	#1016
BruceBanner Site Supporter Join Date: Dec 2015 Posts: 5,404	The Kid's Kupcake (100% made by my youngest). Image taken with a Lensbaby Velvet 56, composite of 3 varying apertures, pixelshifted, processed using RNI.

05-20-2019, 03:38 AM - 1 Like	#1017
Gimbal Pentaxian Join Date: Apr 2007 Location: Sweden Posts: 2,106	Originally posted by BruceBanner Sometimes when I review my shots and see motion on things that really shouldn't have had any motion at all I wonder what's going, tripod not tight enough, a bit of wind etc, the clunking of the shutter even though it's being handled via ES? Details in the scene doesn’t necessarily have to move to trigger the motion correction algorithms. If the light changes even the slightest during the exposure you will also get “motion” detected due to variances in the images. Luckily (as already mentioned ) you can tweak the settings quite a lot in RT.

05-20-2019, 04:03 AM - 1 Like	#1018
biz-engineer Pentaxian Join Date: Feb 2015 Photos: Gallery Posts: 12,232	Originally posted by BruceBanner Sometimes when I review my shots and see motion on things that really shouldn't have had any motion at all I wonder what's going, tripod not tight enough, a bit of wind etc, the clunking of the shutter even though it's being handled via ES? I don't know how concrete is the floor where you install your tripod, but when I take a step back from the camera I could notice camera motion with my 150450@450 zoomed in live view, my own weight slightly changed the floor. So when I take pixel shifted shots of a still subject, I focus then go on the other side of the room and use the IR remote to trigger the PS shot, then I could observe that PS worked as expected with or without MC.

05-20-2019, 04:09 AM	#1019
MarkJerling Moderator Site Supporter Join Date: May 2012 Location: Wairarapa, New Zealand Photos: Gallery \| Albums Posts: 20,406	Originally posted by BruceBanner The Kid's Kupcake (100% made by my youngest). Image taken with a Lensbaby Velvet 56, composite of 3 varying apertures, pixelshifted, processed using RNI. Great shot! It looks so real I want to grab it right off the screen!

05-20-2019, 04:19 AM - 1 Like	#1020
BruceBanner Site Supporter Join Date: Dec 2015 Posts: 5,404	Originally posted by Gimbal Details in the scene doesn’t necessarily have to move to trigger the motion correction algorithms. If the light changes even the slightest during the exposure you will also get “motion” detected due to variances in the images. Luckily (as already mentioned ) you can tweak the settings quite a lot in RT. Originally posted by biz-engineer I don't know how concrete is the floor where you install your tripod, but when I take a step back from the camera I could notice camera motion with my 150450@450 zoomed in live view, my own weight slightly changed the floor. So when I take pixel shifted shots of a still subject, I focus then go on the other side of the room and use the IR remote to trigger the PS shot, then I could observe that PS worked as expected with or without MC. Good points to both of you, I suspect it could be one or the other, or a combination of the two! In times gone by I did tell the kids to "FREEZE!" if they came into the room I was shooting a longer exposed PS shot hahaha, I am more the wiser now and prioritise a lot of these shots when they're in school Originally posted by MarkJerling Great shot! It looks so real I want to grab it right off the screen! Thanks, yes this one came out really good I think, quite chuffed with it I am (and of course my 9yr old kids piping and decoration efforts!).