K3II: Pixel Shift First Test

bibz · 05-26-2015, 06:01 PM

The reason this works (from what I see) is that its taking better/more accurate readings from each pixel. Instead of algorithms producing the 'probably' colour its getting measured information at all zones. The fake 24mp images (in this case 16+8+8) are now true 24mp, with each pixel checked 4 times. Theres still quite some processing going on but the huge leap in data allows for much higher accuracy. The processing can be done in the cameras memory, the system only spitting out the corrected information, so no big file sizes. This is probably what needs to be done for DNG files as they have to comply with existing standards. The PEF can be a new sub-format and include all 4 raw shots of data for later processing allowing for development of both the blending technology and the full image data before output.

A very novel and very pentax approach to increasing the image quality of digital capture. Every camera (with ibis) should have this.

normhead · 05-26-2015, 06:31 PM

Originally posted by bibz

A very novel and very pentax approach to increasing the image quality of digital capture. Every camera (with ibis) should have this.

Can we see an example of how it works first? The examples i saw were slight improvements, but not what I would expect going to 36MP would be.

mee · 05-26-2015, 06:44 PM

Originally posted by Tjompen1968

That looks really nice. Some weird patterns on a few places but overall nice result.

Yes I think that is where there was a large amount of movement.. for instance on this image at the bottom, left corner right above the vent fan there is a pattern over a tree limb that appears to have been swaying in wind.

http://www.yaotomi.co.jp/blog/used/PENTAX_K-3II_1001.JPG

Interested in how this behave around moving water.. lake or river..

That and MACRO with no movement is going to give phenomenal results for crop body..

eraser · 05-26-2015, 06:55 PM

Just doing some playing around with pixel shift and the DNGs and PEFs appear to be the same size for me. Got some samples as soon as I find somewhere to upload the fullsize.
Also Lightroom 5.7 appears to read the DNGs just fine.
JPG off: 12.5MB
JPG on: 13.4MB
DNG: 129.1
PEF: 126.9

Last edited by eraser; 05-26-2015 at 08:50 PM.

JohnBee · 05-26-2015, 07:17 PM

Originally posted by eraser

Just doing some playing around with pixel shift and the DNGs and PEFs appear to be the same size for me. Got some samples as soon as I find somewhere to upload the fullsize.
Also Lightroom 5.7 appears to read the DNGs just fine.
JPG off: 12.5MB
JPG on: 13.4MB
DNG: 129.1
PEF: 126.9
Added some 100% crop examples, the compression of the forum uploader ruins image quality a bit I think.

hmmm... are you sure this is even working?
ie, I don't see any of the image characteristics demonstrated by Pentax in these samples whatsoever.
Perhaps they were sharpened?

eraser · 05-26-2015, 07:32 PM

Originally posted by JohnBee

hmmm... are you sure this is even working?
ie, I don't see any of the image characteristics demonstrated by Pentax in these samples whatsoever.
Perhaps they were sharpened?

I think it's working, the differences are very noticeable when viewed at 100%. Uploading to the forums killed the detail I think.

JohnBee · 05-26-2015, 07:39 PM

Originally posted by eraser

I think it's working, the differences are very noticeable when viewed at 100%. Uploading to the forums killed the detail I think.

Oh okay, that must be it. That said I'd love to see one of those up close.
Would you be interested in uploading a full sized, unprocessed image to my ftp/webserver?
Do you have and ISO100 sample?

If so, I'd be more than happy to host the full size image(s) on the forums afterwards.

eraser · 05-26-2015, 07:41 PM

Originally posted by JohnBee

Oh okay, that must be it. That said I'd love to see one of those up close.
Would you be interested in uploading a full sized, unprocessed image to my ftp/webserver?
And if so, I'd be more than happy to host the full size image on the forums afterwards.

Sure, I just shot three sets jpg/dng/pef samples. The first is up on dropbox, however long that will last.
Also down for request of whatever people want tested.
Edit: Will take an ISO 100 sample

JohnBee · 05-26-2015, 07:55 PM

Originally posted by eraser

Sure, I just shot three sets jpg/dng/pef samples. The first is up on dropbox, however long that will last.
Also down for request of whatever people want tested.
Edit: Will take an ISO 100 sample

Awesome!
PM sent

eraser · 05-26-2015, 08:36 PM

DeadJohn · 05-26-2015, 08:54 PM

Originally posted by mee

Yes I think that is where there was a large amount of movement.. for instance on this image at the bottom, left corner right above the vent fan there is a pattern over a tree limb that appears to have been swaying in wind.

http://www.yaotomi.co.jp/blog/used/PENTAX_K-3II_1001.JPG

Interested in how this behave around moving water.. lake or river..

That and MACRO with no movement is going to give phenomenal results for crop body..

That sample image is very informative. Look for things likely to move and you'll find artifacts when zoomed to 1:1., not just trees.

Check the red Ferris wheel visible behind the center building. Also follow the road and sidewalk. Many people on the sidewalk have pixellated ghosts. The ghosts of fast moving cars are less noticeable because the ghost is so far away from the main car, but you can see a lot near the center road (where distance and viewing angle have the car move fewer pixels across the frame).

---------- Post added 05-27-15 at 12:10 AM ----------

Originally posted by eraser

PEF and DNG are uploading, in the meantime some JPGs. Same procedure for all shots, first shot is PS off, AF for first then switched to MF for all following shots in that set with PS ON. Some full size samples (Thanks to JohnBee for hosting):

Thank you very much for sharing those.

The last pair looks like there might have been some tripod movement. http://199.168.184.48/eraser/IMGP0112.JPG was with pixel shift off and it looks like there's motion blur (or maybe I'm just looking at an area that's not in focus). http://199.168.184.48/eraser/IMGP0113.JPG is with pixel shift on, and shows artifacts in tree bark and other things that don't move with the wind.

Your other samples, though, reinforce that pixel shift is for printing not on screen viewing. Fairly low res viewing on a computer screen doesn't need pixel shift's increased resolution, and heavy screen cropping with pixel shift on will make the stacking artifacts obvious. Looking forward to what happens with sharpening. I suspect that our usual sharpening methods no longer apply.

eraser · 05-26-2015, 09:16 PM

Originally posted by DeadJohn

That sample image is very informative. Look for things likely to move and you'll find artifacts when zoomed to 1:1., not just trees.

Check the red Ferris wheel visible behind the center building. Also follow the road and sidewalk. Many people on the sidewalk have pixellated ghosts. The ghosts of fast moving cars are less noticeable because the ghost is so far away from the main car, but you can see a lot near the center road (where distance and viewing angle have the car move fewer pixels across the frame).

---------- Post added 05-27-15 at 12:10 AM ----------

Thank you very much for sharing those.

The last pair looks like there might have been some tripod movement. http://199.168.184.48/eraser/IMGP0112.JPG was with pixel shift off and it looks like there's motion blur (or maybe I'm just looking at an area that's not in focus). http://199.168.184.48/eraser/IMGP0113.JPG is with pixel shift on, and shows artifacts in tree bark and other things that don't move with the wind.

Your other samples, though, reinforce that pixel shift is for printing not on screen viewing. Fairly low res viewing on a computer screen doesn't need pixel shift's increased resolution, and cropping with pixel shift on will make the stacking artifacts more obvious. Looking forward to what happens with sharpening. I suspect that our usual sharpening methods no longer apply.

Looks like possibly some tripod movement yea.

Here is an example of an artifact I found in IMGP0107:

**jbinpg** · 05-26-2015, 10:49 PM

Thank you for the samples, Alex. The PS DNG from the first set with the fence is noticeably better in my raw developer than the matching JPG with PS off. Everything is crisper and the colours are not oversaturated and burnt out, e.g., the dandelion in the JPG. I am convinced that PS works as advertised. Full JPG export is 38.7 MB from the 129.8 MB DNG. I applied my standard sharpening and local contrast boost I use for my K-3 and, to be honest, the result is exceptional.

Alas, the PEF will not open in my raw developer of choice (darktable) - unknown format. I imagine it will be the same for other raw developers until an update to handle the new format is released.

Jack

Gimbal · 05-27-2015, 01:17 AM

Originally posted by jbinpg

Alas, the PEF will not open in my raw developer of choice (darktable) - unknown format. I imagine it will be the same for other raw developers until an update to handle the new format is released.
Jack

Wait a minute, are you saying that you can develop the DNG-file in a non Pentax program?

**jbinpg** · 05-27-2015, 01:27 AM

Yes, sir. Loads just like a regular DNG.

05-26-2015, 07:17 PM	#50
JohnBee Veteran Member Join Date: Sep 2008 Location: Newrfoundland Photos: Albums Posts: 4,667	Originally posted by eraser Just doing some playing around with pixel shift and the DNGs and PEFs appear to be the same size for me. Got some samples as soon as I find somewhere to upload the fullsize. Also Lightroom 5.7 appears to read the DNGs just fine. JPG off: 12.5MB JPG on: 13.4MB DNG: 129.1 PEF: 126.9 Added some 100% crop examples, the compression of the forum uploader ruins image quality a bit I think. hmmm... are you sure this is even working? ie, I don't see any of the image characteristics demonstrated by Pentax in these samples whatsoever. Perhaps they were sharpened? Last edited by JohnBee; 05-26-2015 at 07:25 PM.

05-26-2015, 08:36 PM	#55
eraser Junior Member Join Date: Jun 2007 Location: Victoria, BC Posts: 32	PEF and DNG are uploading, in the meantime some JPGs. Same procedure for all shots, first shot is PS off, AF for first then switched to MF for all following shots in that set with PS ON. Some full size samples (Thanks to JohnBee for hosting): http://199.168.184.48/eraser/IMGP0090.JPG http://199.168.184.48/eraser/IMGP0091.JPG http://199.168.184.48/eraser/IMGP0092.DNG http://199.168.184.48/eraser/IMGP0093.PEF http://199.168.184.48/eraser/IMGP0097.JPG http://199.168.184.48/eraser/IMGP0098.JPG http://199.168.184.48/eraser/IMGP0099.DNG http://199.168.184.48/eraser/IMGP0100.PEF http://199.168.184.48/eraser/IMGP0102.JPG http://199.168.184.48/eraser/IMGP0103.JPG http://199.168.184.48/eraser/IMGP0104.DNG http://199.168.184.48/eraser/IMGP0105.PEF http://199.168.184.48/eraser/IMGP0106.JPG http://199.168.184.48/eraser/IMGP0107.JPG http://199.168.184.48/eraser/IMGP0108.DNG http://199.168.184.48/eraser/IMGP0109.PEF http://199.168.184.48/eraser/IMGP0112.JPG http://199.168.184.48/eraser/IMGP0113.JPG Last edited by eraser; 05-27-2015 at 06:23 AM.

05-26-2015, 08:54 PM	#56
DeadJohn Pentaxian Join Date: Jan 2011 Location: New York Posts: 4,833	Originally posted by mee Yes I think that is where there was a large amount of movement.. for instance on this image at the bottom, left corner right above the vent fan there is a pattern over a tree limb that appears to have been swaying in wind. http://www.yaotomi.co.jp/blog/used/PENTAX_K-3II_1001.JPG Interested in how this behave around moving water.. lake or river.. That and MACRO with no movement is going to give phenomenal results for crop body.. That sample image is very informative. Look for things likely to move and you'll find artifacts when zoomed to 1:1., not just trees. Check the red Ferris wheel visible behind the center building. Also follow the road and sidewalk. Many people on the sidewalk have pixellated ghosts. The ghosts of fast moving cars are less noticeable because the ghost is so far away from the main car, but you can see a lot near the center road (where distance and viewing angle have the car move fewer pixels across the frame). ---------- Post added 05-27-15 at 12:10 AM ---------- Originally posted by eraser PEF and DNG are uploading, in the meantime some JPGs. Same procedure for all shots, first shot is PS off, AF for first then switched to MF for all following shots in that set with PS ON. Some full size samples (Thanks to JohnBee for hosting): Thank you very much for sharing those. The last pair looks like there might have been some tripod movement. http://199.168.184.48/eraser/IMGP0112.JPG was with pixel shift off and it looks like there's motion blur (or maybe I'm just looking at an area that's not in focus). http://199.168.184.48/eraser/IMGP0113.JPG is with pixel shift on, and shows artifacts in tree bark and other things that don't move with the wind. Your other samples, though, reinforce that pixel shift is for printing not on screen viewing. Fairly low res viewing on a computer screen doesn't need pixel shift's increased resolution, and heavy screen cropping with pixel shift on will make the stacking artifacts obvious. Looking forward to what happens with sharpening. I suspect that our usual sharpening methods no longer apply. Last edited by DeadJohn; 05-26-2015 at 09:14 PM.

05-26-2015, 09:16 PM	#57
eraser Junior Member Join Date: Jun 2007 Location: Victoria, BC Posts: 32	Originally posted by DeadJohn That sample image is very informative. Look for things likely to move and you'll find artifacts when zoomed to 1:1., not just trees. Check the red Ferris wheel visible behind the center building. Also follow the road and sidewalk. Many people on the sidewalk have pixellated ghosts. The ghosts of fast moving cars are less noticeable because the ghost is so far away from the main car, but you can see a lot near the center road (where distance and viewing angle have the car move fewer pixels across the frame). ---------- Post added 05-27-15 at 12:10 AM ---------- Thank you very much for sharing those. The last pair looks like there might have been some tripod movement. http://199.168.184.48/eraser/IMGP0112.JPG was with pixel shift off and it looks like there's motion blur (or maybe I'm just looking at an area that's not in focus). http://199.168.184.48/eraser/IMGP0113.JPG is with pixel shift on, and shows artifacts in tree bark and other things that don't move with the wind. Your other samples, though, reinforce that pixel shift is for printing not on screen viewing. Fairly low res viewing on a computer screen doesn't need pixel shift's increased resolution, and cropping with pixel shift on will make the stacking artifacts more obvious. Looking forward to what happens with sharpening. I suspect that our usual sharpening methods no longer apply. Looks like possibly some tripod movement yea. Here is an example of an artifact I found in IMGP0107: Attached Images

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
SR and Pixel Shift Technology	C_Jones	Pentax K-3 & K-3 II	9	05-10-2015 06:24 AM
Pixel Shifting Samples from Pentax on the K3II	mikeodial	Pentax DSLR Discussion	99	04-26-2015 05:00 PM
Pixel shift practical applications	Spodeworld	Pentax K-3 & K-3 II	20	04-25-2015 10:01 AM
K-3 II pixel shift resized. Wow factor.	JimmyDranox	Pentax K-3 & K-3 II	21	04-25-2015 09:37 AM
Pixel shift and bracketing	Spodeworld	Pentax K-3 & K-3 II	9	04-23-2015 08:56 PM

05-26-2015, 06:01 PM	#46
bibz Senior Member Join Date: Nov 2009 Location: Newcastle, AU! Posts: 276	The reason this works (from what I see) is that its taking better/more accurate readings from each pixel. Instead of algorithms producing the 'probably' colour its getting measured information at all zones. The fake 24mp images (in this case 16+8+8) are now true 24mp, with each pixel checked 4 times. Theres still quite some processing going on but the huge leap in data allows for much higher accuracy. The processing can be done in the cameras memory, the system only spitting out the corrected information, so no big file sizes. This is probably what needs to be done for DNG files as they have to comply with existing standards. The PEF can be a new sub-format and include all 4 raw shots of data for later processing allowing for development of both the blending technology and the full image data before output. A very novel and very pentax approach to increasing the image quality of digital capture. Every camera (with ibis) should have this.

05-26-2015, 06:31 PM	#47
normhead Pentaxian Join Date: Jun 2007 Location: Near Algonquin Park Photos: Gallery \| Albums Posts: 40,451	Originally posted by bibz A very novel and very pentax approach to increasing the image quality of digital capture. Every camera (with ibis) should have this. Can we see an example of how it works first? The examples i saw were slight improvements, but not what I would expect going to 36MP would be.

05-26-2015, 06:44 PM	#48
mee Veteran Member Join Date: May 2010 Posts: 7,403	Originally posted by Tjompen1968 That looks really nice. Some weird patterns on a few places but overall nice result. Yes I think that is where there was a large amount of movement.. for instance on this image at the bottom, left corner right above the vent fan there is a pattern over a tree limb that appears to have been swaying in wind. http://www.yaotomi.co.jp/blog/used/PENTAX_K-3II_1001.JPG Interested in how this behave around moving water.. lake or river.. That and MACRO with no movement is going to give phenomenal results for crop body..

05-26-2015, 06:55 PM	#49
eraser Junior Member Join Date: Jun 2007 Location: Victoria, BC Posts: 32	Just doing some playing around with pixel shift and the DNGs and PEFs appear to be the same size for me. Got some samples as soon as I find somewhere to upload the fullsize. Also Lightroom 5.7 appears to read the DNGs just fine. JPG off: 12.5MB JPG on: 13.4MB DNG: 129.1 PEF: 126.9 Last edited by eraser; 05-26-2015 at 08:50 PM.

05-26-2015, 07:32 PM	#51
eraser Junior Member Join Date: Jun 2007 Location: Victoria, BC Posts: 32	Originally posted by JohnBee hmmm... are you sure this is even working? ie, I don't see any of the image characteristics demonstrated by Pentax in these samples whatsoever. Perhaps they were sharpened? I think it's working, the differences are very noticeable when viewed at 100%. Uploading to the forums killed the detail I think.

05-26-2015, 07:39 PM	#52
JohnBee Veteran Member Join Date: Sep 2008 Location: Newrfoundland Photos: Albums Posts: 4,667	Originally posted by eraser I think it's working, the differences are very noticeable when viewed at 100%. Uploading to the forums killed the detail I think. Oh okay, that must be it. That said I'd love to see one of those up close. Would you be interested in uploading a full sized, unprocessed image to my ftp/webserver? Do you have and ISO100 sample? If so, I'd be more than happy to host the full size image(s) on the forums afterwards.

05-26-2015, 07:41 PM	#53
eraser Junior Member Join Date: Jun 2007 Location: Victoria, BC Posts: 32	Originally posted by JohnBee Oh okay, that must be it. That said I'd love to see one of those up close. Would you be interested in uploading a full sized, unprocessed image to my ftp/webserver? And if so, I'd be more than happy to host the full size image on the forums afterwards. Sure, I just shot three sets jpg/dng/pef samples. The first is up on dropbox, however long that will last. Also down for request of whatever people want tested. Edit: Will take an ISO 100 sample

05-26-2015, 07:55 PM	#54
JohnBee Veteran Member Join Date: Sep 2008 Location: Newrfoundland Photos: Albums Posts: 4,667	Originally posted by eraser Sure, I just shot three sets jpg/dng/pef samples. The first is up on dropbox, however long that will last. Also down for request of whatever people want tested. Edit: Will take an ISO 100 sample Awesome! PM sent

05-26-2015, 10:49 PM	#58
jbinpg Site Supporter Join Date: Mar 2008 Location: Prince George, BC Photos: Gallery \| Albums Posts: 3,546	Thank you for the samples, Alex. The PS DNG from the first set with the fence is noticeably better in my raw developer than the matching JPG with PS off. Everything is crisper and the colours are not oversaturated and burnt out, e.g., the dandelion in the JPG. I am convinced that PS works as advertised. Full JPG export is 38.7 MB from the 129.8 MB DNG. I applied my standard sharpening and local contrast boost I use for my K-3 and, to be honest, the result is exceptional. Alas, the PEF will not open in my raw developer of choice (darktable) - unknown format. I imagine it will be the same for other raw developers until an update to handle the new format is released. Jack

05-27-2015, 01:17 AM	#59
Gimbal Pentaxian Join Date: Apr 2007 Location: Sweden Posts: 2,106	Originally posted by jbinpg Alas, the PEF will not open in my raw developer of choice (darktable) - unknown format. I imagine it will be the same for other raw developers until an update to handle the new format is released. Jack Wait a minute, are you saying that you can develop the DNG-file in a non Pentax program?

05-27-2015, 01:27 AM	#60
jbinpg Site Supporter Join Date: Mar 2008 Location: Prince George, BC Photos: Gallery \| Albums Posts: 3,546	Yes, sir. Loads just like a regular DNG.