Diffraction Compensation - What and How?

rawr · 07-08-2014, 06:39 PM

Originally posted by normhead

the "unique thing" Pentax is going to introduce with it's FF

But Sony is already doing it on FF with the A7r. And Pentax are already doing it on the 645Z. I don't think it is news or a big deal.

Unless I am reading the benefits of this technology entirely wrong, diffraction compensation might be buried somewhere as dot-point number 17 on a Pentax marketing brochure for FF, but won't be the sort of thing to sway buyers: 'Gee, the Pentax K-1 has diffraction compensation for selected lenses! I'm not going to buy a Nikon D610 now - I'll buy the Pentax!'.

normhead · 07-08-2014, 07:31 PM

Originally posted by rawr

But Sony is already doing it on FF with the A7r. And Pentax are already doing it on the 645Z. I don't think it is news or a big deal.

Unless I am reading the benefits of this technology entirely wrong, diffraction compensation might be buried somewhere as dot-point number 17 on a Pentax marketing brochure for FF, but won't be the sort of thing to sway buyers: 'Gee, the Pentax K-1 has diffraction compensation for selected lenses! I'm not going to buy a Nikon D610 now - I'll buy the Pentax!'.

If it's simply sharpening yes, if it actually better defines diffraction softened objects, it could be major.

Imageman · 07-09-2014, 09:22 PM

Its conclusive, Pentax diffraction compensation is as I suggested, simply sharpening the image using a well know and used algorithm such as unsharp mask.

Quote -

The main new feature introduced via this firmware update (for k3) is Diffraction Correction, an addition to the K-3's existing suite of Lens Correction options. This feature was originally introduced in the Pentax 645Z. Diffraction correction attempts to compensate for the degradation in image quality that occurs at small aperture settings, such as F11 or F16. Like the other Lens Correction settings, diffraction correction only applies to JPEG files and it is only available with Pentax DA, DA L, D-FA, and FA Limited lenses.
Note that on the Pentax K-3, the diffraction correction option is only accessible via the first tab of the main menu; it is not being added to the "control panel" quick-access screen as on the 645Z.
After performing some quick image tests, we discovered that the effect of this new feature is essentially an unsharp mask (a sharpening algorithm).

rawr · 07-09-2014, 09:53 PM

Originally posted by Imageman

Its conclusive, Pentax diffraction compensation is as I suggested, simply sharpening the image using a well know and used algorithm such as unsharp mask.

Maybe. But one should be able to test 'simple' USM vs deconvolution sharpening, to see what is at work with the Pentax solution.

Could be done by shooting in RAW+ at f16-f22 eg a newpaper page or another suitable subject, then processing the RAW using (a) normal USM sharpening, as seen in Lightroom or DxO; and (b) a deconvolution tool like Focus Magic or Topaz InFocus or AstraImage, and then comparing (a) and (b) with what the in-camera JPG looks like when diffraction compensation is turned on ... Maybe when I have the time I'll do some comparisons like that.

Last edited by rawr; 07-09-2014 at 10:45 PM.

**Digitalis** · 07-10-2014, 12:37 AM

Originally posted by Imageman

Its conclusive, Pentax diffraction compensation is as I suggested, simply sharpening the image using a well know and used algorithm such as unsharp mask.

I f*cking said so. The sad part is that it is also affecting RAW files as well.

rawr · 07-10-2014, 01:24 AM

Originally posted by Digitalis

The sad part is that it is also affecting RAW files as well.

Perhaps not. As I understand them, the PDF notes issued with the firmware update, seem to relate mainly to the in-camera processing of images. They say:

'Diffraction Correction is not on RAW Development items.
(Setting is reflected when shooting image but it cannot be changed)'

which I take to mean that diffraction correction is both (1) not available on the in-camera RAW development menu options, and (2) diffraction correction shows up on the review screen when shooting an image because the image displayed is the JPG preview baked into all RAW's, not the state of the actual RAW.

The fact that the notes also say that the corrections don't apply if the lenses aren't Pentax, or if a teleconverter is used, also strongly suggests to me that diffraction correction isn't applied to RAW's automatically, but is only an option.

It will be interesting to see what new fields show up in the K-3 EXIF that relate to this function once EXIFTOOL has been updated.

Stone G. · 07-10-2014, 01:44 PM

I won't discuss the value, if any, of this upgrade. Just curious to see the effect of this "compensation-on-demand". As others have pointed out, this doesn't change the raw files; it only works in-camera and only for Pentax lenses that the camera will reckognize.

Here's my test target: A folding curtain some 3 meters away. AA-filter simulation was turned off with these images (in fact, I don't think I ever have it on):

Test Target (Downsized DA 18-135 @ 135mm full image)

Then some 100% non-resized comparisons with Compensation off to the left and on to the right. First my DA*200 lens:

Pentax-DA* @ f/22

Not much of a difference to my old eyes and no need to show comparisons at larger apertures. Quite different with the DA 18-135 lens:

Pentax-DA 18-135 lens@ 135mm and f/5.6

Pentax-DA 18-135 lens@ 135mm and f/11

Pentax-DA 18-135 lens@ 135mm and f/22

Pentax-DA 18-135 lens@ 135mm and f/32

You may judge for yourself, but there is an effect and it does somehow seem to differentiate between different lenses at sam f-number.

Na Horuk · 07-10-2014, 02:31 PM

Originally posted by Stone G.

You may judge for yourself, but there is an effect and it does somehow seem to differentiate between different lenses at sam f-number.

Yeah, its not super advanced deconvolution, but it is lens-specific sharpening profiles, which is still better than a one-size-fits-all sharpening solution. I didn't notice and odd artefacts caused by this feature, so its probably ok for the jpeg-shooters.

Imageman · 07-12-2014, 09:31 AM

The fact is its an unsharp mask however you slice it.

Manufacturers are businessmen and they are not in the business of pushing boundries unless theres a lot of profit in it. This was never going to be anything other than a simple cheap sharpening which any competent photographer who can sharpen in post processing can do already and doesn't need.

Its for jpeg shooters who never sharpen or do anything, its for holiday snaps, and its a marketing exercise so as not to miss out on the latest wrinkle that the competitors offer.

Its about looking good in the shop so the salesmen have something else to show customers and its about keeping market share at zero or little cost to the manufacturer.

Don't assume Pentax want to push boundaries in research and development. make no mistake, if Pentax could make more money selling 6 megapixel point and shoots instead of dSLRs with all their associated tooling and research plus lens design plus accessories, they would drop dSLR production faster that a dog drops a bone.

Profit and cheap production is all that matters.

Ive said as much all along.

wombat2go · 07-12-2014, 09:51 AM

Originally posted by Imageman

The fact is its an unsharp mask however you slice it.

But what else could it be?

rrstuff · 07-14-2014, 12:39 AM

I wonder if they might be limited by processing power on board of a camera. The idea of using deconvolution for sharpening is simple and is used by many camera manufacturers - at least those that also produce microscopes.

Nicolas06 · 07-18-2014, 01:54 PM

For this diffraction thing, if it is possible at high computationnal cost to do a good algorithm, well it might be far easier than you think to do. Provided the model work, the embedded camera engines use specialized hardware, they can work like a grapphic card, doing thousand of operation in parallel and being very specialized. So more or less for a specialized task they can do it far faster than a full PC... And a specialized hardware is not that costly once you produce it in large quantity and have you R&D done. I mean this would not cost more than today specialized chips.

The question is for now, is it worth it? No we agree, this is just a sharpening mask. After all, if you know you gear you avoid shooting past f/11 most of the time anyway.

But the day we start to have 50-100MP APSC sensors where the diffraction will be visible at f/5.6 or even f/4, this might be worth to do it... And this will also be bundled in raw processing software like DxO or Lightroom.

lytrytyr · 07-18-2014, 06:15 PM

Originally posted by Nicolas06

But the day we start to have 50-100MP APSC sensors where the diffraction will be visible at f/5.6 or even f/4, this might be worth to do it.

Pentax is already there with the Q!

vonBaloney · 07-18-2014, 06:20 PM

Originally posted by lytrytyr

Pentax is already there with the Q!

Yup...

Nicolas06 · 07-23-2014, 12:22 AM

Originally posted by lytrytyr

Pentax is already there with the Q!

Yes with the Q you should shoot at f/4 max instead of f/11 max. I don't know if all Q lens can do that with optimal quality but the crop factor allow this f/4 to have a good enough deph of field. The problem is more on the opposite.... Where are the f/1 lenses? and f/0.7 lenses to get shallow deph of field?

07-10-2014, 01:44 PM	#37
Stone G. Pentaxian Join Date: Jun 2010 Location: North Zealand, Denmark Photos: Gallery \| Albums Posts: 1,516 Original Poster	Some First Examples I won't discuss the value, if any, of this upgrade. Just curious to see the effect of this "compensation-on-demand". As others have pointed out, this doesn't change the raw files; it only works in-camera and only for Pentax lenses that the camera will reckognize. Here's my test target: A folding curtain some 3 meters away. AA-filter simulation was turned off with these images (in fact, I don't think I ever have it on): Test Target (Downsized DA 18-135 @ 135mm full image) Then some 100% non-resized comparisons with Compensation off to the left and on to the right. First my DA200 lens: Pentax-DA @ f/22 Not much of a difference to my old eyes and no need to show comparisons at larger apertures. Quite different with the DA 18-135 lens: Pentax-DA 18-135 lens@ 135mm and f/5.6 Pentax-DA 18-135 lens@ 135mm and f/11 Pentax-DA 18-135 lens@ 135mm and f/22 Pentax-DA 18-135 lens@ 135mm and f/32 You may judge for yourself, but there is an effect and it does somehow seem to differentiate between different lenses at sam f-number.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Rumoured K-3 f/w update: diffraction correction and mirror flap improvement	JPT	Pentax K-3 & K-3 II	42	07-11-2014 10:34 AM
Flash Compensation, Overall Compensation and Ambient Lights	Rashedul	Flashes, Lighting, and Studio	3	01-11-2013 04:19 PM
Diffraction and our Q	hnikesch	Pentax Q	17	12-05-2012 10:59 PM
Hyperfocus and diffraction help.	ladybug	Troubleshooting and Beginner Help	14	05-13-2012 05:05 PM
Exposure compensation..umm what?	Raptorman	Pentax DSLR Discussion	14	01-18-2010 12:33 PM

07-08-2014, 06:39 PM	#31
rawr Veteran Member Join Date: Jun 2009 Posts: 11,913	Originally posted by normhead the "unique thing" Pentax is going to introduce with it's FF But Sony is already doing it on FF with the A7r. And Pentax are already doing it on the 645Z. I don't think it is news or a big deal. Unless I am reading the benefits of this technology entirely wrong, diffraction compensation might be buried somewhere as dot-point number 17 on a Pentax marketing brochure for FF, but won't be the sort of thing to sway buyers: 'Gee, the Pentax K-1 has diffraction compensation for selected lenses! I'm not going to buy a Nikon D610 now - I'll buy the Pentax!'.

07-08-2014, 07:31 PM	#32
normhead Pentaxian Join Date: Jun 2007 Location: Near Algonquin Park Photos: Gallery \| Albums Posts: 40,451	Originally posted by rawr But Sony is already doing it on FF with the A7r. And Pentax are already doing it on the 645Z. I don't think it is news or a big deal. Unless I am reading the benefits of this technology entirely wrong, diffraction compensation might be buried somewhere as dot-point number 17 on a Pentax marketing brochure for FF, but won't be the sort of thing to sway buyers: 'Gee, the Pentax K-1 has diffraction compensation for selected lenses! I'm not going to buy a Nikon D610 now - I'll buy the Pentax!'. If it's simply sharpening yes, if it actually better defines diffraction softened objects, it could be major.

07-09-2014, 09:22 PM	#33
Imageman Veteran Member Join Date: Dec 2013 Photos: Albums Posts: 461	Its conclusive, Pentax diffraction compensation is as I suggested, simply sharpening the image using a well know and used algorithm such as unsharp mask. Quote - The main new feature introduced via this firmware update (for k3) is Diffraction Correction, an addition to the K-3's existing suite of Lens Correction options. This feature was originally introduced in the Pentax 645Z. Diffraction correction attempts to compensate for the degradation in image quality that occurs at small aperture settings, such as F11 or F16. Like the other Lens Correction settings, diffraction correction only applies to JPEG files and it is only available with Pentax DA, DA L, D-FA, and FA Limited lenses. Note that on the Pentax K-3, the diffraction correction option is only accessible via the first tab of the main menu; it is not being added to the "control panel" quick-access screen as on the 645Z. After performing some quick image tests, we discovered that the effect of this new feature is essentially an unsharp mask (a sharpening algorithm).

07-09-2014, 09:53 PM	#34
rawr Veteran Member Join Date: Jun 2009 Posts: 11,913	Originally posted by Imageman Its conclusive, Pentax diffraction compensation is as I suggested, simply sharpening the image using a well know and used algorithm such as unsharp mask. Maybe. But one should be able to test 'simple' USM vs deconvolution sharpening, to see what is at work with the Pentax solution. Could be done by shooting in RAW+ at f16-f22 eg a newpaper page or another suitable subject, then processing the RAW using (a) normal USM sharpening, as seen in Lightroom or DxO; and (b) a deconvolution tool like Focus Magic or Topaz InFocus or AstraImage, and then comparing (a) and (b) with what the in-camera JPG looks like when diffraction compensation is turned on ... Maybe when I have the time I'll do some comparisons like that. Last edited by rawr; 07-09-2014 at 10:45 PM.

07-10-2014, 12:37 AM	#35
Digitalis Site Supporter Join Date: Mar 2009 Location: Melbourne, Victoria Photos: Gallery Posts: 11,694	Originally posted by Imageman Its conclusive, Pentax diffraction compensation is as I suggested, simply sharpening the image using a well know and used algorithm such as unsharp mask. I f*cking said so. The sad part is that it is also affecting RAW files as well.

07-10-2014, 01:24 AM	#36
rawr Veteran Member Join Date: Jun 2009 Posts: 11,913	Originally posted by Digitalis The sad part is that it is also affecting RAW files as well. Perhaps not. As I understand them, the PDF notes issued with the firmware update, seem to relate mainly to the in-camera processing of images. They say: 'Diffraction Correction is not on RAW Development items. (Setting is reflected when shooting image but it cannot be changed)' which I take to mean that diffraction correction is both (1) not available on the in-camera RAW development menu options, and (2) diffraction correction shows up on the review screen when shooting an image because the image displayed is the JPG preview baked into all RAW's, not the state of the actual RAW. The fact that the notes also say that the corrections don't apply if the lenses aren't Pentax, or if a teleconverter is used, also strongly suggests to me that diffraction correction isn't applied to RAW's automatically, but is only an option. It will be interesting to see what new fields show up in the K-3 EXIF that relate to this function once EXIFTOOL has been updated.

07-10-2014, 02:31 PM	#38
Na Horuk Veteran Member Join Date: Feb 2012 Location: Slovenia, probably Photos: Gallery \| Albums Posts: 11,186	Originally posted by Stone G. You may judge for yourself, but there is an effect and it does somehow seem to differentiate between different lenses at sam f-number. Yeah, its not super advanced deconvolution, but it is lens-specific sharpening profiles, which is still better than a one-size-fits-all sharpening solution. I didn't notice and odd artefacts caused by this feature, so its probably ok for the jpeg-shooters.

07-12-2014, 09:31 AM	#39
Imageman Veteran Member Join Date: Dec 2013 Photos: Albums Posts: 461	The fact is its an unsharp mask however you slice it. Manufacturers are businessmen and they are not in the business of pushing boundries unless theres a lot of profit in it. This was never going to be anything other than a simple cheap sharpening which any competent photographer who can sharpen in post processing can do already and doesn't need. Its for jpeg shooters who never sharpen or do anything, its for holiday snaps, and its a marketing exercise so as not to miss out on the latest wrinkle that the competitors offer. Its about looking good in the shop so the salesmen have something else to show customers and its about keeping market share at zero or little cost to the manufacturer. Don't assume Pentax want to push boundaries in research and development. make no mistake, if Pentax could make more money selling 6 megapixel point and shoots instead of dSLRs with all their associated tooling and research plus lens design plus accessories, they would drop dSLR production faster that a dog drops a bone. Profit and cheap production is all that matters. Ive said as much all along.

07-12-2014, 09:51 AM	#40
wombat2go Veteran Member Join Date: Jan 2010 Location: Michigan Photos: Gallery Posts: 2,207	Originally posted by Imageman The fact is its an unsharp mask however you slice it. But what else could it be?

07-14-2014, 12:39 AM	#41
rrstuff Veteran Member Join Date: Sep 2013 Location: Pasadena, CA Photos: Albums Posts: 1,126	I wonder if they might be limited by processing power on board of a camera. The idea of using deconvolution for sharpening is simple and is used by many camera manufacturers - at least those that also produce microscopes.

07-18-2014, 01:54 PM	#42
Nicolas06 Veteran Member Join Date: Nov 2013 Posts: 4,854	For this diffraction thing, if it is possible at high computationnal cost to do a good algorithm, well it might be far easier than you think to do. Provided the model work, the embedded camera engines use specialized hardware, they can work like a grapphic card, doing thousand of operation in parallel and being very specialized. So more or less for a specialized task they can do it far faster than a full PC... And a specialized hardware is not that costly once you produce it in large quantity and have you R&D done. I mean this would not cost more than today specialized chips. The question is for now, is it worth it? No we agree, this is just a sharpening mask. After all, if you know you gear you avoid shooting past f/11 most of the time anyway. But the day we start to have 50-100MP APSC sensors where the diffraction will be visible at f/5.6 or even f/4, this might be worth to do it... And this will also be bundled in raw processing software like DxO or Lightroom.

07-18-2014, 06:15 PM	#43
lytrytyr Pentaxian Join Date: Nov 2011 Photos: Gallery \| Albums Posts: 4,310	Originally posted by Nicolas06 But the day we start to have 50-100MP APSC sensors where the diffraction will be visible at f/5.6 or even f/4, this might be worth to do it. Pentax is already there with the Q!

07-18-2014, 06:20 PM	#44
vonBaloney Pentaxian Join Date: Oct 2011 Location: Albuquerque, NM Posts: 6,029	Originally posted by lytrytyr Pentax is already there with the Q! Yup...

07-23-2014, 12:22 AM	#45
Nicolas06 Veteran Member Join Date: Nov 2013 Posts: 4,854	Originally posted by lytrytyr Pentax is already there with the Q! Yes with the Q you should shoot at f/4 max instead of f/11 max. I don't know if all Q lens can do that with optimal quality but the crop factor allow this f/4 to have a good enough deph of field. The problem is more on the opposite.... Where are the f/1 lenses? and f/0.7 lenses to get shallow deph of field?