[JPT]

The linked Japanese article has some more information about the diffraction correction added in firmware 1.10.

Some points of interest:
- the correction applied is tailored to the lens used, not just a generic unsharp mask
- supported lenses are DA, DFA, FA limited
- diffraction correction is only possible because of the power of the latest generation of processors
- it makes a noticeable difference with apertures of f8+
- the correction is very effective up to f16 and makes f22 more usable

http://m.dc.watch.impress.co.jp/docs/review/minirepo/20140717_658292.html
(I don't know why, but I have trouble posting links to Google Translate on my phone - sorry)

[Imageman]

Im still unconvinced.


They can claim that their system is superior to simple unsharp mask all they want but ive looked at their claims and they don't stand up.


The example they give is using a bird pic with their system on and off.
I took their off pic at f22 and applied a simple unsharp mask to it, this will surely expose the inadequacy of unsharp mask and I will be unable to replicate their superior results if their claims are to be believed.


Look for yourself at the results.


The left hand image is the unsharpened f22 image


The middle image is their method of diffraction correction.


The right hand image is my result after 5 seconds with unsharp mask.


You decide if their system is any better than unsharp mask.

[rawr]

I found the following text in the DCWATCH article most interesting:

"Conventional's that there be advantageous to move to 35mm full-size sensor to make the depth of field while maintaining the high resolution, if you remain in the APS-C, only reduced to 16 million pixels about the number of pixels of was common sense. Waseda union of diffraction correction and APS-C image sensor 24 million pixels of has the potential to overturn it."

The pitch by the author seems to be that in-camera diffraction correction removes an advantage larger sensor full-frame cameras have enjoyed over APS-C (due to the slower onset of diffraction effects on larger sensors). Now you can maintain deep depth of field whilst also retaining high resolution on APS-C, just like on full-frame. So diffraction correction is kind of a 'full-frame killer'.

It's an interesting spin. I hope Ricoh don't take that perspective too much to heart, however.

[dms]

Frankly it seems like "black magic" if they actually claim to reduce effects of diffraction, Diffraction is an effect that is irreversible--the information loss is just that--lost. Unsharp mask, etc.provide the appearance of greater sharpness--but that is not the same thing.

[Adam]

not just a generic unsharp mask

Perhaps they apply sharpening to the raw pixel data and vary the strength based on the lens- but in any case, there's no such thing as magic. It would be interesting to see what's in that black box

[Michaelina2]

Look for yourself at the results.

OK.... On my monitor, the Pentax method appears superior to your application of the un-sharp mask.

...just sayin'... M

[Ole]

The advantage over applying an unsharp mask in post processing is that no post processing is needed. The jpg images are ready as they come out of the camera, also when using F11 and F16, a big time saver. I usually shoot RAW and post process. I only shoot JPG when I don't want to spend time post processing and in that case the new feature is handy.

[rawr]

It would be interesting to see what's in that black box

Somewhere it was mentioned that Ricoh are taking advantage of the newer Fujitsu Milbeaut CPU in the K-3 to do the diffraction correction.

And indeed, Fujitsu says their latest 7th generation Milbeaut chip includes all sorts of optical correction cabailities right there in the CPU:

Features

• A new image processing algorithm with greatly improved optical correction capabilities
• Faster processing thanks to a newly developed Integrated Image Processor circuit
• Improved high dynamic range (HDR) photograph quality using a JPEG-HDR™ format developed by Dolby Laboratories.

Other Specifications

• CPU: ARM Cortex-A5MP
• Maximum image processing speed equivalent to12fps at 24M pixels
• Hardware assist capable of feature extraction
• Improved lens correction, lens distortion correction, lens resolution correction
• Accelerated multi-frame operation

So it is probably accurate to say that they are using something perhaps a little more complex than just USM sharpening. Probably all sorts of other signal processing is linked to how the diffraction correction works.

[Imageman]

OK.... On my monitor, the Pentax method appears superior to your application of the un-sharp mask.

...just sayin'... M

Im quite happy for you to see a difference and for theirs to look better to you.


I would be very interested for you to have a go at sharpening there are many different sharpening methods and your attempt could be closer to theirs than mine.


I dont see the difference that you see however on my monitor.


What were talking about here is the suggestion by some seemingly knowledgeable people that a very sophisticated solution is being applied to the image in camera, that somehow deconvolves light ray bundles to end up with the original image before it was degraded by diffraction.


Whereas my contention is its simply an application of cheap and cheerful sharpening.


Whats your view.

[JPT]

I wonder whether there is any detrimental effect on the camera's speed of operation when the correction is set to "on".

Anyway, the camera that would benefit most from this technology is the Q. That camera is constantly close to its diffraction limit.

[aglet]

I also prefer the middle example of the image.
The USM method increases overall contrast appearance more than I might want and is a sharpening method I rarely use unless I also want to increase that fine local contrast effect.

[Imageman]

There are I believe 2 important issues coming out of this.


First, this option offered by Pentax is quite obviously desirable for the casual shooter who has no desire to post process. I would not be so crass as to try to dissuade anyone from using an option available to them that assists them in image production.


Second, this question of whether unsharp mask is used in the Pentax approach is academic, what is not academic however is the fact that this Pentax solution is available for only a small selection of lenses.


If it turns out as I believe that a simple sharpening method gives comparable results to the method used by Pentax, then this simple sharpening method will be applicable to all lenses and not just the few that Pentax have programmed into the K3 firmware.


I believe that the k3 owner is being short changed here into believing that lens diffraction correction can only be applied to a handful of lenses by suggesting that you need the Pentax method and nothing else will work satisfactorily.


If as I believe a comparable post processing sharpening technique exists that will perform similarly to the "official" method, then that can be used on all manual m42 lenses, manual k mount lenses, and everything else capable of mounting to the k3, including bellows and extension rings and reversing rings irrespective of aperture size and lens design.


We should be looking to expand options here rather than restrict choice.

[John Cafarella]

The middle one shows more texture in the white areas. Definitely better.

Now if Adobe could build that into LR....

[Michaelina2]

Whats your view.

My view, in response to your request for feed-back, is to simply observe that the Pentax approach produced a better result on my monitor than your application of an un-sharp mask. In other words, the Pentax method seems to do what is promised, that's all. Perhaps, you feel the feature 'ought' to do more, but it is what it 'is'. The two terms have different meanings... eh?

Cheers... M

[Imageman]

I have to say im truly shocked.


Everybody is seeing a huge difference that I just don't see with my poor eyes.


Your all seeing the Pentax as a much better image than the unsharp mask version.


I have decided to let you prove that the Pentax image is streets ahead of unsharp mask with a little challenge.


It should be very easy for you as you found the Pentax version to be so much better than unsharp mask.


Your task is to look at 4 images and spot the Pentax diffraction corrected image amongst the unsharp mask sharpened images.


One of is the Pentax diffraction corrected image, and the other 3 are sharpened slightly differently with unsharp mask.


This should be a walkover for you guys, and I fully expect everyone to immediately spot the Pentax image. This time no clues.


Name the Pentax image is it 1 2 3 or 4