[bwDraco]

Pentax says it has used a diffractive optical element in the SAFOX IXi+ autofocus system (emphasis added):

The K-30 features the new state-of-the-art SAFOX IXi+ AF sensor module, which assures responsive, high-precision autofocus operation by incorporating a diffraction lens to minimize chromatic aberration in accordance to the characteristics of each individual lens.

I know Canon has two lenses that use diffractive optics to correct for chromatic aberration more effectively allowing more compact lens designs at the cost of some sharpness. Canon describes this technology as follows:

Conventional glass lens elements disperse incoming light, causing chromatic aberration. The DO element’s dispersion characteristics are designed to cancel chromatic aberrations at various wavelengths when combined with conventional glass optics.

Any thoughts on the use of diffractive optics in the AF system? Will this actually deliver increased AF accuracy? Are there any possible disadvantages of using a DO element in the AF module?

--DragonLord

[Anvh]

Well it's the cause of the AF focus problems with the k5 and the Kr in low light.

[Giklab]

Well it's the cause of the AF focus problems with the k5 and the Kr in low light.

It is? I thought it was because IR swamped the AF sensor.

[Anvh]

Indeed and with a diffractive optics also the IR will be better corrected.
IR is also light after all.

[rawr]

Pentax says it has used a diffractive optical element in the SAFOX IXi+ autofocus system

Pic related, from the K-30 images Adam put up:



You can see how important it is going to be for the alignment of all those baffles and lens elements to be precise!

[geru2000]

Any thoughts on the use of diffractive optics in the AF system? Will this actually deliver increased AF accuracy? Are there any possible disadvantages of using a DO element in the AF module?

--DragonLord

Well it's the cause of the AF focus problems with the k5 and the Kr in low light.

Anvh, I'm confused by your answer to the quote from DragonLord?

The way I read your statement it's the DO in the Kr and K5 AF system causing their problems yet neither has DO ?

...or....did you mean Defraction is causing the problem?

[Anvh]

I meant that defraction was the problem, well at least partially.
Lets say that it should help quite a bit with the problem, maybe not solve it with just this but who knows.

[Ray Pulley]

It is? I thought it was because IR swamped the AF sensor.

There has been an ongoing myth that IR is the cause of FF in Pentax AF systems as long as I can remember. This has been tested and debunked many times (Ricehigh is NOT a credible source IMO). The PDAF system has an IR cut filter in the optical path to solve any IR mis-focus issues.

The problem is apparently uncorrected CA in the AF system, which has also been discussed but it is difficult to prove or even test for. Different wavelengths focus at different planes in the system, and the AF system is only accurate for the range of wavelengths it was calibrated and corrected for.

The K5 introduced a color sensor in the AF system to try and offset this issue, but just as with AWB, it isn't always able to perfectly correct for every condition. My 2 K5's focus pretty well in low tungsten light after the firmware fix was introduced, but other cameras (like the Kr) do not have the color sensor in the AF systems and therefore they cannot easily be corrected for this problem.

Just like a prism, CA is a refraction problem, so a optic that has very specifically designed diffraction properties can offset the refraction. This is not the only way to correct for CA, but there must be some other reason (cost, size, etc.) to choose this technology over the other methods. Perhaps it is also more accurate?

Anyway, I think that this is a very positive development for Pentax, but only time will tell how effective it is.

Frankly, I have never understood why a simple offset could not be entered for use any time the user or AWB selected tungsten? This might not be a perfect solution, but it would probably satisfy most users. It could even be programmed to be triggered at certain light values as read by the meter, or to be applied in a gradual manner as the light is reduced.

With all of the noise and grief over this issue going back to at least the first Pentax DSLRs, it is hard to imagine why a firmware work-around wasn't ever implemented?

Ray

[Anvh]

There should be a sensor in the K5 to check the wavellength so the problem should never ever have been there to begin with...

[Ray Pulley]

There should be a sensor in the K5 to check the wavellength so the problem should never ever have been there to begin with...

Yes it does have a color sensor as part of the PDAF system. Apparently the original release had some sort of problem in the software that takes the sensor readings into account in the AF algorithm.

However, I think that this sensor probably isn't any better than auto white balance, which sometimes does not get the color balance right either.

Ray

[Class A]

There should be a sensor in the K5 to check the wavellength so the problem should never ever have been there to begin with...

Any SAFOX AF with a "+" in the name has such a sensor.

Falk (falconeye) deduced that the metering chip is doing double duty for metering and as a wavelength sensor, gauging the colour temperature of the light in an AF area.

One definite problem was that the metering chip was not nearly as sensitive as the AF system so the camera lost information about the colour temperature in low light.

Judging from Falk's report I reckon there was also an unresolved software bug, but the main source of the problem was the CA in the AF system's optical path (the wavelength sensor, i.e., the metering chip, was designed to allow software compensation against the CA, but this approach failed in low light and depending on the widest f-ratio on the lens, this could mean "not so low light" in reality).

The new "diffraction lens" approach is designed to combat the low light AF issues at its root cause (just confirming what Ray already said).

[Anvh]

I wonder if they could use the meter sensor in the viewfinder, it's used for the colour temperature and the light meter and more light goes to that sensor then the AF so it could be more accurate.

[Class A]

I wonder if they could use the meter sensor in the viewfinder, it's used for the colour temperature and the light meter and more light goes to that sensor then the AF so it could be more accurate.

The sensor used for the light meter is the one I've been calling "metering chip".

The fact that it is not sensitive enough, is one of the reasons for the K-5's low light AF problems. If you are dealing with slow lenses then the light does not need to be that low (e.g., 5EV) to cause AF issues.

[Anvh]

hmmm the K20D and the K7 didn't had these problems but.... they had far more problems with white balance, the K5 is pretty good even under tungsten so how can it not be sensitive while with WB it has a wider range?

I think Ray Pulley is simply right.

[Class A]

hmmm the K20D and the K7 didn't had these problems but....

The K-5 has different AF optics. They are more sensitive but apparently also have higher dispersion. This, most likely, made the K-5 different from its predecessors, not the metering sensor.

they had far more problems with white balance, the K5 is pretty good even under tungsten so how can it not be sensitive while with WB it has a wider range?

Range and sensitivity are two different things. Differences in WB could be down to software improvements only. Furthermore, isn't it the case that Auto-WB uses the actual image data, rather than consulting the metering sensor? It makes no sense to use the metering chip if you can use the real sensor data to obtain an Auto-WB value.

I think Ray Pulley is simply right.

Where do you see a difference between what he said and I said?
I agree with most of what he said, however, he is not right when he says "K5 introduced a color sensor in the AF system ..." because the K-7 had the same system before (with different AF optics).