[bwDraco]

For over a decade, Pentax has only made iterative improvements on the SAFOX AF system. More recent updates added more rows and columns but did not change the layout of the AF array. A closer examination of the SAFOX 13 AF grid reveals both similarities and differences from previous generations of the SAFOX autofocus system.

SAFOX 13 continues to use fundamentally the same layout of rows and columns as that of SAFOX 12. Among the selectable points, the center area is still a 5x5 grid of cross-type points, flanked by two columns of linear points on each side. The number of selectable points in each of these side areas has, however, increased from 4 to 8. The total number of selectable AF points is now 41, up from 33 in SAFOX 12 (K-1 and K-1 Mark II) and 27 in SAFOX 11 (K-3 and K-3 II).

But more significantly, SAFOX 13 adds an additional 60 assist AF points, dramatically increasing AF point density and making each point much smaller. Looking at pages 60 and 61 of the manual, when selecting a particular point to use, you can choose whether or not to use the four adjacent assist AF points. Select mode uses these additional points, while Select (S) does not. It also means that the Spot mode is much more precise than before. The AF grid also covers a much wider portion of the frame. All of these improvements are likely to be the result of both improved optics and a higher-resolution AF sensor.

Despite the similarities to earlier generations of SAFOX, the smaller AF points and higher point density represent a very welcome change. It addresses a common complaint with earlier Pentax cameras that the AF points are too large for precise selection. Combined with the high-resolution metering sensor enabling AI subject recognition, new AF algorithms, and an overhauled drive system which gives the AF system more time to track movement, I can see a dramatic improvement in AF.C tracking performance in the K-3 Mark III.

You Tube

You Tube

Questions, thoughts, or comments?

Draco

---

Last edited by bwDraco; 04-04-2021 at 02:04 PM. Reason: For some reason, posts on mobile don't show my signature

[vrphoto]

Tuppence-worth of comment - the major temptation to going down the early-adopter route was more accurate centre focus point - as well as depth-of-field bracketing, increased stabilisation, better high ISO performance (did succumb to temptation).

[bwDraco]

I jumped ship to Olympus over a year ago, but still have all my Pentax glass. I'm really intrigued as to what the K-3 Mark III has to offer. All the specs look very promising to me and it might just be enough to get me to dust off the D FA* 70-200mm f/2.8 lens and invest in the Pentax K system again.

Draco

[bwDraco]

Thinking more about how Pentax has implemented the new AF system, and looking at this page... I get the impression that the basic design of the secondary image registration (SIR) optics has not changed dramatically, except for optimizations to expand the AF grid to cover more of the frame and improve overall performance. (Pentax engineers note that they did have to redesign the entire mirror box, including by using a larger AF sub-mirror.) The SIR optics are the beam splitters that project the light to the actual AF sensor and is what determines the layout of the AF points. It also still has a "chromatic aberration compensating lens", which is almost certainly the diffractive optical element introduced in SAFOX IXi+. (More information about SIR optics in this Photography Stack Exchange thread.)

The SIR beam-splitter lens is referred to as the "separator lens" in this diagram from the Pentax Japan web page for the original K-3, which also depicts the diffractive optical element:



What is likely to have seen a more significant change is the underlying AF sensor. The sensor behind the SIR optics is what actually detects the phase difference to determine how to adjust focus. Using a higher-resolution sensor would allow each point to be smaller and make focusing more precise. This, in turn, enables a denser AF grid with many more points than before. In conjunction with other improvements, this should significantly improve subject tracking performance. A new focus sensor would also explain improved low-light focusing performance, down to EV -4 with f/2.8 lenses.

The expanded frame coverage of the AF array does come at a cost: page 117 of the manual states that with certain lenses, some or all of the outer columns of AF points may not be usable. While this is not unusual for high-density, wide-area AF systems—on the Canon EOS-1D X Mark III, lenses are divided into as many as 10 groups for the purpose of determining AF functionality—this is the first time we've seen a limitation of this type on a Pentax camera. The lenses affected are primarily consumer-grade zooms and wide-angle optics.

Draco

---

Last edited by bwDraco; 04-04-2021 at 02:04 PM. Reason: For some reason, posts on mobile don't show my signature

[jeffdrew]

Thinking more about how Pentax has implemented the new AF system, and looking at this page... I get the impression that the basic design of the secondary image registration (SIR) optics much has not changed dramatically, except for optimizations to expand the AF grid to cover more of the frame and improve overall performance. (Pentax engineers note that they did have to redesign the entire mirror box, including by using a larger AF sub-mirror.) The SIR optics are the beam splitters that project the light to the actual AF sensor and is what determines the layout of the AF points. It also still has a "chromatic aberration compensating lens", which is almost certainly the diffractive optical element introduced in SAFOX IXi+. (More information about SIR optics in this Photography Stack Exchange thread.)

The SIR beam-splitter lens is referred to as the "separator lens" in this diagram from the Pentax Japan web page for the original K-3, which also depicts the diffractive optical element:



What is likely to have seen a more significant change is the underlying AF sensor. The sensor behind the SIR optics is what actually detects the phase difference to determine how to adjust focus. Using a higher-resolution sensor would allow each point to be smaller and make focusing more precise. This, in turn, enables a denser AF grid with many more points than before. In conjunction with other improvements, this should significantly improve subject tracking performance. A new focus sensor would also explain improved low-light focusing performance, down to EV -4 with f/2.8 lenses.

The expanded frame coverage of the AF array does come at a cost: page 117 of the manual states that with certain lenses, some or all of the outer columns of AF points may not be usable. While this is not unusual for high-density, wide-area AF systems—on the Canon EOS-1D X Mark III, lenses are divided into as many as 10 groups for the purpose of determining AF functionality—this is the first time we've seen a limitation of this type on a Pentax camera. The lenses affected are primarily consumer-grade zooms and wide-angle optics.

Draco

Thx for your opinion & analysis. I have also begun assimilating some OMD and like their approach in several areas, like AF, EVF, carry weight and overall tech. My Penf & K3ii go out on walks together now! ��

[bwDraco]

In the spirit of this nearly decade-old thread, here are some AF grid diagrams. The circles are selectable AF points; those that are filled are f/2.8 high-precision points. Small dots are assist AF points which cannot be selected. The two are not to scale; SAFOX 13 covers a wider area of the frame than SAFOX 12. Note that the two assist points left and right of each of the center three points in the center column are also f/2.8 high-precision points. Sorry if this is a bit messy; drawing is not my strong suit. Regardless, this should highlight how similar the new AF system is to the previous generation.



Draco

[gatorguy]

"when selecting a particular point to use, you can choose whether or not to use the four adjacent assist AF points. Select mode uses these additional points, while Select (S) does not."

I for one have no idea why I should choose one or the other, and what the cons are for choosing incorrectly. Perhaps one of our more knowledgeable members might explain?

[Raffwal]

One thing I have thought about is that why (apparently) no DSLR manufacturer has used an on-sensor contrast detect AF to calibrate individual lenses. I assume it would be possible. One of the reasons I have used the last several years mainly my m43 gear and not Pentax is the lackluster accuracy and reliability of the phase detect AF used by Pentax. (I know that many users disagree and I have not used the latest models since K-5 Mk I, so let's not go there).

[sprint113]

One thing I have thought about is that why (apparently) no DSLR manufacturer has used an on-sensor contrast detect AF to calibrate individual lenses. I assume it would be possible. One of the reasons I have used the last several years mainly my m43 gear and not Pentax is the lackluster accuracy and reliability of the phase detect AF used by Pentax. (I know that many users disagree and I have not used the latest models since K-5 Mk I, so let's not go there).

I believe Nikon offers that under their Automatic AF fine tune feature on their recent high end models.

You Tube

You Tube

.

[thibs]

In the spirit of this nearly decade-old thread, here are some AF grid diagrams. The circles are selectable AF points; those that are filled are f/2.8 high-precision points. Small dots are assist AF points which cannot be selected. The two are not to scale; SAFOX 13 covers a wider area of the frame than SAFOX 12. Note that the two assist points left and right of each of the center three points in the center column are also f/2.8 high-precision points. Sorry if this is a bit messy; drawing is not my strong suit. Regardless, this should highlight how similar the new AF system is to the previous generation.



Draco

Very nice, thanks !

[gatorguy]

"when selecting a particular point to use, you can choose whether or not to use the four adjacent assist AF points. Select mode uses these additional points, while Select (S) does not."

I for one have no idea why I should choose one or the other, and what the cons are for choosing incorrectly. Perhaps one of our more knowledgeable members might explain?

No one yet? I suppose we have two weeks to figure it out.

[One3rdEV]

A brief quick comment about AF Fine Adjust regarding the prior post (#9): The Video regarding Nikon Learn & Explore is excellent in being professionally 'concise' and 'informative' in providing guidance for how to perform af fine adjustment !

The manual technique presented in the video is essentially the same as what Ricoh-Pentax has officially posted here:
How to optimize focusing accuracy with large-aperture lenses / explore | RICOH IMAGING

[To the moderators: I would recommend placing the links to these two resources (Ricoh-Pentax info) and (Nikon Learn & Explore Video) in an appropriate place with in the forum e.g. sickies or help or how to area.]

[Kobie]

No one yet? I suppose we have two weeks to figure it out.

Sounds the same as currently using SEL vs SEL with expanded on.

[phoebus]

No one yet? I suppose we have two weeks to figure it out.

For me, 'single point only' makes sense for wildlife amongst trees/grass etc, to get best possible chance of focussing on the desired target.

'A bird in focus is worth two of a bush' or some saying like that...