[dcpropilot]

I found this piece in the development story to be very interesting. Previous lenses detect the focus in steps, while this one is "stepless"?

"Installed in the lens barrel are a position sensor which detects the position of the focus lenses steplessly, rather than in steps, and a sensor which detects the SDM’s rotation speed. Obtaining position and speed data simultaneously optimizes the balance between autofocusing speed and focusing precision, despite the lens’s large open aperture of F1.4 and its very shallow depth of field".

[monochrome]

I found this piece in the development story to be very interesting. Previous lenses detect the focus in steps, while this one is "stepless"?

"Installed in the lens barrel are a position sensor which detects the position of the focus lenses steplessly, rather than in steps, and a sensor which detects the SDM’s rotation speed. Obtaining position and speed data simultaneously optimizes the balance between autofocusing speed and focusing precision, despite the lens’s large open aperture of F1.4 and its very shallow depth of field".

Step #1 in a competitive AF system. What will the next body platform do to optimize stepless AF?*


* It is possible the core logic in SAFOX (which goes back at least to MZ-S) will be replaced

[beholder3]

If I remember correctly very old AF lenses from Canon used a system where position was read physically with copper contacts from the lenses. This obviously had to have a limited number of readable positions.

These posts here also show some clear stepwise system:

Programmatic focus control and absolute/relative focus position

It also should be remembered that these piezodrives have less torque as speed goes up.

Optimizing the whole system seems to be a fiddly task as every one wants fast AND precise at the same time.

[AstroDave]

Optimizing the whole system seems to be a fiddly task as every one wants fast AND precise at the same time

Some years ago, when I was a contractor at NASA, NASA came up with a buzz phrase for new missions: Faster, Better, Cheaper (I think this was supposed to make Congress feel better about the NASA budget)

The (unspoken to the brass) response from those of us in the trenches was "pick two out of three"!

[Kozlok]

I think Pentax chose better and faster. Nothing about the DFA50/1.4 seems to be in the ”cheaper” camp.

[Mikesul]

How can anything digital be stepless? The clock, the individual bits etc.

[thibs]

Step #1 in a competitive AF system. What will the next body platform do to optimize stepless AF?*


* It is possible the core logic in SAFOX (which goes back at least to MZ-S) will be replaced

At least as far as names goes AFAICT, SAFOX came with SF-X. Happy?

[boriscleto]

If I remember correctly very old AF lenses from Canon used a system where position was read physically with copper contacts from the lenses. This obviously had to have a limited number of readable positions.

These posts here also show some clear stepwise system:

Programmatic focus control and absolute/relative focus position

It also should be remembered that these piezodrives have less torque as speed goes up.

Optimizing the whole system seems to be a fiddly task as every one wants fast AND precise at the same time.

The lenses contain a ROM chip that holds the program curve (brightness/shutter speed/aperture) for prime lenses or up to three program curves for zooms at their different focal length settings. The distance information is hardwired on the lens barrel. A number of tiny parallel conducting/non-conducting strips are glued to the rotating part of the lens. The same number of contacts are attached to the fixed part of the lens. As the lens barrel turns during focusing, different combinations of the strips are detected by the contacts. There are four parallel strips in the F 50/1.7, resulting in at most 16 different distance settings. But for that particular lens not all combinations are used.

A similar technique must be used for the focal length.

Features and Operation of the KAF Mount | The K-Mount Page

KAF2 added a CPU for MTF. 4-bit for non-power zoom and 8-bit for power zoom.

Features and Operation of the KAF2 Mount | The K-Mount Page

Dimitrov doesn't have anything about KAF3 or KAF4...

He also had a page about the first 4 generations of SAFOX...

Pentax Auto Focus Systems | The K-Mount Page

[kiwi_jono]

How can anything digital be stepless? The clock, the individual bits etc.

Exactly. It could be done with analogue technology but I highly doubt they have done that. Perhaps its marketing talk for the resolution is so fine that its basically indistinguishable from analogue.

[monochrome]

At least as far as names goes AFAICT, SAFOX came with SF-X. Happy?

Huh. Then I have a camera body with the first iteration, the last iteration and several intermediate iterations.

I hope many people are happy in February.

[Mikesul]

Exactly. It could be done with analogue technology but I highly doubt they have done that. Perhaps its marketing talk for the resolution is so fine that its basically indistinguishable from analogue.

It has to be that. Processors are now so fast we do not see the step by step process of computing. They probably are only claiming that the adjustments are now so fast in the lens that we do not see them, not that they do not proceed in steps.

[monochrome]

It has to be that. Processors are now so fast we do not see the step by step process of computing. They probably are only claiming that the adjustments are now so fast in the lens that we do not see them, not that they do not proceed in steps.

The Seiko Seikosha shutter in the K2 was described as stepless. What I believe that meant, and this means, is the intermediate stops are infinite (or so small as to be deemed infinite).

[RobA_Oz]

Analogue is sometimes just the infinitesimal deceiving our ability to resolve.

[mee]

It has to be that. Processors are now so fast we do not see the step by step process of computing. They probably are only claiming that the adjustments are now so fast in the lens that we do not see them, not that they do not proceed in steps.

Maybe they went from regular floats to double precision? You get enough decimal points in your calcs ehhhh.. its virtually indistinguishable.



Same thing with audio production.. lots of digital recreations of analog gear these days. Some of it is quite believable... a lot of it is 'good enough.'

[mlt]

Are there any pictures online that show all four motor (Original SDM, DC, PLM, and new ring type SDM) types for comparison? The Ricoh picture showing the SDM next to the new ring type motor is good for those two, but it would be interestng to see pictures of the other two motor types along side.