[Ray Pulley]

The same reason that stops me from entertaining the thought that the moon is strung to the earth by an invisible rubber band.

It make sense to use the f/5.6 sensors first as they see more subject distance range. It also makes sense for the AF procedure to take a bit longer with the f/2.8 sensors as the focus movements will have to be more precise. However, I do not believe that camera internal processing adds further delays or that higher resolution sensors take noticeably more time to evaluate.

You are free to believe what you want, but you haven't supplied any reason for your belief. Normally such belief is called religion

What seems clear enough to me is that adding more calculations and 2x as many sensors to read, coordinate and evaluate cannot be faster unless some other optimizations or changes were made. It is certainly possible that when the algorithms were re-written for the new features they were optimized and improved such that the extra processing has little or no speed impact.

Canon uses a dedicated AF processor in the models we have been discussing, which we know Pentax does not employ, So, consider the fact that the main processor now has to read and process data from 2x as many sensors as before, so there has to be more overhead and processing cycles on the main (2+ year old) prime engine in the K5.

Whether the extra processing is noticeable or not cannot be separated from the overall AF system performance, which also includes the performance of the attached lens, so we will never know.

I hope that the system is more accurate than the K5Mk1 and at least as fast overall. It is my belief that the lenses are the thing holding back the Pentax AF speed, especially the SDM versions, so it could well be that the camera will still be faster than the lenses and the overall system will not slowdown. I am looking forward to well executed tests that will answer some of these questions.

Ray

[philbaum]

You are free to believe what you want, but you haven't supplied any reason for your belief. Normally such belief is called religion

What seems clear enough to me is that adding more calculations and 2x as many sensors to read, coordinate and evaluate cannot be faster unless some other optimizations or changes were made. It is certainly possible that when the algorithms were re-written for the new features they were optimized and improved such that the extra processing has little or no speed impact.

Canon uses a dedicated AF processor in the models we have been discussing, which we know Pentax does not employ, So, consider the fact that the main processor now has to read and process data from 2x as many sensors as before, so there has to be more overhead and processing cycles on the main (2+ year old) prime engine in the K5.

Whether the extra processing is noticeable or not cannot be separated from the overall AF system performance, which also includes the performance of the attached lens, so we will never know.

I hope that the system is more accurate than the K5Mk1 and at least as fast overall. It is my belief that the lenses are the thing holding back the Pentax AF speed, especially the SDM versions, so it could well be that the camera will still be faster than the lenses and the overall system will not slowdown. I am looking forward to well executed tests that will answer some of these questions.

Ray

Ray - thanks for the interesting discussion. The focus time also varies according to the lens, e.g. my 50-135 seems slower than my DA17-70, which has a really short focus throw. I hope at some point, Ricoh attacks both the speed and reliability of lens focus mechanisms when they overhaul SDM.

[Class A]

You are free to believe what you want, but you haven't supplied any reason for your belief. Normally such belief is called religion

Dear Ray,

it is you who is believing something.

You wrote

Of course, being further apart makes the calculation take a little longer as well,

That's your belief.
I simply see no reason for such a belief.

So for the record: I am the non-believer here.

[awaldram]

So the assumption is because of the 2.8 sensor AF must be slower?

I see it differently

Fit a slower than f2.8 lens and AF iterations will be identical with the f5.6 sensors in use.
Fit f2.8 and faster the f2.8 sensor activate and double the precision data is fed to the the lens so half the iterations are required for greater accuracy.

i.e f5.6 sensor historically 2.8 lens af pressed move to 3.5m target adjust back 2 CM forward 1 cm lock achieved 3.49M
f2.8 sensor now AF press move to 3.51m target move back 1.7cm forward .6 lock achieved 3.499m

[JinDesu]

So the assumption is because of the 2.8 sensor AF must be slower?

I see it differently

Fit a slower than f2.8 lens and AF iterations will be identical with the f5.6 sensors in use.
Fit f2.8 and faster the f2.8 sensor activate and double the precision data is fed to the the lens so half the iterations are required for greater accuracy.

i.e f5.6 sensor historically 2.8 lens af pressed move to 3.5m target adjust back 2 CM forward 1 cm lock achieved 3.49M
f2.8 sensor now AF press move to 3.51m target move back 1.7cm forward .6 lock achieved 3.499m

I think the assumption is based on Canon's implementation of the F2.8 sensor. They utilize the F5.6 to get close, and then F2.8 to lock.

So from that we gather that the F2.8 is more precise, but takes longer to confirm the precision, while the F5.6 readily moves from OOF to almost-focus.

[rawr]

So from that we gather that the F2.8 is more precise, but takes longer to confirm the precision, while the F5.6 readily moves from OOF to almost-focus.

We gather and we presume, but no one really knows anything concrete about how the AF works or performs.

I mean to date no one anywhere has even seen a sample pic from the camera, let alone had an opportunity to do an AF test.

And I second or third the notion that Pentax need to release a White Paper detailing all the groovy new features of the K-5 II and K-5IIs, including its AF.

[Ray Pulley]

Dear Ray,

it is you who is believing something.

You wrote

Of course, being further apart makes the calculation take a little longer as well,

That's your belief.
I simply see no reason for such a belief.

So for the record: I am the non-believer here.

My last response to you on this subject:

Note your words from an earlier post, especially the word "believe":

"However, I do not believe that camera internal processing adds further delays or that higher resolution sensors take noticeably more time to evaluate"

Since you clearly have a belief that internal processing will not add any further delays, you must have some reason for your belief, but continue to simply state your "belief" with no supporting logic.

Not a very convincing way to make a point, IMO.

Ray

[Class A]

Not a very convincing way to make a point, IMO.

You've got this completely backwards.

You make a claim; it was about AF speed but for argument's sake let's assume you said "Of course, there is a tiny teapot circling the sun. It is too small for anyone to see, but surely it is there."

I then reply and ask what makes you think there is a tiny teapot circling the sun (as I cannot even begin to think about your potential reasons for making such a claim) and later on state that "I don't believe that there is a tiny teapot circling the sun" (read "I don't join you in believing that there is a tiny teapot circling the sun").

You then accuse me of not providing enough evidence to support my claim. You are now demanding proof from me that there is no tiny teapot circling the sun. Sorry, it doesn't work that way.

Note,

• Neither of us has access to the internals of the AF algorithm you are talking about. Hence I simply cannot do more than "believe". However, note that my "belief" is grounded in my experience as a software engineer. Also,
• I didn't make a claim. I challenged your claim. If anyone needs to provide evidence or make convincing arguments it is you (-> argumentum ad ignorantiam). Furthermore,
• I cannot even begin to imagine how your claim could be upheld. That makes it hard for me to provide counter arguments (over and above to what I already wrote).

[pavpen]

If you think of the AF array as looking back at the lens through these slits, it can only see points located on a very specific part of the lens (there are two matching slits, one on each side).

I’m doing some research before taking plunge with pentax when I came across this thread. And this is my thought about the discussable topic.
If I understand correctly the aperture of the Main Lens (ML) defines the amount of light that will transfer to CMOS (as well as degree of distortions and other IQ factors.) So ML transfers photographing object to the certain distance from the back end of ML (plan of CMOS) and AF must make the projecting image sharp, which means “AF system” analyzes not areas on ML but the splitted light from object at the distance of CMOS. So, my point is F/2.8 sensors are sittuating further apart only because the F/5.6 sensors were placed closer by spatial design.
Cheers.

[Class A]

I’m doing some research before taking plunge with pentax when I came across this thread.

The K-5's AF is very performant in general and apparently belongs to the most accurate in its class.

It can struggle in very low artificial light, leading to front-focus. Furthermore, if you are interested in capturing fast action, there are other cameras (such as the Canon 7D) which have much quicker AF system with more AF areas better suited to track subjects.

The K-5 II's AF will not have the low-light issue. I may be a slight improvement regarding speed, but I'm sure it won't reach the performance of a 7D.

So, my point is F/2.8 sensors are sittuating further apart only because the F/5.6 sensors were placed closer by spatial design.

I'm not sure what your question/statement is.

F/2.8 AF sensors have a larger baseline, making them more accurate.

[pavpen]

The K-5's AF is very performant in general and apparently belongs to the most accurate in its class.

It can struggle in very low artificial light, leading to front-focus. Furthermore, if you are interested in capturing fast action, there are other cameras (such as the Canon 7D) which have much quicker AF system with more AF areas better suited to track subjects.

The K-5 II's AF will not have the low-light issue. I may be a slight improvement regarding speed, but I'm sure it won't reach the performance of a 7D.

Sorry for skim introduction sentence. I meant to say that i’m considering +/- of K-5 versus K-5ii, so there is no question about the brand name for me.

I'm not sure what your question/statement is.

F/2.8 AF sensors have a larger baseline, making them more accurate.

What I was trying to say (and this is a pure speculation) is that when engineer modifies an existing part he has limited area on sensor plate (with F/5.6 sensors already in place) to work with for F/2.8 sensors – the plate real estate is limited. Again this is my personal assumption.