Reading mode: Light Dark

Pentax AF: What's needed?

Go to last post
J

Jeff Charles

Veteran Member
Messages
7,514
Solutions
1
Reaction score
2,034
Location
MX
What features and capabilities will Pentax need to include in the K-3 to match the AF systems in the Nikon D300S and the Canon 7D? Those new features and capabilities would be in addition to lenses with faster AF motors.

I'm most interested in the thoughts of people who have used one of those cameras and who can say from experience how their AF systems perform compared to the K-5's.
--
Jeff
 
Here is my take based upon what I have garnered from the forums.
4 aspects:

1) Predictive focus working just as well to track a subject moving through focal planes using a single focus point.

2) concept of "helper points" - you designate a primary point to use but can also enable 'helper' points. The key here is the camera will always use the designated point if it detects focus -- the helper points are only used if the primary point can't obtain focus.

3) When all focus points are allowed, Tracking across multiple points: as the subject moves across the viewfinder the focus system accurately passes "active" point to track the subject (vs. switching to something in the background). Not many sports shooters do this but it's extremely helpful for birders shooting small BIF.

4) Customization - ability to select different arrays for which points are used, change balance between focus speed & accuracy etc.

I think by far the biggest gap seems to be #1 - the ability and reliability to track a subject through focal planes - ESPECIALLY when subject speed and direction is not constant. There are other nuances, but I think the above represent the areas of biggest gap.
 
Jeff Charles said:
What features and capabilities will Pentax need to include in the K-3 to match the AF systems in the Nikon D300S and the Canon 7D? Those new features and capabilities would be in addition to lenses with faster AF motors.

I'm most interested in the thoughts of people who have used one of those cameras and who can say from experience how their AF systems perform compared to the K-5's.
Click to expand...
.

1) More, smaller AF points, with cross-hatch making up a greater percentage.

2) AF module with enough core speed and throughput to run algorithms very quickly, allowing possibly more sophisticated algorithms with modern approaches to predictive focus.

3) Funding (and commitment) to implement the above. :)

.
--
Here are a few of my favorite things...
---> http://www.flickr.com/photos/95095968@N00/sets/72157626171532197/
 
John_A_G said:
Here is my take based upon what I have garnered from the forums.
4 aspects:

1) Predictive focus working just as well to track a subject moving through focal planes using a single focus point.

2) concept of "helper points" - you designate a primary point to use but can also enable 'helper' points. The key here is the camera will always use the designated point if it detects focus -- the helper points are only used if the primary point can't obtain focus.

3) When all focus points are allowed, Tracking across multiple points: as the subject moves across the viewfinder the focus system accurately passes "active" point to track the subject (vs. switching to something in the background). Not many sports shooters do this but it's extremely helpful for birders shooting small BIF.

4) Customization - ability to select different arrays for which points are used, change balance between focus speed & accuracy etc.

I think by far the biggest gap seems to be #1 - the ability and reliability to track a subject through focal planes - ESPECIALLY when subject speed and direction is not constant. There are other nuances, but I think the above represent the areas of biggest gap.
Click to expand...
Good list. I agree about the predictive focus. I believe that would address the problem of the subject changing distance between the time focus is acquired and when the shot is taken.
--
Jeff
 
moving_comfort said:
3) Funding (and commitment) to implement the above. :)
Click to expand...
That's critical. However, there's no discounting the AF engineering expertise that Canon and Nikon have built up over the years. Can Pentax catch up by dedicating a lot of resources to the problem, or is it an arcane body of knowledge that takes time, and trial and error, to acquire?

--
Jeff
 
moving_comfort said:
2) AF module with enough core speed and throughput to run algorithms very quickly, allowing possibly more sophisticated algorithms with modern approaches to predictive focus.
Click to expand...
I've always thought SR and AF were at odds with each other for processing time. The latest Milbeaut image processor allows for dual processors, so maybe a lot of AF issues will be resolved when Pentax puts it and dual CPUs into the next camera.

Thank you
Russell
 
Are more focus points actually required? As long as they are in the right place in the frame the current number should be ok. I note that Canon have said that one of the reasons for the low (supposedly) number of points on the 6d is to allow the centre point to focus at very low EV, just like Pentax are claiming for the k-5 refreshes.
Brian
 
Just to clarify this, the Canon engineer said the centre point needed to be large to give the low light performance. This is normally the knock on Pentax.
Brian
 
Russell Evans said:
moving_comfort said:
2) AF module with enough core speed and throughput to run algorithms very quickly, allowing possibly more sophisticated algorithms with modern approaches to predictive focus.
Click to expand...
I've always thought SR and AF were at odds with each other for processing time. The latest Milbeaut image processor allows for dual processors, so maybe a lot of AF issues will be resolved when Pentax puts it and dual CPUs into the next camera.
Click to expand...
I think they have their own dedicated chips. Same as the image processing has its own chip, which is different from the CPU
 
Jeff Charles said:
moving_comfort said:
3) Funding (and commitment) to implement the above. :)
Click to expand...
That's critical. However, there's no discounting the AF engineering expertise that Canon and Nikon have built up over the years. Can Pentax catch up by dedicating a lot of resources to the problem, or is it an arcane body of knowledge that takes time, and trial and error, to acquire?
Click to expand...
How about RICOH? Anyone know what their AF performance is like?

--
Shooting since '59 and still waiting for a keeper
 
jonska said:
Jeff Charles said:
moving_comfort said:
3) Funding (and commitment) to implement the above. :)
Click to expand...
That's critical. However, there's no discounting the AF engineering expertise that Canon and Nikon have built up over the years. Can Pentax catch up by dedicating a lot of resources to the problem, or is it an arcane body of knowledge that takes time, and trial and error, to acquire?
Click to expand...
How about RICOH? Anyone know what their AF performance is like?
Click to expand...
They likely have no phase-detection AF expertise, since they do not make a DSLR.

--
Jeff
 
Jeff Charles said:
moving_comfort said:
3) Funding (and commitment) to implement the above. :)
Click to expand...
That's critical. However, there's no discounting the AF engineering expertise that Canon and Nikon have built up over the years. Can Pentax catch up by dedicating a lot of resources to the problem, or is it an arcane body of knowledge that takes time, and trial and error, to acquire?
Click to expand...
Sometimes but not always a tech deficiency that's directly attributable to a staffing issue can be mitigated by funding.

Meaning: 30% raise + stock options = a few of their key engineers become your key engineers. :)

It's often hampered by non-compete clauses though.

.
--
Here are a few of my favorite things...
---> http://www.flickr.com/photos/95095968@N00/sets/72157626171532197/
 
Jeff Charles said:
jonska said:
Jeff Charles said:
moving_comfort said:
3) Funding (and commitment) to implement the above. :)
Click to expand...
That's critical. However, there's no discounting the AF engineering expertise that Canon and Nikon have built up over the years. Can Pentax catch up by dedicating a lot of resources to the problem, or is it an arcane body of knowledge that takes time, and trial and error, to acquire?
Click to expand...
How about RICOH? Anyone know what their AF performance is like?
Click to expand...
They likely have no phase-detection AF expertise, since they do not make a DSLR.

--
Jeff
Click to expand...
The GRD IV has a hybrid contrast/phase detection AF system. Nonetheless, the phrase apples to oranges comes to mind.

--
My photographs on flikr: http://www.flickr.com/photos/71572637@N03/
 
James O'Neill said:
I think they have their own dedicated chips. Same as the image processing has its own chip, which is different from the CPU
Click to expand...
The CPU is responsible for I/O in the camera, so any dedicated chips will still be dependent on the CPU and a single CPU will be a bottleneck for processes that need to happen concurrently. The reason the Milbeaut supporting dual CPUs is important, is that it probably isn't worth the power compromise, or the price increase, of having dual CPUs in a body, if the main component, the Milbeaut, isn't able to use them.

Or, at least, that's what I'm thinking.

Thank you
Russell
 
Russell Evans said:
James O'Neill said:
I think they have their own dedicated chips. Same as the image processing has its own chip, which is different from the CPU
Click to expand...
The CPU is responsible for I/O in the camera, so any dedicated chips will still be dependent on the CPU and a single CPU will be a bottleneck for processes that need to happen concurrently. The reason the Milbeaut supporting dual CPUs is important, is that it probably isn't worth the power compromise, or the price increase, of having dual CPUs in a body, if the main component, the Milbeaut, isn't able to use them.

Or, at least, that's what I'm thinking.
Click to expand...
My assumption has always been that Pentax have their own SAFOX ASICs and these just signal to the the main part of the camera "focus ready", "Still focusing". This would be the only way to do it in a film camera, and I assmued that what is in the current DSLRs developed from film. It would also mean they could use the same AF in different cameras without having to develop it if they change CPU.
 
briantrying said:
Are more focus points actually required?
Click to expand...
I found this thread on the subject in the Nikon D4 - D1 / D800 Forum: http://forums.dpreview.com/forums/read.asp?forum=1021&message=40755332&changemode=1

OT: The first post in that thread mentions the Canon Elan 7E, which I once owned. That camera had eye-controlled focus-point selection, a great feature. You'd just look at a focus point, and the camera would switch to it. Apparently, it did not work well for some people, which is probably why Canon dropped it.

--
Jeff
 
James O'Neill said:
My assumption has always been that Pentax have their own SAFOX ASICs and these just signal to the the main part of the camera "focus ready", "Still focusing". This would be the only way to do it in a film camera, and I assmued that what is in the current DSLRs developed from film. It would also mean they could use the same AF in different cameras without having to develop it if they change CPU.
Click to expand...
Haven't there been firmware upgrades that have improved AF? That would mean something is programmable. A straight ASIC solution would have all the functions hardwired into the chip. If not, I still would think in order to drive power efficiencies, all the modules would share a CPU.

http://kmp.bdimitrov.de/technology/AF/index.html

"If the lens is focused to far, the two images have a greater distance on the CCD line, if it is focused to narrow these two images have a shorter distance then in the ´zero´ position. Out of this image-, phase-shifting the computer calculates the number and direction of rotations for the auto focus motor. It is not possible to stop right now from a full speed focusing to the exact point. So the speed is reduced before the right focus point is reached. The computer of the camera subdivide the number of turns to reach the right focus point. A photo interrupter in the lens count the impulses and send it back to the camera.

You might be right though, as your thoughts are pretty logical. Awaldern posted a logical electronics schematic of, I think, the K10D in the forum discussion about SR and wireless flash a ways back, that I can't now find of coarse. IIRC, it showed the AF module feeding back into CPU. But, maybe I have that wrong as well.

Thank you
Russell
 
You must log in or register to reply here.

Keyboard shortcuts

f
Forum
m
My threads
Mobile site
About
Editorial content
Cameras & Lenses
Community
All content, design, and layout are Copyright © 1998–2025 Digital Photography Review All Rights Reserved.
Reproduction in whole or part in any form or medium without specific written permission is prohibited.
When you use DPReview links to buy products, the site may earn a commission.
©GPS Media - Guides, Products, Services.
Back
Top