[Urkeldaedalus]

The Leica M9 costs $7,000 and it doesn't even have autofocus or a movie mode. What a ripoff!

[reeftool]

Someday I will understand how the number of AF points affects me who always uses center point focus.

I agree completely. I grew up in photography using a manual camera with a split prism in the center and I always focus in the center. If I want the focus off center it is far simpler to use the focus lock feature in af.s. Focus on what I want and then re-frame with the shutter still half pressed. What could be simpler? This is a perfect example of a retarded feature. If exists mostly to spice up the spec sheet. If they did away with it everybody would scream but in reality, hardly anybody uses it. It has ruined many shots for me because it is easy to accidently move that dial in routine handling of the camera and the little red square showing up in the wrong place doesn't always register right away.

[Blue]

The Leica M9 costs $7,000 and it doesn't even have autofocus or a movie mode. What a ripoff!

By god, its an AF system speed Pentax can beat.

[Class A]

Focus on what I want and then re-frame with the shutter still half pressed. What could be simpler?

Have you not seen the posts in this thread that explain the utility of manually selectable AF points?

What is wrong about avoiding focus errors (the closer you are, the bigger the problem) introduced by recomposing and being able to focus while following a moving subject that is off-centre?

[shadmanaj]

"Many AF points" does not imply "automatic AF point selection". See my post above for only one useful application where you manually select one AF point. There are more useful applications.

Yes, I did read it, and I absolutely agree with you. I am just stating that lesser no. of AF points on the K-5 does not necessarily mean it will have inferior AF performance than an 18 or 19 point system.
I have read about Canon 7D and Nikon D300s having similar AF performance, despite the vast difference in AF points.
I doubt we'll see Focus lock or zone focus soon, though.

[knumbnutz]

Hi Adam,
I dont think it makes tooo much difference the amount of points.
Most use centre point but those who do shoot sport might have wanted more, but if it works whether that because its fast ,predictive or both, only time will tell.

IF the K-5 is the same leap above the K-7 as the K10 was to the K20 then it WILL be a great little camera.

Surely they must need some real world testers in Australia ok i'll put my hand up....just for the good of the people of course !!

Cheers Neil

[Ivan Glisin]

You pick a focus point, usually the center point, as the starting point. You place your subject (e.g. a bird or a soccer player or something) there and tell the camera to start tracking. Now when the subject moves across the viewfinder the camera will automatically switch which AF point it uses to keep focus on the subject. Soccer player runs to the right, the camera will start switching to the AF points to the right of center, etc. Then there are algorithms and settings that determine how long it will follow the predicted path of the subject before giving up, so if your bird flies behind a tree or a telephone pole, or your soccer player runs behind someone on the field, the camera will keep trying to predict where he/she was headed so that when they reappear it can reacquire focus.

Unfortunately, this could only work for an isolated subject on a simple background. But for this to work reliably with background and foreground distractions there should be much more information available to that algorithm then AF system can provide.


I feel that I can safely make this claim because several years ago I was experimenting with data available in the AF system and provided by the lens in order to devise some sort of software-based focus limiter (especially for macro and telephoto lenses). I gave up eventually and one of the main reasons was because available information was not consistent between the lenses (plus several other problems). But this is what I have learned:


Each individual AF point is actually not a point, it's an AREA and there is no way for a relatively large area to reliably track any small object. Next, distance information is NOT associated with individual AF areas, there is just in/out of focus status. Overall distance scale is transmitted by lenses but it is very rough (like "closeup", "near", "medium", "far" and "infinity" if I remember well). In addition, one additional piece of information could be determined in AF-C mode on how correction is acquired, say, if reacquiring AF with new status "far" required forward correction "medium-to-far" or backward "infinity-to-far". And that's pretty much all.


Although that is enough information to track a subject moving towards or away from the camera or across the frame with plain background, it is absolutely NOT enough to track subjects moving across the frame with background or foreground distractions. The functionality as described in your post would require either pattern analysis or more accurate AF distance associated with each AF point, both not available in any AF system in consumer cameras today.

[opiedog]

while the 18/19 point AF rumor gave me something to be excited about the k5, the reality of 11 pt brought it back to earth.

but, the # of AF points doesn't matter as long as AFC works and tracks. 18 or 51 points may provide additional benefits, so what? works for some folks, may not work for others.

my biggest concern is the MSRP. as others have stated IF the rumors of the D7000 is true, spec wise it is neck-and-neck with the k5 at $500 LESS. THAT IMHO would be a problem.

again these are all rumors. i'll wait 2 weeks to establish my own judgement.

[Class A]

But for this to work reliably with background and foreground distractions there should be much more information available to that algorithm then AF system can provide.

The Nikon D300s uses color and brightness information for subject tracking. I haven't used that camera, but I guess this could go quite some way to track subjects.

[hcarvalhoalves]

Well, but "X" AF points with "Y" tracking algorithms can't beat the human brain. You will never have a camera that locks focus on a subject and keeps chasing it forever, no matter how many dollars you are willing to pay.

I happen to know by experience, as I'm a software developer and have worked with computer-assisted vision systems already. It only works under limited constraints: you need the correct light angles to being able to detect borders, you need the correct colors to provide enough contrast, and subject can't move faster than you processing speed, otherwise you can't follow the difference between frames. And this is under controlled variables (industry), imagine shooting photos in the wild.

So it doesn't matter, there's a limit for automagic focusing systems. Those scenarios of "chasing a bird" or "chasing a soccer player" going thru obstacles and the system still locking focus are unreal. What the computer really does is keeping the same focus for some frames in the hope the subject will get back into frame again. Don't believe it's actually "chasing" the subject and "predicting" where it'll show up.

[Class A]

Don't believe it's actually "chasing" the subject and "predicting" where it'll show up.

Of course predictive AF does that. It uses the previously measured speed and trajectory of a subject to extrapolate where it should appear next (approximately). If predictive AF only reacted instead of anticipated, it would always be too slow.

I appreciate your personal software development experience but at the same time I doubt that you know how the Nikon algorithms look like and how useful the D300s AF system is in practice. Of course there are limitations to any technology but I'd be surprised if didn't beat continuous Pentax AF buy a large margin.

A human brain is able to follow a subject through a complex scene but how many of us are able to manually keep it in focus at the same time?

In any event, this automatic AF business is of limited interest to me personally. I'd like more (and smaller) AF areas for manual selection in order to avoid recomposition/re-framing and to be able to pan subjects that aren't in the centre.

[hcarvalhoalves]

Of course predictive AF does that. It uses the previously measured speed and trajectory of a subject to extrapolate where it should appear next (approximately). If predictive AF only reacted instead of anticipated, it would always be too slow.

No, simply because a computer has no particular idea of what a "subject" is, neither can it "measure speed and trajectory", which is impossible without stereo vision. Although the marketing name is "predictive", it's not exactly how you're thinking it is. It's just a threshold function that decides if it should keep focus or start chasing focus, and in which direction. Let's say it's a "good guess" for when you actually press the shutter.

I appreciate your personal software development experience but at the same time I doubt that you know how the Nikon algorithms look like and how useful the D300s AF system is in practice.

It's possible, in the particular case they are doing something completely novel amongst the entire industry. Which I doubt, otherwise they would be making more money licensing the technology instead of making cameras. Computer-vision is considered the computer science problem of the century, and hard enough even when you have clusters of computers available, having to solve it with limited processors (like the ones you embed in cameras) doesn't make it any easier.

I'm not claiming that AF-C on D300 isn't better than Pentax one either, it's very possible that it is. Still, I reinforce that everybody should take any claims of "automagic" focusing systems with a grain of salt. The systems tend to be way simpler than how the consumer tend to believe they are in reality.

[rawr]

Agree on the difficulty inherent in tracking and 'predictive' AF.

I remember reading with dismay a recent reviewer of the 7D or something who was shooting a rugby match, and was claiming the camera AF was able to pick and follow the individual player the photog was interested in as he broke out of a scrum and ran off with the ball.

Hmm, I thought. How would the camera know (a) the player the photog was interested in, (b) know to focus on him in preference to all the other players on the field around him in front/behind/to the sides, (c) be able to intelligently keep tracking him when every other player on the field was running towards him from all directions and he himself was running into other players.

Maybe such wizardry is possible nowadays, but I am sceptical.

[Arpe]

So it doesn't matter, there's a limit for automagic focusing systems. Those scenarios of "chasing a bird" or "chasing a soccer player" going thru obstacles and the system still locking focus are unreal. What the computer really does is keeping the same focus for some frames in the hope the subject will get back into frame again. Don't believe it's actually "chasing" the subject and "predicting" where it'll show up.

One of those systems, Canon 7D I think, you can tell the camera how long to wait before refocussing if it loses track. So if you're panning with the player and another player crosses the shot, the camera will not try to refocus on the new player, unless you stay aiming at the new player for longer than your predetermined time. Damn fine idea.

[Freak]

I would like to see Pentax confirm Lens Profiles for Adobe & Others, as well as GUARANTEE that tethering and the remote control utility is compatible with the new K-5... Without this Pentax do not listen to the iPhone/iPad and LR tethering threads. Without those additons no upgrade for me from my K-7.

And while we are there: The AF Lamp is useless on the K-7, i want (LISTEN PENTAX) On/Off and possibily AUTO for ALL menu options..

Auto Assist Lamp - On/Off/Auto
DFS - On / Off / Auto
Video = 24/25/30fps Video should be standard. Not 25 or 30fps, all freaking 3 standards, it's really not that hard.

These little niggles are simple and often Firmware fixable, it's a complete joke if Pentax have not addressed them.