[WPRESTO]

IR is basically radiant heat. Are they developing something beyond "face recognition" that will identify a warm body = a person in the picture? Awful lot of things could fool such a system, such as a dark boulder that's been sitting in an August sun, or an incandescent ceiling light above a birthday party. But---

[Simen1]

Body heat IR is in the 8-15 mm long wave part of the IR spectrum (LWIR). Imaging sensors made of silicon are very little sensible to IR with longer wave length then 1 mm (almost visible red glowing hot) Near IR (NIR).

In the NIR part of the spectrum normal lenses work almost like in the visible red part (with some focal skew), but in the LWIR part most glass lenses are very little transparent. It needs special quartz glass to be transparent. These also has very different refraction indices then in the visible part, causing very wrong focus if you try to focus with LWIR for visual imaging. NIR would work better but still be a little off. That is why I think (N)IR is just a first approach focal method. Precision focus must be done in the visible part of the spectrum.

[Dan Rentea]

And by the way, I'm not claiming that improvements were "so big". I'm merely disagreeing with your claim that they're "barely noticeable".

K-3 II autofocus is better than K-5 II autofocus when comes to tracking . I had both camera and there was a visible improvement in Af, no doubt in this regard.

- the D700 being a $3000/2500 euro body
- the 24-70 also being much more expensive than the Pentax 16-50

You are right, but try to compare the Af performance from Canon 7D (the first one, not Mark II) with the Af from K-3 II.
K-3 II can't keep up with 7D when comes to tracking, even if you mount a cheap lens on 7D. If we put 7D Mark II into discussion...then it's game over for Pentax (I'm referring only to the Af tracking performance). But if we speak about image quality, then K-3 II is better on this regard, no question about that.

So...depends on everyone's needs. If you shoot sports most of the time, then Nikon/Canon are far away from Pentax in terms of Af tracking capabilities. If you don't shoot sports, Pentax K-3 II can be a good option for the moment, and let's hope it will be a great option in 4-6 years when it will reach some maturity in the market.

[Kunzite]

K-3 II autofocus is better than K-5 II autofocus when comes to tracking . I had both camera and there was a visible improvement in Af, no doubt in this regard.

I can't contradict you, as I haven't used the K-3 II, nor the K-3 long enough to form an opinion (and I trust your assessment). I will say, though, that what I found myself really wanting in the K-5IIs regarding tracking was a way of distinguishing the subject from the background. The K-3 was the first Pentax which introduced that.

As for the comparison with 7D-series cameras, unfortunately I don't have enough relevant experience with those, either. Regardless, I didn't expect Pentax to be able to get to that level - which means camera and lens-wise - on their first try.
The K-3 is just a first and very significant step regarding advanced AF tracking. Expect progress, but Pentax/Ricoh will play catch-up for a while. It isn't easy.

Getting back to my point about price, one thing Canon and Nikon had but Pentax hadn't was a higher end, higher margin camera on which more expensive technologies can be introduced, tweaked, tested - with the results trickling down to more affordable products. The D7200's 51 points AF is the result of what was done with the D3, D300 and D700.

[Dan Rentea]

I can't contradict you, as I haven't used the K-3 II, nor the K-3 long enough to form an opinion (and I trust your assessment). I will say, though, that what I found myself really wanting in the K-5IIs regarding tracking was a way of distinguishing the subject from the background. The K-3 was the first Pentax which introduced that.

I have 2 nephews (twins, 6 years old). I did see a big improvement on K-3 II Af tracking capabilities.

Getting back to my point about price, one thing Canon and Nikon had but Pentax hadn't was a higher end, higher margin camera on which more expensive technologies can be introduced, tweaked, tested - with the results trickling down to more affordable products. The D7200's 51 points AF is the result of what was done with the D3, D300 and D700.

That is why I said on different topics that Pentax will need at least 4 to 6 years to reach some maturity.

[Nicolas06]

Problem is, I don't want any of those lenses. I explained this earlier on.

But I don't want to hijack this thread any further. It's just that I have a seeing-before-believing view on all of this.

Then you can't really complain on the camera performance if the lenses don't follow.

I understand your point and if a beginer comes to me with action/sport needs I'll propose them to buy a D7100/D7200 rather than a K3, K3-II. This is because even if the camera improved we are not there yet... and there a lack of affordable fast performing lenses in pentax land. I think this will change but this is going to need at least 2-3 more years likely more 4-5 years until all lenses are available and the camera is fully optimized for predictive AF.

But if the beginer has other requirements, it might be very well a better choice to go for Pentax. WR is a very popular request for example.

[MadMathMind]

Huh interesting. Not sure how much this will help (since IR is shifted?), but its good that Ricoh is trying to innovate in the AF field, which was often pointed to as a Pentax weakness. Even though many of us never had problems to begin with, and some tests showed that Pentax flagships have comparable AF speeds as other cameras in the same price range

Today I heard a Nikon user complain about his AF woes. I hear it commonly from Canon users, so I don't think Canon's single AF is terribly better than Pentax's either. (The biggest weakness, up to the K-3, has been the small number of focus points.) Anyway, this Nikon shooter said that he loves his 85mm f/1.8 but has issues at f/2.0 nailing that focus. I thought it was interesting. I told him to use Live View for that sort of thing because it doesn't have the same issues.

[Canada_Rockies]

From some comments I have read on the web - including other forums (forae?) - some cameras trade off accuracy for speed. They use a "close enough" algorithm. This no doubt works when the lens is set for better than minimal depth of field, but with something like an 85/1.8 the DOF can be hair thin.

[reh321]

Today I heard a Nikon user complain about his AF woes. I hear it commonly from Canon users, so I don't think Canon's single AF is terribly better than Pentax's either. (The biggest weakness, up to the K-3, has been the small number of focus points.) Anyway, this Nikon shooter said that he loves his 85mm f/1.8 but has issues at f/2.0 nailing that focus. I thought it was interesting. I told him to use Live View for that sort of thing because it doesn't have the same issues.

How does focusing under Live View differ from focusing while looking thru the viewfinder??

[thibs]

Contrast instead of phase.
Slower but more precise.

[Na Horuk]

Anyway, this Nikon shooter said that he loves his 85mm f/1.8 but has issues at f/2.0 nailing that focus. I thought it was interesting. I told him to use Live View for that sort of thing because it doesn't have the same issues.

Good idea. And I do agree, which is why I said AF was "pointed to as a weakness", not "AF was a weakness." But we should also keep in mind that SDM did have some problems, and screwdrive is fairly loud on some lenses. And even though some Pentax cameras had problems like misfocusing in tungsten, there were cases where Canikon cameras had problems with the AF module, as well. Still, if Pentax wants to get recognition, it has to have AF that leaves no excuses to Canikon.

How does focusing under Live View differ from focusing while looking thru the viewfinder??

Viewfinder is Phase detect AF, which is the classical type. It uses AF module and has some points to select from. Points are tuned to some aperture rating
Live view cameras generally use Contrast detect, which uses CPU and simply maximizes edge contrast. It can do this for any area in frame. It is slower, but never suffers from back or front focusing, misalignment, etc. But it can still focus on the "wrong" object or it can simply not find enough contrast and will just hunt for focus. Some mirrorless cameras now have sensors with phase detect af points built-in, so that even live view would have some PD AF (but this technology is not very good so far; not as good as the highly evolved classical PD AF of DSLRs)

[traderdrew]

Today I heard a Nikon user complain about his AF woes. I hear it commonly from Canon users, so I don't think Canon's single AF is terribly better than Pentax's either. (The biggest weakness, up to the K-3, has been the small number of focus points.) Anyway, this Nikon shooter said that he loves his 85mm f/1.8 but has issues at f/2.0 nailing that focus. I thought it was interesting. I told him to use Live View for that sort of thing because it doesn't have the same issues.

Some people complain about their cameras no matter what. Im hoping this IF technology can work together with phase detection and help me nail wildlife in action.

[monochrome]

Recently Western culture has celebrated an ordered meritocracy, meaning lists of things are compared and contrasted, ranked and the Best or Top-Ranked is declared. We have been acculturated to accept only the Best consumer products, and seek rankings as part of our selection process. We have become the 'Consumer Reports' culture.

The fundamental failing of ordering everything lies in equivocal acceptance of what is 'Good', as follows:

1. Good = Standard, a universally accepted set of qualities against which all Candidates shall be judged.
2. Better = Comparative, a ranking of one Candidate against another in their relative adherence to or achievement of the Standard
3. Best = Superlative, the single highest scoring Candidate in the continuation of the orderly ranking of Candidates in their adherence to or achievement of the Standard.

Of course what is considered Good is not universally accepted, so there can actually be no Best.

For that reason it will be quite helpful in future if a Poster should say, rather, "Because I shoot primarily moving subjects, Canon AF is Best for me," than, say, "Canon has the best AF and Pentax sux," or some such.

The first adherent to this preference should be DxOMark.

[Nicolas06]

The first adherent to this preference should be DxOMark.

DxO give you every detail for each sensor they tested. They also quite clear they only tested the sensor. If is up to you to optimize your reading of DxO results depending of the level of geekness you require.

If you read first level, you can honestly just read their unified number. Sensor with significantly higher ranking tend to perform much better. For 99% of people that's more than enough.

Now if you want to mess up with the details of the sensor score because you get all the information you could ever want. But you know the sensor is just one aspect of the camera and should not be the only factor of choice. There are better sensors out there (and without surprise they score better) but there no bad sensors.

Other aspects like the camera echosystem, AF performance or ergonimics might have much more impact on the final result.

[zapp]

Nice

Wouldn't it be possible to use also the IR sensor for an hybrid AF with Time of Flight measurement through infrared laser emission ? LG has a product like this (with an hybrid AF designed by STMicroelectronics) and the laser is used to make a coarse measure of the object position before making a finer regular focusing.

No. TOF works in close range and for a limited number of sensor elements. Beyond a couple meters and off center it would be useless. For longer distance you would need to pump out a lot of energy and accuracy of small TOF is not useful. Phase detection is much faster as is. TOF sensors not necesseraly need IR rays. The approach here is to check the IR spectrum to reduce errors. Very different approach with a passive sensor. TOF is active.