[VHDEL]

Definetly not scientific.

I had a chance to try me fathers K20D yesterday.
Both cameras I used the same DA* 16-50 F2.8 set at 50mm
Not the best lens, but it's all I had using SDM.

I've offen thought my copy of the K10 has substandard AF performance and the
tests confirmed this for me.

I had my 4 1/2 yr old son run towards me, twice for the K10 & K20 just to see if there is some consistency with my ability using both cameras.
For a pint sized midget, he runs faster than my 6 yr old daughter.

Cameras were hand held using the following setting:

Center point focusing
AF-C
3 FPS cont shooting mode
SR on
Av mode
Matrix metering
AWB
JPEG fine
Battery at least 80% charged on both cameras

The shutter speed on both cameras varied between 1/1500sec to 1/2000sec for all the images taken.

The K10 on both sequences had a hard time trying focus on target, always towards the rear.
I can assure you that all 4 tests the focus point always remained on his chest.

The K20 did a far better job throughout both sequences, although some were OOF.

I only wish I had of hired the Canon 40D 17-55 F2.8 combo just to compair.
I still may just do that to satisfy my curiosity.

I have put a link to Photobucket although the file size is limited, at least you get an overview of the sequence.

The other link has the larger files.
All I have done to the origional images is used Photoshop & resized for web
(jpeg medium).
At least with these's files you can zoom in on the target in the distance without to much pixelation as you get with photobucket images.

I'll let you draw your own conclusions, for me the K20D is a major improvment.
But is it comparable to the other major players.

Photobucket images.
VHDEL/AF Test/K10D Set 1 - Photobucket - Video and Image Hosting
VHDEL/AF Test/K10D Set 2 - Photobucket - Video and Image Hosting
VHDEL/AF Test/K20D Set 1 - Photobucket - Video and Image Hosting
VHDEL/AF Test/K20D Set 2 - Photobucket - Video and Image Hosting

Larger images
Index of /~del/Af Test

Cheers

Del

[BrendanPK]

Well I know this so far, evening/night LAX games have always been fun for me, and progressivly the AFC has gotten better from the *ist-D though IMHO the *ist-D could beat the K10D sometimes when I used the AF-button and the shutter release just for that. (I find the K**D design for AF-button placement hard to reach when shooting Vert, that are most used for sports)

I had my first chance w/ the K20D the other day and I must say I got WAY more (99.5-100% perfect) in-focus shots and thoes that were acceptiable (85% there) for news-print (not a mag spread) than I ever did w/ the K10D or *ist-D. Add to the high ISO and WOW.

[RiceHigh]

Thanks for sharing. Your results shared are very useful and interesting!

As a short remark, the difference and amount of OOF for the K10D pics are obvious for your son running near the camera, at closer distance whereas for far away object, the difference is small - but that's very normal anyway.

Once again, it has been proved that the SAFOX VIII of the K20D has been much improved over the one in K10D, thanks. (But WHY don't Pentax tell us so? SAFOX VIII != SAFOX VIII ?!)

Definetly not scientific.

I had a chance to try me fathers K20D yesterday.
Both cameras I used the same DA* 16-50 F2.8 set at 50mm
Not the best lens, but it's all I had using SDM.

I've offen thought my copy of the K10 has substandard AF performance and the
tests confirmed this for me.

I had my 4 1/2 yr old son run towards me, twice for the K10 & K20 just to see if there is some consistency with my ability using both cameras.
For a pint sized midget, he runs faster than my 6 yr old daughter.

Cameras were hand held using the following setting:

Center point focusing
AF-C
3 FPS cont shooting mode
SR on
Av mode
Matrix metering
AWB
JPEG fine
Battery at least 80% charged on both cameras

The shutter speed on both cameras varied between 1/1500sec to 1/2000sec for all the images taken.

The K10 on both sequences had a hard time trying focus on target, always towards the rear.
I can assure you that all 4 tests the focus point always remained on his chest.

The K20 did a far better job throughout both sequences, although some were OOF.

I only wish I had of hired the Canon 40D 17-55 F2.8 combo just to compair.
I still may just do that to satisfy my curiosity.

I have put a link to Photobucket although the file size is limited, at least you get an overview of the sequence.

The other link has the larger files.
All I have done to the origional images is used Photoshop & resized for web
(jpeg medium).
At least with these's files you can zoom in on the target in the distance without to much pixelation as you get with photobucket images.

I'll let you draw your own conclusions, for me the K20D is a major improvment.
But is it comparable to the other major players.

Photobucket images.
VHDEL/AF Test/K10D Set 1 - Photobucket - Video and Image Hosting
VHDEL/AF Test/K10D Set 2 - Photobucket - Video and Image Hosting
VHDEL/AF Test/K20D Set 1 - Photobucket - Video and Image Hosting
VHDEL/AF Test/K20D Set 2 - Photobucket - Video and Image Hosting

Larger images
Index of /~del/Af Test

Cheers

Del

[VHDEL]

Thank's for your replies.

I forgot to mention that all the test were at F2.8.
I did have a feeling that the closing speed between me & the target was going to be greater than if I had shot a much longer focal lenght.


It just so happens my father has just received his DA*50-135 today.
I would like to try that lens also @ the 135mm end, although I don't like running up the shutter actuations on me dads new cam.

Cheers

[stewart_photo]

...(snip)... I've offen thought my copy of the K10 has substandard AF performance and the tests confirmed this for me. ...(snip)...

And what norm are you using to judge the K10D's AF performance as substandard? One camera and one or two tests are not a norm. Regardless, there are always limits to how well any particular camera can focus on a fast moving subject. That is just as true for the K20D as it is for the K10D. Continuous autofocus works best with gradually changing focus points, such as a sports figure running across in front of you with the overall distance from the camera not changing all that quickly from moment to moment. Continuous autofocus also works well with distant subjects where a wider depth of field allows for relatively gradual focal point changes.

However, fast moving subjects, especially those moving directly towards and away from the camera where focus points change most quickly, can easily fool continuous autofocus because the camera focuses where the subject was shortly after the the shutter button was pressed while a fast moving subject can easily move outside that focus point by the time the exposure is made several hundreds of a second later. This is why your K10D focused behind your son running quickly towards you. The K20D might handle these situations slightly better then the K10D (so you claim), but the K20D certainly has limits as well.

Several years ago, manufacturers promoted "predictive autofocus" as a solution to this common problem. With predictive autofocus, the camera tried to predict where a fast moving subject would be at the moment of exposure based on where and how fast the subject was moving when the shutter release was pressed. However, it wasn't long before the exposure delay was shortened enough to outstrip even the capabilities of predictive autofocus, with this feature now dropped from most cameras today.

Of course, none of this means there will never be times whan a fast moving subject outpaces a camera's ability to track. And the occasional fast moving subject outpacing a camera's ability to track doesn't necessarily mean the AF performance of that particular camera is substandard. Instead, it simply means alternative tactics, such as pre-focusing or another vantage point, are required for these relatively rare situations.

stewart

[Arpe]

Thi is good news a I'm about to pull the trigger on a K20D. Glad I skipped the generation of the K10D

[VHDEL]

And what norm are you using to judge the K10D's AF performance as substandard?
stewart

Because I have also tested it against my fathers K10D as well.
His K10 was bought in Sept 2007
My K10 was bought in Jan 2007
Overall his K10 camera has a slightly better AF than mine, although its not as good as the hit rate on his new K20.

Cheers

[rfortson]

I never really tested my K10D but I shot a road race with the K20D and the 50-135 and I only had 2 shots out of ~50 that were not sharp. See my post here with a couple of example shots. I was very impressed with the AF performance of the K20D.

[VHDEL]

Nice shots there Russ,

The 50-135 seems like a great lens.

I do belive the K20 has far more improvment using AF-C which was reasons for my own personal test.
There are most probably better copies of the K10D than mine for AF, but the difference between "MY K10D" & the K20 was chalk & cheese.

Del

[rfortson]

Nice shots there Russ,

The 50-135 seems like a great lens.

Thanks, I've been very pleased with the lens, on both bodies.

There are most probably better copies of the K10D than mine for AF, but the difference between "MY K10D" & the K20 was chalk & cheese.

I agree. There does seem to be significant sample variation with the K10D AF. I never noticed huge problems, but I've read posts where people couldn't track their kid running straight towards them - every shot would be OOF.

[Jodokast96]

There does seem to be significant sample variation with the K10D AF. I never noticed huge problems, but I've read posts where people couldn't track their kid running straight towards them - every shot would be OOF.

I think you are right there. Shooting my kids soccer, I've never missed a shot that wasn't from me moving the focus point off of the subject. Until I found this forum, I would have never thought there were any issues with the focus tracking, based on my experiences.

[tpeace]

[QUOTE=BrendanPK;197196](I find the K**D design for AF-button placement hard to reach when shooting Vert)

Completely agree with you on this. And its worse with a battery grip in place - almost impossible to reach.

re-t