[Eruditass]

Interesting comment Eruditass!

No offence, (don't take it personal) but we are trying to have AF tests for the K5 but not as a comparison with other brands.
OK to compare to any other Pentax camera ... K-r, Kx, K7, K20D, K10D ....

JP

I think it's more interesting to compare them to other brands as I'm sure the K-5 and K-r are improved over their predecessors.

[dgaies]

I think it's more interesting to compare them to other brands as I'm sure the K-5 and K-r are improved over their predecessors.

But the questions remains how much better and is it worth the expense to upgrade to the newer bodies. That's not to say it wouldn't be interesting to compare the AF of the new bodies to other brands, but as the only non-Pentax DSLR I've every shot was the original Canon Rebel (300D), I'm not sure anyone would be interested in that particular comparison

[jpzk]

I think it's more interesting to compare them to other brands as I'm sure the K-5 and K-r are improved over their predecessors.

I agree that it WILL BE more interesting to compare it (K5) with other brands and I am sure some people are just waiting for that.

Also agreed that the K5 has improved over its predecessors but I don't think this would be the right place nor time to do brand vs. other brand comparing.

Don't you think that this would likely become a rather complicated-to-follow thread if we did that?

On the other hand, I would be pleased to see such comparisons ... in another thread altogether, unless people want to do it right here and now.

Cheers.

JP

[Arpe]

Believe it or not I have had sequences of about 5 photos from the K-7 all in focus of guys running not quite straight at me, AFC (on AF button held down and shutter button held down, not 5.2 fps though as it refocusses between shots). But I have found in good light, ie. sunny, it can do this way more easily than dimmer light, ie. cloudy.

So not only is it camera dependant, and lens dependant, it's also light dependant.

If anyone can try these at f2.8 so the misfocus shows better it would be good.

[Kini62]

Just a suggestion, for future tests .... maybe use a tripod?

JP

Aaaaa no. Focus is using a single point should be square on the chest. There is no way to reliably keep a single AF point on the head let alone an eye.

Gene

[falconeye]

I had my son run towards me as fast as he could, which it turns out isn't really all that fast
All taken with the K5 + DA*60-250/4 in one burst. at ~100mm, 1/1000s, f/5.6, iso200, 5-point auto AF point selection and AF-C.

Best K-5 AF.C test to date. Thanks a lot. Sweet boy btw He seemed to have fun too.

Unfortunately, some "Microsoft Windows Photo Gallery 6" software destroyed the precise time stamps in the EXIFs If the photome shutter delay is wrong by a factor 10, then I would deduce from the increment in shutter delay that the sequence took 5.3s. Which would translate into 2.6fps. Can this be correct?

The images are in focus except for the second last. This would be cited as 93% at 2.6fps. This would be a very good result if the speed was that of an adult and the focal length 200mm. But we lack comparison data for this exact situation of course. Anyway, a promising start.

I could try it next time with center point only focus, but I don't think there would be anyway I could keep the center point exactly on his eyes as he's running towards me

It's ok as you did. Center AF is ok too (and a bit better for comparisons). But no way to focus on his eyes. The AF sensors areas are much larger anyway!

More important would be to let him run until he runs by you. The AF will fail at some point and it is this point which is telling. Maybe, it takes longer than the buffer can maintain the speed. Then, it would be better to start later.

It will be interesting too to see low burst and the difference between focus and speed priority.

[dgaies]

Best K-5 AF.C test to date. Thanks a lot. Sweet boy btw He seemed to have fun too.

Unfortunately, some "Microsoft Windows Photo Gallery 6" software destroyed the precise time stamps in the EXIFs If the photome shutter delay is wrong by a factor 10, then I would deduce from the increment in shutter delay that the sequence took 5.3s. Which would translate into 2.6fps. Can this be correct?

The images are in focus except for the second last. This would be cited as 93% at 2.6fps. This would be a very good result if the speed was that of an adult and the focal length 200mm. But we lack comparison data for this exact situation of course. Anyway, a promising start.

It's ok as you did. Center AF is ok too (and a bit better for comparisons). But no way to focus on his eyes. The AF sensors areas are much larger anyway!

More important would be to let him run until he runs by you. The AF will fail at some point and it is this point which is telling. Maybe, it takes longer than the buffer can maintain the speed. Then, it would be better to start later.

It will be interesting too to see low burst and the difference between focus and speed priority.

I have a feeling I know what the 'Microsoft Windows Photo Gallery 6" issue was I forgot to check off "save orientation" in the menu so the pictures needed to be rotated 90 degrees. Instead of doing it properly, I just selected all 15 and rotated them in windows explorer window. I have a feeling that's why the EXIF info is messed up. I still have the originals so I can repost them later with the EXIF info properly intact.

5 seconds sounds about right. From where I was standing to the slide was about 10-12 meters and I doubt he can run faster than 2 m/s (also, the focal length was closer to 100mm)

I think the next shot in the sequence (which I didn't post) was out of focus... however I think he might have hit the MFD of the lens as he literally ran to me, not by me.

Anyway, I'm sure I'll have many more chances to try it out with different settings... my son seemed to be a willing participant

[falconeye]

FWIW they were skating - not walking. Little faster movement.

But I wouldn't know how to keep a camera hand held with center point on eyes while the subject is moving. I perhaps don't understand (which is very possible), but can't imagine that level of physical control. Especially at over 200mm.

Unfortunately, not fast enough in such bright light conditions.

The subject often can't be held in a single AF spot. This is why the new 5 spot AF mode is helpful.

Your test is interesting. The AF seemed to struggle when the biker disappeared behind a tree. We don't have a "tree in focus" shot. Can you describe a bit what exactly it did?

[falconeye]

Anyway, I'm sure I'll have many more chances to try it out with different settings... my son seemed to be a willing participant

Seems so

No need to repost the images. Just the time stamps are messed up. The EXIF file change time stamp (the one which includes seconds) of the first and last would suffice. Not the EXIF image time as it doesn't include seconds.

Was this the 60-250mm? Then just do it with 200mm and f/4 next time. If you need higher speed, he could hold the camera and you run

[dotchoucou]

The best would be to try the camera at an indoor shooting range place.
The system of the target coming back towards the shooter to check the accuracy would be perfect (constant speed) and the light should always be the same.
If one can put the camera on tripod with an angle it could be a nice testing setup I guess.

[dgaies]

Seems so

No need to repost the images. Just the time stamps are messed up. The EXIF file change time stamp (the one which includes seconds) of the first and last would suffice. Not the EXIF image time as it doesn't include seconds.

Was this the 60-250mm? Then just do it with 200mm and f/4 next time. If you need higher speed, he could hold the camera and you run

I already replaced them, no big deal

The time difference between the first and last file was 3 seconds, per the "DateTimeDigitized" tag. I suppose without any more precision that means we know it was somewhere between 2-4 seconds. However since there were 3 files tagged with the same time stamp (at 50 seconds) and the last 2 with the same time stamp (at 53 seconds), it is probably fair to say 3 seconds is pretty closer to the actual amount of time elapsed for the 15 frames.

This was taken with the 60-250. I'll have him stand farther away next time and do it at 200mm and f/4... as for me running, no one wants to see pictures of that!

[falconeye]

I already replaced them, no big deal

The time difference between the first and last file was 3 seconds, per the "DateTimeDigitized" tag. I suppose without any more precision that means we know it was somewhere between 2-4 seconds. However since there were 3 files tagged with the same time stamp (at 50 seconds) and the last 2 with the same time stamp (at 53 seconds), it is probably fair to say 3 seconds is pretty closer to the actual amount of time elapsed for the 15 frames.

This was taken with the 60-250. I'll have him stand farther away next time and do it at 200mm and f/4... as for me running, no one wants to see pictures of that!

50 ... 53, so the shutter delay must be something else. I agree that 3.0s is the best guess. Which would translate to 4.7fps which would be pretty good.

As for the 200mm: no, not farther away. Same distance but at 200mm/250mm would be perfect if you include the last shots near you (he shouldn't slow down). The last ones may be parts of him only (if you're lucky, his face ). And the last ones may not be sharp. But the last sharp one is a winner.

[jpzk]

Believe it or not I have had sequences of about 5 photos from the K-7 all in focus of guys running not quite straight at me, AFC (on AF button held down and shutter button held down, not 5.2 fps though as it refocusses between shots). But I have found in good light, ie. sunny, it can do this way more easily than dimmer light, ie. cloudy.

So not only is it camera dependant, and lens dependant, it's also light dependant.

If anyone can try these at f2.8 so the misfocus shows better it would be good.

PLeased to know that!

Unfortunately for me, and that is because my technique sucks, I am not that lucky with the K7.
Whether the K5 will solve my problems, I don't think so but hopefully I will get it right one of these days.

JP

[jpzk]

Aaaaa no. Focus is using a single point should be square on the chest. There is no way to reliably keep a single AF point on the head let alone an eye.

Gene

Is that so?
OK, forget about BIF with focus on the eyes then, right? ...

JP

[jrforman]

Your test is interesting. The AF seemed to struggle when the biker disappeared behind a tree. We don't have a "tree in focus" shot. Can you describe a bit what exactly it did?

I can't really say what it did. This was shot while on a walk with my dog - overcast early morning light - I saw a biker and thought, "let me try this". No real thought to it at all. What I can say is that there isn't a frame missing. I was panning with the camera, keeping the person centered, and it seemed to skip over the tree - perhaps couldn't get a focus until I was beyond the tree?

I would be the first to say that we need better testing methods. And I would like to try with the five point method you mentioned. And with my dog at home.