[Kguru]

Hi, I posted this at DPR but got the feeling I'm getting cold shoulder treatment overthere (not sure what I did though) so I'll try my luck here.
=====

Godfrey wrote:
>
> In my testing, the manual in-focus confirmation in Pentax DSLRs is
> good enough to give excellent results with lenses at f/4 or smaller
> openings. It is NOT accurate enough to critical focus at f/1.4 to f/2.8

Godfrey ... hope you don't mind me quoting this old post of yours.
I've just found Godfrey's statement above spot on with my FA43.

Light was good but for testing I shot at f/2.8 thus shutter 1/2000s. First shot: focused at infinity before auto-focused at object (a black on white signboard at ~15ft). Second shot: focused at a near point before auto-focused at object so to make it focus from opposite direction. Thrid shot: turned focus ring away before refocused manually at object. The manual shot was better focused than both 1st and 2nd.

Please help explain for me why: light was good, aperture was open, the auto-focus mechanism should have no problem seeing the object clearly (meself no 20/20 had no problem seeing it!) so why couldn't it focus well at f/2.8 yet ok at higher numbers? Thanks.

---

Last edited by Kguru; 09-17-2007 at 06:55 PM.

[FotoPete]

Do you notice any decentering in the images at large aperture? Or perhaps your camera had a certain AF setting that wasn't agreeable with your scene? Like was the camera parallel with the signboard? Maybe the camera's AF spot coinceded with something else.

Anyhow this kinda reinforces the point that a camera can't always take a photo for you and is only making decisions based on fulfillment of certain parameters. Manual is the way to go!

[Ole]

Autofocus isn't 100% accurate, no camera is. The error is normally masked by the depth of field, but at large f-stops like 1.4 or 2.0 the depth of field is very shallow and may not in all instances be sufficient to hide the error. Use manual focusing if spot-on focus is required with large f-stops.

[Kguru]

Do you notice any decentering in the images at large aperture? Or perhaps your camera had a certain AF setting that wasn't agreeable with your scene? Like was the camera parallel with the signboard? Maybe the camera's AF spot coinceded with something else.

I don't think it's decentering since decentering would have occured with manual focus too wouldn't it?

I use center AF, and while on this let me ask about the AF spot. At center of viewfinder there's this rectangular panel for center AF; the red dot, when it appears, only takes up about 1/10th of this panel. Theoretically where should I aim the focus point on, where the red dot is or anywhere in the center panel?

[Kguru]

Autofocus isn't 100% accurate, no camera is. The error is normally masked by the depth of field

Thanks for your advice Ole ... sounds very plausible.
But at the same time it's disappointing that 21st-century electronics cannot match a middle age's on-the-way-down eyesight.

[roentarre]

Theoretically where should I aim the focus point on, where the red dot is or anywhere in the center panel?

Switch to selective AF and dial the wheel to move to the AF point you prefer.

In general, Fa 43 does not encounter the AF problem unless in the dark lit environment. Again, photography still requires a bit of work and manual focus is sometimes much quicker than a lens hunting constantly.

When I take this shot, it hunts like hell. Manual focus saves me

[blende8]

Autofocus isn't 100% accurate, no camera is. The error is normally masked by the depth of field, but at large f-stops like 1.4 or 2.0 the depth of field is very shallow and may not in all instances be sufficient to hide the error. Use manual focusing if spot-on focus is required with large f-stops.

Or, you can send lens+camera for adjustment.
Perhaps there is a small amount of front- or backfocus.
Have you tested for this?
This can be adjusted.

Also it is possible that there is some spherical abberation, which causes the focus plane to be somewhere else at f2.8 compared to f5.6.

[Kguru]

Perhaps there is a small amount of front- or backfocus.
Also it is possible that there is some spherical abberation, which causes the focus plane to be somewhere else at f2.8 compared to f5.6.

I have tested for FF or BF problem, by shooting a newspaper page at a slant angle. On the odd one or two occasions focus was out but probably due to refocusing travel too small for it to act accurately. I was happy with the rest of the tests.

It may well be similar to what I read about the DA21 focusing at f/5.6 etc ... so let me read up again and do some more tests. Thanks.

[Kguru]

Roentarre ... thanks for the really nice pics.
I'm thoroughly impressed by your lens list, certainly gives you a "voice of authority"

[jeffkrol]

Roentarre ... thanks for the really nice pics.
I'm thoroughly impressed by your lens list, certainly gives you a "voice of authority"

Always best to read this:
http://doug.kerr.home.att.net/pumpkin/AF_accuracy.pdf
There is also a reference somewhere about the upper level Canon's using a different AF sensor for large aperature lenses and the camera will switch from one to another when needed. I have to find that reference again..
No excuses but some background.