[falconeye]

I have 1.02 firmware.
Download k5_e102w EXE, fast and secure downloading from easy-share.com

ogl, thanks for the link. Actually, I have a spare copy too. But would it install over the newer one? I admit I actually didn't try it out ...

[falconeye]

Hope you didn't use the automatic horizon correction in which case the camera needed calibration...

By the way Falk . On your nice pic of the sunset the horizon is not level. Something wrong with the electronic level on your K5 ?

The horizon is unlevel by 0.35°. I agree it is visible and the eye's ability to see this is stunning. I didn't use the horizon correction. But would it be exact enough to bring down 0.35° to say, 0.1°? The photo upload was a quick shot to accompany the blog article and I skipped the horizon correction

[falconeye]

An interesting finding Falk.
In real terms there is little noticeable improvement from your tests.
I'm going to be trying out my own K-5 v1.02 before and after firmware to see the difference myself.

Great!
But maybe, change something from my scenario. Like using a dark background. I imagine a black background would be too much either.

[falconeye]

What really matters, AFAIC, is how the K-5 performs with 1.03 in absolute terms. Comparisons to 1.02 might be interesting in order to find out what Pentax did, but isn't it much more relevant to determine in what situations 1.03 still fails?
[...]
I'd like to learn how the K-5 compares to other cameras like the K20D or K-7 in terms of front-focusing in low light. The question for me is: Is the K-5's AF worse than that of previous Pentax cameras in some situations or is it not?

You're right.
However, what actually bothers me is that the K-5 AF locks (and pretty fast so) in situations where it is unable to aquire the correct focus.

It would be a big improvement if the K-5 just gave up then. After all, a photographer needs to trust his tool.

I wouldn't be surprised to learn that other cameras like K-7 or competitors actually won't lock focus either in similiar situations then.


It cannot be a problem for the firmware to know that colorimetric info is insufficient and make the AF give up. Or at least, to make this behaviour a menu option.

The K-5 meter gives up too: there are longest exposures the meter cannot exceed (it is -1EV at f/1.8 and +2.3EV at f/5.6). Interestingly, this is about where the focus shift happens with my white target. I.e., the firmware knows where the meter operates out of spec. It doesn't create random meterings but conservative ones (leading to underexposure). So, why not make the AF algorithm include code to fall back to a conservative choice (like no lock) in the same situation?

It is ridiculous that the AF continues to lock up to 4 EV below where the meter still works... (because it obviously needs the meter for a correct focus).

[deejjjaaaa]

You're right.
However, what actually bothers me is that the K-5 AF locks (and pretty fast so) in situations where it is unable to aquire the correct focus.

It would be a big improvement if the K-5 just gave up then. After all, a photographer needs to trust his tool.

I wouldn't be surprised to learn that other cameras like K-7 or competitors actually won't lock focus either in similiar situations then.

the right way will be to provide an option in setup and let owner decide whether he wants AF to lock regardless or give up and signal that correct focus probably can't be achieved.

[deejjjaaaa]

ogl, thanks for the link. Actually, I have a spare copy too. But would it install over the newer one? I admit I actually didn't try it out ...

I think you can edit the firmware number in bin file using some hex editor from 1.02 to 1.03 and install 1.02 that way...

[Duplo]

I know this may be a stupid question.

But have you tested more than one K-5?

Is it possible that the difference between the positive experience I have had with the 1.03 firmware and yours is down to a a faulty camera body or lens on your end?

[jbinpg]

Indeed, that is the only question remaining for me as well. Replicability is part of the scientific method so hopefully Pentax is following up Falk's findings with a whole slew of cameras to determine if/how much variation there is among K-5 bodies.

Jack

[the swede]

You're right.
However, what actually bothers me is that the K-5 AF locks (and pretty fast so) in situations where it is unable to aquire the correct focus.

It would be a big improvement if the K-5 just gave up then. After all, a photographer needs to trust his tool.

I wouldn't be surprised to learn that other cameras like K-7 or competitors actually won't lock focus either in similiar situations then.


It cannot be a problem for the firmware to know that colorimetric info is insufficient and make the AF give up. Or at least, to make this behaviour a menu option.

The K-5 meter gives up too: there are longest exposures the meter cannot exceed (it is -1EV at f/1.8 and +2.3EV at f/5.6). Interestingly, this is about where the focus shift happens with my white target. I.e., the firmware knows where the meter operates out of spec. It doesn't create random meterings but conservative ones (leading to underexposure). So, why not make the AF algorithm include code to fall back to a conservative choice (like no lock) in the same situation?

It is ridiculous that the AF continues to lock up to 4 EV below where the meter still works... (because it obviously needs the meter for a correct focus).

Very good post! They should have just make the AF to give up when there's not enough light to meter. As it is now, we are made to believe the AF should work correct because its still active in poor lighting. Then we get angry when it doesnt focus correct.

Sounds to me like bad marketing

[Class A]

It is ridiculous that the AF continues to lock up to 4 EV below where the meter still works... (because it obviously needs the meter for a correct focus).

The meter is needed in certain lighting situations but there should be a range of colour temperatures where the AF should obtain (almost) correct focus even in very low levels.

It would thus be a shame to lose the AF sensitivity in all situations just because it produces incorrect focus in some situations.

I wonder whether the power spectrum distribution of your blue light bulb is recognised as "cool" light by the colorimetric meter. Maybe the camera expects the power spectrum distribution of actual daylight to use the respective AF calibration curve? I have a hunch that while your three lights stimulus seems logical, it may not tell us the full story about the AF's dependency on different spectral power distributions.

It appears logical to me that the AF module's sensitivity should be usable under some light. Maybe the "fall back" value for the colour temperature when the colorimetric sensor turns blind is

• chosen incorrectly
• outside the range you cover with your three lights
• not used because of a remaining bug in the firmware
• dependent on a particular K-5 copy (unlikely, but would explain contradicting observations)

But have you tested more than one K-5?

The report clearly states that only one K-5 has been tested. It seems safe to exclude the lens from being faulty. It is possible that various K-5 copies show different behaviour. In this case the pre-V1.03 FF issues would have been caused by both faulty hardware and suboptimal firmware. Seems a bit unlikely, but is of course possible.

[Duplo]

The report clearly states that only one K-5 has been tested. It seems safe to exclude the lens from being faulty. It is possible that various K-5 copies show different behaviour. In this case the pre-V1.03 FF issues would have been caused by both faulty hardware and suboptimal firmware. Seems a bit unlikely, but is of course possible.

I know it does, but had to ask since I have had a positive experience with the upgrade.

The lens may be OK and it may not, just as the body may or may not be OK.
But the combination of the two specific components may not be working.

What i asked was if this possibility has been tested and confirmed irrelevant, I think it is a fair question to ask, given the conclusions reached.

I wonder if things would improve if the combo were sent to entax for a check.

[dankoBanana]

Pentaxs lack of ability to tackle this problem (and the K-r FF problem, and the K-5 flash issue) is making me reconsider the purchase of a K-5 and I'm even thinking about switching systems ... thats lame :ugh:

[Smeggypants]

The report clearly states that only one K-5 has been tested. It seems safe to exclude the lens from being faulty. It is possible that various K-5 copies show different behaviour. In this case the pre-V1.03 FF issues would have been caused by both faulty hardware and suboptimal firmware. Seems a bit unlikely, but is of course possible.

Well I'm exchanging my K-5 this week now my Retailer has new stocks. I'm getting a 4xxxxxx serial replacement. In about 2 weeks I@m hoping to buy a 2nd K-5.

this means I will have tested 3 different K-5s in the not too distant future. I actually hope my original model, which is now parcelled up ready for collection tomorrow,was a rogue copy.

[Ash]

I have a refurbished 38xxxx model from Japan.
No stains (seen yet) and no noticeable FF in low light.
Pre v1.03 there was minimal FF to speak of.
Post v1.03 I notice no FF on my low light shots so far.
One issue, Falk, is that most people want to use their K-5s rather than devote time and energy to testing it. You do it and do it well, and this I believe is a scientific task that is of significant financial value. Most of us would not be skilled enough to perform the correct tests under controlled conditions to reproducibly and consistently quantify FF under those test conditions.

[ogl]

I'll call it the dark is daylight hypothesis.

Falk, I'd like remind you that the idea about Pentax AF problems because of CA in AF sensor is idea of GordonBGood and oleg_v.
And it's very reasonable idea.