[falconeye]

Thanks, Falk. Good stuff. Possible error in the summary:

2. Should probably read "in front of the sensor plane"?

Jack

Jack, you're probably right.
I'll double check and fix it in the next version.
It is correct though that the image width is too large, i.e., df>0.

Thanks for spotting this

[Tommot1965]

Jack, you're probably right.
I'll double check and fix it in the next version.
It is correct though that the image width is too large, i.e., df>0.

Thanks for spotting this

Falk

Now you have confirmed what many of us knew to be true, do you have any idea what is causing the issue, and are you confident on a firmware fix. Or is it hardware related

[falconeye]

Thanks falconeye. Wow! You're a testing machine. Your value to this community is appreciated.

Compliments aside, out with it: hardware or software? Guess, if you must. Reasoned deduction preferred.

Thank-you again.

Falk

Now you have confirmed what many of us knew to be true, do you have any idea what is causing the issue, and are you confident on a firmware fix. Or is it hardware related

I said something about it in the paper itself.

What is appearant from the results is that we have two clusters of data which are clearly separate from one another.

Let's call them the good and bad cluster.

The scatter of data within each cluster may be tolerable.

This is what I meant by "the cause must be non-linear".

Even if it is hardware, then a proper detection of the fault situation could be used to improve things in software.

Non-linear hardware faults are frequent. But in many cases irreversible. Think of a broken leg

So, the steep jump from one cluster into the other (focus jump) makes a software issue more likely. Of course, it could be a digital chip which can't be updated. But even then, the firmware could know the condition when the digital chip fails and correct the input it receives from it.

I see an 80% chance that a firmware fix can be made. But only a 60% chance that Pentax takes it serious enough to allocate the good people to the issue. Which makes a fifty/fifty chance we'll ever see a fix. Which is inacceptable, IMHO.

[jpzk]

I said something about it in the paper itself.

What is appearant from the results is that we have two clusters of data which are clearly separate from one another.

Let's call them the good and bad cluster.

The scatter of data within each cluster may be tolerable.

This is what I meant by "the cause must be non-linear".

Even if it is hardware, then a proper detection of the fault situation could be used to improve things in software.

Non-linear hardware faults are frequent. But in many cases irreversible. Think of a broken leg

So, the steep jump from one cluster into the other (focus jump) makes a software issue more likely. Of course, it could be a digital chip which can't be updated. But even then, the firmware could know the condition when the digital chip fails and correct the input it receives from it.

I see an 80% chance that a firmware fix can be made. But only a 60% chance that Pentax takes it serious enough to allocate the good people to the issue. Which makes a fifty/fifty chance we'll ever see a fix. Which is inacceptable, IMHO.

Falk:

Thanks a bunch for your efforts and the complicated paper (I have problems understanding it all ) provided in your blog.
I've downloaded the PDF file on my desktop for further reading ... that will take me days!

So the issue is real and there is a chance (80%) that a firmware fix could be made: are you implying that there is an 80% chance of a firmware update would fix the problem?
I sure would be happy with that.

I'd like to think that since focus jump seems to be the problem, that a firmware will be soon available even with the possibility that hardware (digital chip) is at fault.

Pentax will likely take a serious look at your paper but, as you said, whether they will provide a solution remains to be seen.

I am optimistic though and I believe that they will come up with some way of adressing the issue.

After all, it is all about Pentax not "losing face", isn't it?

JP

[Ray Pulley]

Thanks Falk.

This pretty much sums up what i saw with 3 bodies in my testing although your tests and presentation are much more concise and have more hard data.

Here's hoping that this gets sorted by Pentax soon (along with the similar issue for the Kr) before many of us give up and jump ship.

I REALLY want the K5 to work as it is all the camera I would need for the foreseeable future.

Ray

[JHD]

I see an 80% chance that a firmware fix can be made. But only a 60% chance that Pentax takes it serious enough to allocate the good people to the issue. Which makes a fifty/fifty chance we'll ever see a fix. Which is inacceptable, IMHO.

That's not good, but it is one of those things we just have to accept about this brand. K5 upgrade indefinitely cancelled.

[Class A]

I see an 80% chance that a firmware fix can be made. But only a 60% chance that Pentax takes it serious enough to allocate the good people to the issue. Which makes a fifty/fifty chance we'll ever see a fix. Which is inacceptable, IMHO.

Well, it would be unacceptable, if it were true. We don't know that.

I agree that there is a high chance that a firmware fix can at least significantly improve the situation. If there is a hardware limitation involved, the fix might lead to an approximation of the ideal solution only.

I definitely hope that the chance of Pentax dedicating their best staff to this is higher than 60%. There is less than 5% chance of myself buying the K-5 new unless the issue is addressed. Pentax may need not worry though because most people act differently compared to me.

---

Last edited by Class A; 02-22-2011 at 09:20 PM.

[kales]

thank you very much, falk, this is really a wonderful job, along with the shutter blurr reports.

[Christine Tham]

Hi Falk

Good reading, and I agree with most of the findings in your paper.

However, based on my limited testing, my K10D and K-5 appear to have very similar issues at similar light levels, so I don't think this issue is unique to the K-5, so perhaps we can rule out specific changes in the K-5 as the true cause of this problem?

Also, it appears that different K-5s suffer from the problem to a different extent (with some perhaps not suffering from any significant manifestation of the problem). It's a pity you are not able to test multiple units - I would caution on relying on anecdotal evidence such as "A German dealer said ..."

I think we can agree for example my K-5 and your K-5 would appear to behave differently with regards to AF assist light behaviour even though we were not able to identify build or firmware differences between the two cameras (they do have different build dates).

Lastly, I have not done an exhaustive test of my lenses, but I do believe the problem varies depending on lens, but I don't necessarily believe it varies according to your hypothesis.

Thanks for forwarding the paper to Pentax - I think it will be useful to help them determine what the issue is and whether it is possible to fix (in software or hardware).

I'm starting to think it's more likely a hardware rather than software issue, simply of the variance in reported behaviour across units. But I wouldn't want to second guess what the actual problem is so I'll await the reply from Pentax (if there is any).

[dankoBanana]

There is less than 5% chance of myself buying the K-5 new unless the issue is addressed.

agreed

[the swede]

Hope Pentax talks to us soon! I use my camera ALOT in low light indoor enviroment, and i cant simply use it this way.

[Kenn100D]

@Falk

Does the test come in at the latest firmware version? The old one i think is much better 1.01.

[Gimbal]

Well I'm not convinced, it's an advanced camera and many doesn't understand how the AF work....

[Tommot1965]

I said something about it in the paper itself.

What is appearant from the results is that we have two clusters of data which are clearly separate from one another.

Let's call them the good and bad cluster.

The scatter of data within each cluster may be tolerable.

This is what I meant by "the cause must be non-linear".

Even if it is hardware, then a proper detection of the fault situation could be used to improve things in software.

Non-linear hardware faults are frequent. But in many cases irreversible. Think of a broken leg

So, the steep jump from one cluster into the other (focus jump) makes a software issue more likely. Of course, it could be a digital chip which can't be updated. But even then, the firmware could know the condition when the digital chip fails and correct the input it receives from it.

I see an 80% chance that a firmware fix can be made. But only a 60% chance that Pentax takes it serious enough to allocate the good people to the issue. Which makes a fifty/fifty chance we'll ever see a fix. Which is inacceptable, IMHO.

yes id agree with that statement...

80% chance is good odds, and id take that to the race track any-day..

50-50 for pentax to fix it ....Hmmm, perhaps we really do need a petition from this forum from all of us to ask Pentax to address this issue for us..or we all return our K5's on mass ?

[ursamajor]

Wow! Amazing work! Keep testing! Maybe the next analysed will be the Kr?