MUST READ! G500 or KD-510z Owners

[JimCockfield]

For those experiencing indoor focus lock problems, there is a fix.

Set your EV Compensation to -1.5 (maximum negative EV Compensation) indoors with flash. This will NOT cause underexposed photos (the flash will automatically compensate, giving correct exposure)!

It WILL allow your camera to focus in areas it could not before, regardless of the amount of zoom you use!

Don't believe me? Pick a subject that you are able to lock focus on at wide angle, but not at full zoom. Then change your EV compensation to negative 1.5 and repeat the tests.

You will be able to get focus lock using more and more zoom, as more negative EV compensation is applied.

As negative EV Compensation is applied, it appears to attenuate the signal from the CCD (or supporting chip set), prior to the algorithms controlling flash strength (so the image still looks exposed the same with flash).

I now believe that this is the ENTIRE PROBLEM with the inconsistencies we are seeing between G500 users (some users reporting that their cameras are working fine, and others reporting that their cameras are not able to focus properly in low light, with higher noise, etc.).

What I think is happening:

When the signal is too strong from the CCD, the autofocus algorithms are unable to distinquish the contrast between pixels. This is why some cameras appear to be defective. I had EXACTLY the same problem with a Sony DSC-P10. It would not focus or expose photos accurately indoors, and also had high noise and image artifacts. It also overexposed photos slightly outdoors. I finally gave up on it, and returned it.

So, I think the entire problem with the G500 (and probably other cameras using this Sony 1/1.8" CCD) is a variation in the signal strength from it (or a flaw in the supporting chip set).

This is why some users report exposure problems, artifacts, higher noise, focus problems, etc., and other user's cameras work fine.

I also believe this is why there is a variation in reviews. For example: Steves-Digicams.com found that the G500 focused surprisingly well in low light, yet dcresource.com had the opposite results.

I also believe that this is why we are seeing a huge difference in reviews in other cameras using this Sony 1/1.8" CCD. For example: some reviewers seeing high noise from the Oly C-5050z, and other reviewers indicating low noise.

It's a variation in the signal strength between CCD's, or some supporting component. Period!

I just started a new album on the problem (and solution) here. Note that there will NOT be a problem with underexposed images using this technique, based on my quick tests so far (because the flash compensation algorithms appear to occur after the CCD signal is attenuated).

http://www.pbase.com/jcockfield/konicaev

After verifying my findings, please report this thread to Konica-Minolta, by any means possible (phone calls, escalation, etc.).

Since I am using a Japanese Konica KD-510z (without a U.S. warranty), my ability to report/influence will be limited (Sorry, but I don't speak Japanese).

They should be able to incorporate this into a firmware upgrade, so that the CCD signal is automatically analyzed and attenuated in cameras without user intervention when using flash, dramatically improving autofocus and exposure.

This will save them lots of money (fewer returns, better customer satisfaction, etc.), so it's in their best interest.

Thanks!

[JimCockfield]

It appears that the camera may limit the maximum focus distance when using flash.

In other words, even though you may be able to get a focus lock in low light on subjects 17 feet away at maximum zoom indoors with flash, the camera may be keeping the maximum focus point within the specified flash range.

It was probably designed this way. I'll do more testing and report my findings.

[JimCockfield]

I have been taking a LOT of photos since my initial post (filling up my memory cards a couple of times, recharging batteries, etc.).

Initial Obervations:

* Focus Lock is NOT Focus Lock (as I've suspected all along), and my photos are in focus without the lock, using the default (no EV compensation) settings, with rare exceptions (for example: Christmas Tree lights, or shiny surfaces with reflections can fool the focus system).

With my camera, even when using full zoom, with no focus lock indicated, the subjects are in focus indoors, with virtually any light, on subjects with virtually any contrast available. It works absolutely fanatasic (better than other Digital camera that I've owned in the past).

For those that have a camera that appears to be working great (regardless of lock), test this yourself. With virtually any light present, the focus is accurate, regardless of Focus Lock.

You can also easily verify that the camera is NOT defaulting to a preset (1m, 2m, 4m) distance by taking photos at both close and far ranges, and comparing them to the results you are getting with autofocus without a lock indication.

The autofocus photos without a lock indication will be in focus, when fixed focus distances will not (for example: a darker subject 2 feet away, comparing autofocus with 1m, 2m, 4m).

Only when in virtual dark conditions (or a subject with no contrast), does the camera seem to default to a focus distance that is at a relatively close range to the camera (probably about mid way through the flash range, which allows a higher percentage of in focus photos, even in total darkness). I think this is a great design decision on the part of Konica.

* EV Compensation DOES Impact Focus, so I still believe that those with "bad" or "out of tolerance" cameras may have to "tune" EV indoors with Flash for accurate focus, to match what appears to be a QC problem with the Sony CCD or supporting chip set (with different signal strengths from the CCD impacting the camera's ability to distinquish contrast between pixels).

Of course, the best solution would probably be to return the camera to the vendor, if your camera is not focusing accurately (but do NOT use the Focus Lock indicator as a guide). It is simply is not representative of the camera's ability to focus accurately.

* What is happening:

When you point your camera at a subject, and press the shutter button, the camera "gains up" the image from the CCD momentarily to focus (you can observe the difference by viewing the LCD display while focusing, watching the image brighten momentarily, then darken back to normal).

As more negative EV is used, you will see more "gaining up" of the signal (the opposite of what you would expect using negative EV). However, too much "gaining up" results in out of focus images with my camera. Mine appears to be properly "tuned" from the factory, for the CCD installed in it.

The camera is probably designed to do this (gain up momentarily) in low light conditions, so that it can "see" the image from the CCD better in order to focus. EV Settings DO impact focus accuracy, but mine appears to be tuned correctly already.

With my camera, using zoom at further distances (i.e., 17 feet away indoors with flash), with negative EV set to max, photos will be out of focus (even though the camera "says" focus is locked). Yet, with no EV Compensation, these photos are IN FOCUS with my camera. I had initially thought that the problem may be that the camera was limiting max focus distance with flash, but this is NOT the case. I was incorrectly assuming that the Focus Lock indication was real (it's not).

Again, Focus Lock is NOT focus lock (but I have NO IDEA what the indicator is telling us). With EV set to normal (in my camera), Focus is reasonably accurate, even at full zoom, with no lock indication (it's not a true focus lock indicator).

It's probably some combination of internal parameters, and appears to be a firmware bug (perhaps some programmer even had some code in the firmware that they forgot to take out, using the indicator for some other purpose, measuring a combination of some values). This is speculation on my part. BTW, this indicator works differently using flash, compared to flash forced off.

* Suggestions:

Test this yourself. For those that appear to have cameras that do not focus accurately, try "tuning" the camera using EV Compensation indoors with flash, so that the best signal strength for accuracy from the CCD is used, as the CCD "gains up" while focusing. Again, DO NOT use the focus lock as a guide, as it does not appear to work as a true focus lock indication.

Use a tripod, focus on an object with a reasonable amount of contrast, then "tune" the EV (plus or minus), until focus is accurate, without relying on the focus lock indication. Look at the photos as your guide to accurate focus while tuning the camera.

* Initial Conclusion:

Based on my initial tests (and the descriptions of problems being encountered by some users), I still believe the problem is related to the CCD or supporting chip set, where there is a variation in the signal strength being used by the camera for contrast detect focus.

IMO, this should be addressed in a firmware upgrade to the camera (to automatically adjust the CCD signal, for the most accurate focus), along with correction of other problems found, for example:

* Correct the Focus Lock indication so that it can be reliably used as a guide (although my cameras works so well, that I wouldn't need it anyway). Again, my camera focuses SUPERBLY!

* Modify the camera's use of Auto ISO, so that sensitivity is being varied between ISO 50 and 200 indoors (400 would be "pushing it" for this CCD), with finer steps in between - not just 50, 100, 200), depending on lighting conditions, while correctly updating the EXIF to reflect the actual ISO speed chosen.

We are already aware of a bug, where the camera is actually using ISO 50 for photos without flash, regardless of lighting conditions. Yet, the EXIF data shows ISO 100.

This can be easily verified in controlled conditions, since exposure (aperture and shutter speed) without flash, matches up identically to what the camera uses when set manually to ISO 50 in a user profile. The exposure does NOT match up to ISO 100 Shutter/Aperture, as reflected in the EXIF.

Jim Dawson found this problem first, and it's since been verified by others in controlled conditions.