Web Analytics RiceHigh's Pentax Blog: September 2007

Sunday, September 23, 2007

DA 21 Mis-Focusing - A Design Problem?

Quite some time ago, I wrote here for the DA 21 Limited Front Focusing (FF) and Back Focusing (BF) problems as reported by some users, in my fourth article in my this blog (and actually the first time for me to talk about the issue on the net - the contents of the first three articles were actually talked about and discussed previously in three different posts at the Steve's Pentax forum which got me banned shortly afterwards overthere).

Later on, the famous professional fashion photographer, Mr. Benjamin Kanarek, who was a new Pentax user (as you know, he switched from Canon), found that his DA 21 Limited lenses as acquired from Pentax France/Germany were having back focusing and also front focusing problems and he finally got four copies of the DA 21 after that I have not seen that he reports again about mis-focusing of his DA 21 on his K10D anymore. Nonetheless, I'm not sure if his this copy really works for him now or he has just aborted to try as the problem might still persist.

So, reports of those FF and BF problems of the K10D, in particular for those associated with the DA 21 Limited which are even more frequently reported, either with the K10D or the K100D or both, just have been continuing to appear up till now (yet an recent example here, with comparison photos posted). Recently, particular German users have found that the mis-focusing actually associated and being affected by the aperture f-number as set by the user or as selected by the camera, which a German user summarises the case clearly and precisely in the following DPR forum thread:-


So, the reported strange phenomenon is that if the aperture selected is smaller than f/5.6 (i.e., f/6.3 and onwards), the AF accuracy will be better, even if the user set a f/8 value first, say, then performs the AF, and then change the f-number back to f/3.2 wide opened, say, and take the shot - the focus achieved is usually better than when the user uses the camera normally, e.g., f/3.2 is selected when doing the AF and then take the shot at f/3.2.

Indeed, the above finding really sounds very strange and odd and I could hardly believe that for the first time I heard about it. However, by following down the above DPR thread, one will not be difficult to see quite some different DA 21 users have confirmed the same phenomenon after they tried and tested their gear, no matter briefly or extensively with care. So, the strange problem, along with the yet clumsy and odd "workaround", have been confirmed.

Actually, I saw before from an issue (#116) of a local magazine that the resolution of the DA 21 lens on a K10D as tested and measured using the Imatest is rather low, with results not more than 1,600 LW/PH (Line Widths per Picture Height) at image centre and around 1,400 only at image corners, at best. To compare the figures, I checked the Issue 115 of the same magazine for the test results of the (just 6MP) Nikon D40 with (just) the Nikkor kit lens 18-55 II AF-S and was rather surprised to learn that at 18mm, it yields results at about 2,100 LW/PH at image centre and about 1,500 to 1,600 LW/PH at the image corners, depending on the f-stop used, which simply means that the resolution of the kit lens on the D40 just blows away the DA 21 Limited prime on the K10D fully and also throughout the whole picture frame as well !(?)

Well, the possible reasons for the low resolution as verified above can be two: 1. The inaccurate AF system of the K10D which gives mis-focusing which in turn decreases the effective resolution significantly; 2. Blurry in-camera jpegs of the K10D (first discovered and reported by Phil Askey) which are primitively having lower resolution than other DSLRs of the competitors or even the K100D (now it is worse than the D40, as verified in the mentioned tests above, in just two consecutive issues of the same bi-weekly magazine).

As a side issue, a poster who responded to the above DPR thread also pointed out that the final focus point achieved by the AF system of his DA 21 varied depending on whether the lens was turned from infinity or from the closest position, which this yet strange but surely undesirable phenomenon was first reported by me nearly two years ago in my this post in 2005. It was really rather disappointed that now the DA 21 on the latest Pentax DSLRs is still the same, with the same old problem, even after two years I had reported the issue.

Do note that in my above old report I had indeed fully tested my $180 MZ-30 (and my more expensive MZ-S too) that my FA 43 Limited did have NO such problem exhibited on these two film bodies whereas my *ist DS and D are having the mentioned problems obviously and do note also that I tested and compared the MZes and the *ist Ds at the same environment, with the same FA 43 Limited lens, for the same target(s) and at the same time. So, this case is really rather sad and absurd.

All in all, I really don't understand why the problem is still here and it seems that Pentax have not listened and still don't listen and the bug is still here! (but Why?) The Bottom line is: Don't forget that the MZ cameras do NOT have similar problems! (including the rather cheap $180 "bottom-of-the-line" MZ-30). So the key point still is: IF Pentax listens.. (and also if They didn't produce inferiorly performing systems in their DSLRs, namely, the AF and AE systems used, which they did do it right and were capable to make better stuff before, in their MZ line of film SLRs..)

See Also:-

AF's Dependency on Lenses and Yellow Light

Focus Calibrations of Pentax DSLRs and Lenses

Saturday, September 15, 2007

First Formal Test of the DA* 16-50/2.8

A Polish gear testing site has carried out the first DA* lens formal test and the test report is published as follows:-


(A free online Polish to English translator can be found here)

Yet, below is a better human-translated summary on the test:-


Well, let's go on to see about the key findings of the test:-

1. Resolution:

See the following report page for the 50% MTF test results, measured in the lp/mm:-


As a short note for what the test is measuring and what information the results can tell, it is to measure the maximum resolvable line-pairs per millimetre (lp/mm) for the black and white lines in the resolution chart for the difference between the black and white levels not less than 50% (ideally, black is at 0% and white is at 100%, so the maximum difference can be 100%).

From the first chart, we can see that MTF / resolution figures are good at the image centre. It can be seen that the optimal focal length is at 30mm, which is the middle of the zoom range. The maximum obtainable resolution figure is about 44 lp/mm. At both extreme ends of the zoom range, i.e., at 16mm and 50mm, resolution drops at about the same amount. As for the optimal f-stop, it is at f/5.6 where the best resolution can be obtained for all focal lengths. At f/2.8 wide opened, the resolution deteriorates and can be as low as at about 27 lp/mm at 50mm.

So, one would ask me how to judge those figures. Is 44 lp/mm or 27 lp/mm good or bad? First, you can compare Photodo charts and figures to have an idea. Do note that Photodo tested the Pentax lenses in a different way for the MTF. They used fixed test target with patterns of different lp/mm and then measure the MTF percentage. So, the higher the percentage, the better is the result. On the other hand, remember the latest popular measurement unit of LW/PH (Line Widths per Picture Height). Normally, a 2,000+ LW/PH figure is considered as high resolution for a 10MP DSLR, e.g. for those Imatest figures or those eye judged resolution charts by DPR. To convert the unit, 44 lp/mm simply means 1,408 LW/PH for an APS-C sensor at about 16mm height. 27 lp/mm is equal to 864 LW/PH. Although those figures are not directly comparable. I would say the highest obtainable resolution of 44 lp/mm is fair to good enough but the 27 lp/mm should be bad.

Well, now that we go to the second chart, which summarises the results for the edge performance. As you can see, the performance drops further. The poorest performance occurs at 16mm and followed by 50mm and then 30mm is still the optimal focal. But now that the highest lp/mm figure at edge is just 39 lp/mm (at 30mm at f/5.6) and the worst now drops to 16 lp/mm (!?), at 16mm at f/2.8, which undoubtedly is a very poor result.

Finally, it can be noted from the captures of the resolution charts for the existence of colour moires of which this phenomenon was also found by other reviewers before, for the K10D.

2. Chromatic Aberration:


They conclude that the CA is terrible, especially at the widest focal and/or wider opened, it performs worse. You can also see the obvious CAs from the sample pictures posted, in addition to the CA chart.

3. Distortion:


The distortion figures obtained are indeed rather terrible at 16mm, at an unbelievable huge amount of -4.1% (barrel) whereas at 30mm there is virtually no distortion (0.1%, pincushion) and at 50mm it is good (0.71%, pincushion). But then at the wide(st) angle, such a huge amount of distortion is ridiculous and IMHO practically it is just unusable! The Pentax DA 16-45 has a distortion of -2.52% at 16mm but yet I already notice its distortion when shooting landscape, I really cannot imagine what will happen for a -4.1% barrel distortion - which is worst than most lenses, including most cheapo kit lenses at 18mm.

4. Blur pattern:


Interestingly, they measured how the image blur at this part of their test. As you can see from the red beam point. At centre (pictures at the left column), the red point reproduction is near perfect and at edges (pictures at the right column), the image formation is sketched in a certain directional way and thus causes blur.

5. Vignetting:


Obvious vignettings are found at 16mm as tested, even at one-stop stepped down at f/4. At f/2.8 and 16mm, an average mean of -1.8EV light loss is recorded at the image corners, which is considered to be significant for an APS-C DSLR. Well, I can see slight de-centering defect, too. The de-centering can be a camera body body for a not perfectly centred CCD sensor, though.

6. Flare Control:


As we all can see from the sample photos, the flare control is poor, at least the testers were disappointed with the results, especially when compared with the other Pentax lenses they have tested before.

7. Autofocus speed and accuracy:


They found that the speed and accuracy are both good enough. This differs from a few of previous less formal tests I have quoted in my blog before.

So, the conclusion? I think I need not to go further to quote their conclusion. I think you have made up your own conclusion already at this point. This DA* lens simply doesn't perform, with worst performance at wide angle and wider opened. So, what's the point of getting this lens? Aren't two of the major selling points of this lens is that it is a fixed f/2.8 faster zoom with a 16mm wider angle? Even for the flare control, where Pentax usually excels, the lens yet doesn't perform here. A real disappointment here and I feel really very sad again here. Actually, this is rather sad for Pentax to produce a "Pentax" lens like this, not even to mention again it is a "Star" lens as labelled. It seems that Pentax has just put her long established Star reputation at risk, which is indeed the most disappointing thing ever seen in Pentax's history.

See Also:-

Shootout Again! DA* 16-50 Vs SIGMA 18-50/2.8

SDM = More AF Errors?

DA* 16-50 Vs DA 16-45, A Shootout

Friday, September 07, 2007

Shootout Again! DA* 16-50 Vs SIGMA 18-50/2.8

Here is a very recently published shootout test for the above two lenses:-


(The report in Traditional Chinese, you can use Babelfish to translate)

Nonetheless, since Babelfish is not something one can fully understand :-D, I summarise the author's findings as follows (you may wish to read in conjunction with the full report with all those testing sample pictures posted):-

1. Purple fringing of the DA* is more obvious than the Sigma at the widest focal settings, for apertures from f/2.8 to f/8, i.e., at 16mm Vs 18mm respectively. The author compares the image corners.. ;

2. Exposures (i.e., for the image brightness) and colour/tone responses are more or less the same as tested at 50mm;

3. Resolution of the Sigma throughout the same aperture range (f/2.8 to f/8) wins the Pentax clearly, at the widest focals as compared. My comment is that this comparison may not be totally fair as the magnification of the Sigma at 18mm is slightly larger. As such if the same crops of the same image scene are compared, the larger magnification one will have the slight advantage. However, the difference seems to be so obvious that the Pentax should really have (much) less resolution. Also, the tester also tested at 34mm/35mm to verify his results on another day and yet again he found that the Sigma is still having better resolution;

4. Focusing accuracy test: The Pentax DA* back focused whilst the Sigma focused flawlessly on the same K10D!(?) The author remarks that this confirms the reported (back) focusing problem of the DA* lens, when SDM is used. He further writes that the focusing error could be the true culprit for the decreased and lower resolution as found. I think he is right for that suggested point as focusing accuracy is very important to keep the (high) effective resolution;

5. Distortion levels of both lenses are more or less the same and both are quite obvious. My comments are: I can see from the test images that the test chart/target are not shot "perfectly" on an optical bench for perfect alignment nor the target is perfectly flat. However, this is understoodable for what a non-professional measurebator could do with the limited resources he has. Still, the test shots can be treated as some kind of basic reference, IMHO;

6. The author is astonished at the high performance/cost to price ratio of the Sigma but however rather disappointed with the Pentax DA Star, which is expensive but yet found to be inferior and lose the shootout for almost all aspects under test;

Well, after reading this test and all the results, I feel rather disappointed (again). The DA* Star is again something let Pentaxians very down in general. It has focusing issue as verified, which has been reported over and over again since launched. The true cause can be either a design flaw, no matter hardware and/or software wise or a quality control issue. The DA* optically does not live up to the standard as supposed and expected. Actually, it has more obvious purple fringing, lower resolution, obvious distortion and so on - and the worse is that it cannot even win but just lose to the Sigma which is being sold at a price less than half of the DA Star.

My impression is that the DA Star is just something Pentax created to hurt seriously their long established reputation of their only true strong point, i.e., excellent optics, especially for the Star series, which should be their top-of-the-line, most luxury glass with the best possible optical quality and performance on the Earth, but still being sold at attractive and reasonable price levels - which all these favourable factors are true and valid for the old film FA* and F* or even A* lenses. But now that it looks rather sad and disappointed for those digital DA* "Star" are found to be performing just like some kinds of rebadged Tokinas which are actually being sold much cheaper. Still, the Pentax Digital "Star" are yet marketed with a "Star" price but lack the true quality and real performance which they should have (or at least as expected or supposed to be), as reported by many of the new users for what have been tested and verified repeatedly from many of the recent reports. (Not even to mention the unique optical characters which each of the old *true* Pentax Star lenses had - Sad. Really..)

Afterall, I believe that the consumers are brilliant, overall speaking. Reputations, no matter good or bad, are built over time and nothing can be hidden or cheated. So, if the DA* really don't perform, I'm afraid that Pentax is really endangering their brand name and just spoil their long-established good name of the Star and when the end of the Star legend comes, their old real fans must be upset (mostly).

How to Get Back to an Older Version of the Firmware?

For quite some good reasons, Pentax users, particularly K10D ones, wish to get back to older version of firmware for their DSLRs. A quick search on major Pentax forums will reveal such queries easily.

One of the major "drives" for those users to "downgrade" is that they encountered focusing problem so that they need to get back to the earlier version 1.10, which has the engineering service mode access method disclosed and in that mode there is a parameter that can be input to compensate for the AF error and as such the Front Focusing (FF) or Back Focusing (BF) problems can be cured partially, at least for particular lens and under specific lighting condition, say, tungsten environment.

Now that some people have found that SDM focusing could produce more AF errors, i.e., BF or FF. So, I and some people have thought about why not try an older firmware and see if there is a difference and if the problem can be cured?

Now, the problem is how to downgrade the firmware, which is not allowed by design. As when the user tries to re-program a "new" copy of firmware to the camera, there must be a version checking process which will only allow upgrade, or at least re-programming of the same version, but not an downgrade. Pentax have assumed that users should and will never look back. But of course, this is not the case in real life.

So, here comes a method for downgrading from version 1.11 to 1.10. In fact, the working principle is simply to cheat the existing firmware for the "new" firmware version. Just change the old firmware file's header, using any Hex Editor (click here for a list of free Hex Editors), to replace the string(s) of actual firmware version to the latest version number, say, it can be even 1.30. It is said that one can then program/write over the existing firmware. And then, if one still want to have complete original old firmware to be installed in his camera, the author said that now the original old firmware, without any modification, can now be then downloaded back to the camera by just using the usual procedures for firmware "upgrade". Well, a copy of the original untouched K10D firmware version 1.10 can be found here.

Good luck and happy downgrading (and actually hacking)! ;-) But do remember try everything at your own risk and you should be the one who knows what you are doing, and why you need to do so, the risk involved and the possible consequence(s) thereafter!

Update (Nov. 25): All the K10D firmware versions can be found here, officially:-


How long Pentax Canada will keep the older versions is yet a question, though, as it seems that it's Pentax's central policy to remove older version of the firmware for downloading once a newer one is available.

Update (Dec. 16): For those who do not want to do the Hex editing, an already modified copy of the firmware version 1.10, which can be used to write over the latest firmware version 1.30, can be downloaded at the page below:-


If you still feel unsafe or uncomfortable or whatsoever with this "hacked" version, just write over it with the original 1.10 that can be obtained in the Pentax Canada link mentioned above and then the downgrade will be perfectly completed.

Wednesday, September 05, 2007

SDM = More AF Errors?

Recently, some different Pentax users, who had got the new DA Star lenses for their K10D, have reported that they found that more AF errors were resulted when the SDM lenses were used on their K10D, i.e., by enabling the SDM function to do AF ultrasonically. Just see the following hot thread:-


As we can see, the logic of the thread open person is simple and clear. If his DA Star lens is used on a non-SDM body with traditional body driven AF, the AF accuracy is good enough at f/2.8 wide opened. And then if another older generation non-SDM prime lens is used on his K10D, the AF is still okay for the same aperture and focal as compared. So, out-focusing only occurred when the SDM lens is used on a SDM body and thus he has drawn up the conclusion that the culprit is the new SDM focusing function, which should be in fault or having some bugs (hardware and/or software wise).

Do note that the Pentax Star lens are with hybrid AF facilities so that it can be used on both latest SDM-enabled Pentax DSLR or older body driven Pentax AF bodies, which is really a good thing as full backward compatibility is ensured. However, if the SDM would actually introduce more AF errors, then this will be a big problem and issue. In fact, this issue (if really existent) must be looked into seriously and tackled by Pentax in a very urgent way and manner. In the meantime, a custom function for letting the user to choose between SDM or body AF motor would be a viable interim solution that can be provided by Pentax quickly and easily, as only a new firmware update with minimal software modification is needed. Until some days later an ultimate technical solution arrives, this firmware option could be the true saver for users who have encountered the problem - and for the rest who do not find any problem, they can leave the setting at default and continue to use the SDM.

Anyway, to learn more about the reported cases and to judge if the problem is really existent, I think the lengthy thread cited above already provides quite some useful information for reference with useful sample photos posted as well. Nonetheless, as there are numerous numbers of opposite arguments followed down the thread (as usual and it is quite normal), I think any reader without a stance can think and judge the case in a more objective way, as far as possible. Of course, in the end, everything can still be concluded with the "bad sample" theory that the reporting users are just being the most unlucky ones. Or, again, simply the old but golden defensing rule of "user errors" can be applied. But, if you're about decided to buy the new DA* lens(es), will you take the risk? Will you bet, or not? (. . until some days the case has been made clearer.)

Update (Sept. 7): I have just a thought which've come over my mind. I think instead of waiting Pentax to confirm and hoping for a solution from them, why not be proactive and try to test the gear oneself and even to find out a possible solution that is actually ready available?

In fact, according to the past track record, Pentax have never admitted those more critical reported gear problems openly or they simply just not even responded publicly to those, even if for the problems that had been widely reported and recognised (clearly for their existence). As such, I think my last imagination or suggestion on the "interim solution" maybe just something not very realistic afterall.

So, what's my thought then? Still remember the old K10D firmware versions of Pentax which do not support the SDM? As such, with those older firmwares, it is most likely that the K10D will do AF *traditionally*, i.e., with the body AF motor to drive the Star zooms to do the AF. I haven't tried that myself though but I think it is most *likely* and it should work.

So, how to get back to the older firmware? Just see my next short article on that.

Update (Sept. 9): PentaxForums.com member SloPhoto has confirmed that older K10D firmware version 1.2, i.e., very probably for all firmware versions prior to the version 1.3, "will focus the DA* lenses using a conventional screw drive". So, this would be really a good news for all those users who have found problems associated with the SDM for their new DA* lenses on their K10D. At least now there is a readily available possible solution for them to try.

Sunday, September 02, 2007

De-centering Defects of Pentax Lenses

Klaus Schroiff of the PhotoZone recently has cancelled his tests on both of the two latest DA* zooms, owing to the pronounced "centering defects" found. In fact, he has already found so many and too many similar defects with Pentax lenses during his tests on Pentax lenses carried out up till now. With those centering defects found, Klaus either re-acquired another sample from the PhotoZone community or he simply cancelled the tests, which is indeed a rather disappointing thing, especially when this keeps repeatedly happened, very unfortunately.

A search of the term "centering defect" in the PhotoZone forum datebase using its search function will reveal how many times this problem has been appeared and reported during the course of his testings in recent months. I'm actually not being totally surprised for having alignment issues with Pentax products from my own "unfortunate" experience, especially for the recent Pentax products which were made and are being made outside Japan.

Nevertheless, I am a bit sceptical about if it is really about the poor QC of the Pentax lenses instead of the mis-alignment of Klaus's K10D body instead. Coincidently, I have asked Klaus the similar questions more than twice in different posts in which he reported the problem when he tested different Pentax lenses (which I can see many of the conversations again when I have done the search on the keywords "centering defects". Below were some of his answers to my questions:-


"This is very unlikely. There were no significant centering defects in the Tamron 70-300, Cosina 100 and the Sigma 10-20 tests - these were among the most recent test runs in Pentax land."


"An asymmetrical centering defect isn't all that severe from a testing perspective.
If I reject a lens it is mostly because the _center_ portion shows both sharp (although suboptimal) edges as well as "edge shadows". In the real life this kind of defect shows up as weak contrast but not as asymmetry."

Here is a capture of the *image center* for a chessboard like resolution chart pattern:-


As we can see from the above diagram, asymmetricity or the phenomenon of directional "edge shadows" does really exist. As Klaus has mentioned in so many of his Pentax lens review and posts about his tests that there are centering defects in common (more or less), objective facts tell me that I should admit that there is a problem with even the Pentax optics, despite I have already tried to think objectively about any other possibly cause and asked for times if "it should be the body". (Well, I do not have very good faith or impression on the quality control and perfect alignment of the K10D body but I can't imagine even a bundle of different Pentax lens models could have serious alignment problem - even for the optical center).

Very interestingly, MTF tests like those conducted by Photodo indeed can already show the presence of centering defects, as cleverly pointed out by this PZ poster:-


Now, the DA* 16-50 and DA* 50-135 samples which have been got from the street have very significant and pronounced centering defects. Despite the tests have been cancelled and Klaus will not publish the test results formally anymore, he is yet kind enough to post briefly the unpublished results in his forum:-


Well, as we can see, actually both DA* perform not so well especially for the poor/low resolution at wider apertures and at the image borders as well as the huge amount of CA present. Whilst the centering defect, which is found to be existent in both DA* lenses, could affect image quality and "spoiled the MTF figures a bit", which has been verified, here, possibly the lenses' optical performances are not that good by themselves. As remarked by the tester, the inferior CA performance is just "Tokina alike".

Last but not least, as those lens samples obtained are new products which have passed the factory QC and those are acquired from the streets, we may think it in the following two additional possible ways:

1. The de-centering amount is just within Pentax's factory specs but should not be considered as a defect by Pentax's own standard - but then Klaus compared the Pentax lenses from Canon, Nikon etc. or even to Sigma and Cosina (on the same K10D) but he found that particular Pentax lenses are worse, including even the latest "luxury" DA Star zooms;

2. Those samples put under test are just lemons and they leaked through the Quality Control system/net from the factory - if so, I'm afraid that the QC system of Pentax is really rather ineffective or even useless as there are so many bad samples out there as picked up in the streets and ultimately been borrowed/bought by Klaus and very unfortunately even been put to the lab test.

All in all, my opinion is simple: for lower priced lenses, such defects are more acceptable as the customers should get what they paid (for). However, for "Star" lenses that marked with a new golden "Star" label on them, with a "Star" price, shouldn't we expect a "Star" quality also, in terms of *real* stringent QC tolerance and inside performance as well.. But, so far from what we can see from all those specific adverse user or even test reports of the new DA Star lenses (which are shown together with quite some evidences), I'm really afraid that our above assumption has not been proved valid and maybe even just wishful thinking of we Pentaxians. Really sad.. :-(