Sunday, March 16, 2008

Improved Continuous AF Tracking of the K20D

A new K20D user has made an interesting and systematic comparison test in comparing the AF-C hit rate and performance with his new K20D against his old K10D :-

(Text in Traditional Chinese, use Babelfish if required)

What's the most interest test he did is the "Goldfish Test" (to be exact, the goldfish is called "Peacock"), he used a stop watch/time counter, set it to five minutes and then put his K10D or K20D in AF-C mode. There are 11 numbers of the goldfish in the tank. Each time he focused on one fish. After the focus confirmation is achieved, he focused on another nearest fish and repeated the test for five minutes. The number of taken shots was then counted and then the shots were then classified into "Good", "Fair" and "Bad" according to the sharpness in focusing or the amount of out-focusing. For example, "Bad" shots were those totally out-of-focus ones which are not usable and should go to the trash bin anyway, "Fair" means that there were slight out-focusing in the shots, but the amounts of errors were yet acceptable.

So, here are the raw results of his:-

K10D: 64 shots taken: 14 "Good", 23 "Fair", 27 "Bad";

K20D: 90 shots taken: 30 "Good", 44 "Fair", 16 "Bad".

Thus, it's obvious that the K20D has smarter continuous AF tracking ability than the K10D. The K20D had taken 40% more photos within the time limit! .. which simply means more photo opportunities which I have been emphasizing on from time to time.

I tabular the following on the successful, failure and in-between rates by simple calculations for our more easy reading:-


So, for ideal "Good" shots, K20D has a 52.47% improvement (by comparing the numbers of "Good" shots) over the K10D although the hit rate of 33.33% is still on the low side IMHO and there is pretty much room for improvement. As for the "Bad" share of shots of total failure, the difference between the two cameras are huge. That means that the "usable" rates for acceptable pictures, in counting both "Good" and "Fair" are 57.8% versus 82.22% for the K10D and K20D respectively. Oh well, for the K10D case, I can't think out a better quick phrase to describe that as what "hit and miss" could tell for a such a low hit rate on even "Acceptable" shots - nearly one picture out of two is trashable for the K10D!(?)

Do note also that the user had also tested the AF-C for *static* objects, which he found NO difference between the K10D and the K20D. But, who would use the AF-C mode for shooting *static* objects then?

Besides, the user has also done some quick checking on the image quality differences between two cameras. We can note the purple fringing difference and the noise level difference in the comparison shot he made and posted along the above same thread which shows that the K20D is just better.

So, now back to the AF issues, as stated in the user's above opening post. In fact, he has been much dissatisfied with the AF performance of his old K10D owing to the poor AF-S accuracy that decreased the resolution (and thus decreased the image quality of his good lenses) and the inferior AF-C performance as well. As such, he just took a gamble to go ahead to acquire the K20D without second thought as he just won't want to switch, he told. Well, just as one responded poster said, he was considered to be "very brave", as he had already found that the K10D to be performing so poorly in the AF department. I do second this viewpoint totally - just imagine IF the K20D's AF is actually identical to the K10D's one, then!?

In fact, I found that Pentax has once again to be 200% unwise (or simply *stupid*, I would say) to tell the world that the K20D uses the SAFOX VIII AF system, which is actually infamous for poor AF speed, low light performance and accuracy. In fact, since it has been verified that various new K20D users have found *improvements* of the AF for the K20D. WHY don't Pentax just tell us it is of an improved version, say, it can be named as SAFOX VIII E (Enhanced) or something, just to tell the difference.

Indeed, to explain such stupidity, it is just the typical and classic Pentax behaviour - they never admitted problems and tried to hide them as far as possible! BUT why not let the old bugs in the past over and be *positive* in promoting their new products? What wrong to tell the world that they have debugged (or maybe just using the word "improved" marketing-wise) the old system (although after so many years) and now they have been able to bring out a better version of the SAFOX VIII. I think it should still be a very positive thing afterall!

Read also: Improved AF accuracy of the K20D (Under White Lights Only)


Anonymous said...

I highly doubt if anything has been changed with the Safox VIII. As you probably know much better than I do, there are many factors in the AF speed, more than just the AF module. The *Ist, the K100D, the K10D all used the Safox VIII but their AF speeds varied. Also I believe the Nikon D3 and the D300 use the same AF module yet their AF speeds, especially the initial AF lock, are reported to be somewhat different.

The bottom line: damn it Pentax, do something to change the fundamentals when you charge more than 40% over the old model.

Julian said...

Wow, an (almost) totally positive article on Pentax. What is going on? ;)

I also agree that Pentax marketing should have insisted they rename the AF system SAFOX IX. Olympus do this kind of thing all the time. Like claiming the E-3 has the fastest AF of any SLR (ever!). With a footnote that says "only with 12-60mm lens" and the reviews actually test it and find it isn't even much different from anyone else's AF.

I guess it is just not the Pentax style to make over the top claims. They prefer to under-promise and over-deliver. More of a Japanese style of thinking and acting, I think.

lol said...

I too am amazed that Pentax didn't make at least a small fuss about the AF improvements on the K20... excess of modesty? ;-)

But still, it's here: AF-C works much better and AF accuracy is excellent (under any kind of light).

Matros said...

"But, who would use the AF-C mode for shooting *static* objects then?"

I am. Actually, I was strongly discouraged by the fact that my 16-50/2.8 can not achieve focus on my new K20D in low-light or just on more or less smooth\regular textured subjects when AF-S selected. It was never happen on my previous K10D. K20D often just do lo-o-ong cycle of AF hunting and finally give up with no focus confirmation. Situation became much better when used AF-C. I don't know if this is a problem of particular lens\camera combination, it will be subject of my further investigation, but for now I have what I have. Seems that my older non-SDM lenes focus much better on K20D than my 16-50. On K10D I haven't such problems at all though.
Totally frustrated with my new purchase... I mean K20D.

Post a Comment

Related Posts

Creative Commons License
RiceHigh's Pentax Blog by RiceHigh is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License.