Saturday, April 09, 2011

Technical Notes: Testing the Pentax K5: no pix this post

My fascinating two-week trial of a Pentax K5, thanks to the generosity of RW (Pentax had nothing to do with it) has gathered enough data and real pictures to reach a conclusion.
The K5 (I never used a K7) presents substantial improvements over the K10/20 platforms on a whole slew of parameters. The train-wreck shutter sound has become very soft, less obvious than the sharp snapping sound of my GF1, which doesn’t have to contend with a flapping mirror. The AF is better than before. It is much less prone to skip over a subject you expect it to focus on (previous Pentax models, pointed at a bicycle, invariably focused on the wall behind it, while this sort of target had never presented a problem for AF systems I first began to use with Nikon in the mid-80s). To be more specific about it, if I held up one finger of my extended left hand and asked the camera to focus on it, it could not, in the field of view of the 21mm lens. It went for the background. Bunch two or three fingers, and it suddenly could find the target. My Olympus E1, a vintage getting pretty old now, could focus on a length of string at that distance.  So could ancient Nikons. The K5 can focus on one finger, though not a piece of string.

Tonal range, SBR—the brightness range of the subject you point at that can be recorded with full data—is significantly improved.

The camera feels nicer in your hand...well, my long thin hands, it might not be the same for someone who doesn’t need XXL Playtex gloves for film development.

AE and AWB are both better, though if you shoot RAW files and then use ACR (the RAW converter for both Photoshop and Lightroom) the camera’s histograms may throw you off until you adapt to them. They seem to reflect a serious highlight recovery effort on the part of the internal software, and shots that are exposed “to the right” according to the camera’s histogram are apt to come into Adobe RAW conversion with highlight clipping. Sometimes a little, sometimes a lot. The difference between the paradigm the in-camera JPEG/histogram is using and the default in ACR seems highly variable. If you use the histogram features, either in digital preview or after the shot, you’ll need to learn how much headroom to allow. Good luck.

The camera was utterly unable to interface in an effective way with a Pentax flash it is supposed to support. The previous cameras performed great in this interface. This would be a major issue if you need to use intelligent-flash features.

I won’t be acquiring one of these. The improvements over the K20D are substantial, but the astonishing thing is, they don’t come up to the level of the Lumix GF1.

Apples and oranges? Of course that’s a problem. But the improved AF of the K5 is sluggish and erratic compared to the GF1. The improved SBR doesn’t impress me except in comparison with earlier Pentax dslr cameras, not the GF1, despite the sensor size difference. Whatever DxO says, I see an improvement over what earlier Pentax bodies did, but precious little over what I’ve grown used to, shooting with the GF1 for over half a year. It wouldn’t be the first time that I’ve found supposedly rigorous technical test parameters don’t actually relate to using equipment in the real world.

The resolution of the K5 files is, well, nice. Nothing to write home about. The problem is, the GF1 files make better 20” wide prints because the superiority of the 14 and 20mm lenses, handled by reliable AF, coupled to the incorporated-into-ACR corrections, beat the hell out of the Pentax 21mm lens (using the latest ACR-incorporated Profile). The Lumix 14mm also surpasses the Pentax 21, and the Pentax 15mm just looks pathetic in comparison to either of them. The 35mm Macro from Pentax remains spectacular, when the camera focuses it correctly.

When I interpolate similar files from the GF1 and the K5 up to around twenty inches to make a print the GF1 wins every time, despite the greater degree of enlargement it needs. What can I say? It’s the print that counts.


Markus Spring said...

Carl, this is strong conclusion, and astonishing for me, as I was convinced for a long time, that the size advantage of the APS-C format would pay out, especially in higher ISO regions. But your night images of New York speak their own language. From friends I've heard that the K5 really shines there, but they haven't made any side by side comparisons.

Luckily enough there is no upgrade body available for my Sony system - going full frame is not really an option - so I can postpone any sleepless nights to a later time.

Carl said...

Markus, the K5 does beat the GF1 at high ISO, but, I was surprised to find, only by a stop or so. Instead of 800 being routine and 1600 acceptable, those numbers go to 1600 and 3200. As with the SBR, there's an advantage, but not an impressive one. These just aren't things I'd drop $1400 on.

Tyler Monson said...

Carl, as you know from our many e-mail exchanges, my experiences with the Pentax K-5 (and DA lenses) have been about the complete opposite of yours.

The images posted on my primary blog since 22 February have been made with the K-5 and DA21. It is fast, quiet, easy to carry, and has not yet disappointed me. I love it!

At the same time, all the pictures posted on another of my blogs since 14 September were made using digital compact cameras between 2001 and 2006.

Each camera style has its advantages and disadvantages, and I've enjoyed them both.

At the same time, with digital photography, I find the camera and lens matter to be far less pertinent than it was with film. The computer has given us so much more control over the process that I find myself selecting the camera that gives me the maximum amount of data to use in the computer's virtual darkroom.


Carl said...

Tyler, pictures posted here beginning March 30 are all from my extensive testing of the K5. Mostly with the 21mm though some are with the 15 or 40. I got some things I really like using this equipment. There's a backlog of them that will probably fill the blog for another week or longer. But my conclusion was that I prefer working with the GF1.

My primary criterion is what final prints look like, and somewhat surprisingly I don't find prints from the K5 and Pentax lenses to be superior to prints from the GF1 and Lumix short primes—rather the opposite, in fact. That makes me really wonder what I might get from a GH2...

Dennis Allshouse said...

The "good luck.." is really ominous.

What I find weird is that Panasonic seems to have baled on the GF1 approach. To me the G2 is intriguing in so far as the viewfinder is reportedly much better than the clip on one.

Carl said...

Dennis, it's just that AF issues continue to swirl. I found AF on the K5 much improved over the K20, but that's not the same as saying it's state of the art.

I don't know what Panasonic was thinking of with the GF2. I *certainly* won't be buying one of those. My wish list would include a more advanced iteration of the GF1 platform. Or more simply, how about a super upgraded EVF--it's a replaceable accessory after all--with the quality of the G-series EVF.

Michael said...

"The camera was utterly unable to interface in an effective way with a Pentax flash it is supposed to support."

Who needs flash when you have Photoshop skills?

Just curious... M

Carl said...

"Who needs flash when you have Photoshop skills?"

In a word, backlight. Digital sensors still have limited dynamic range compared to negative films. Plenty of situations exceed that range, and Photoshop skills don't invent data that isn't there. Well, they can, but then you're making a digital painting, not a photograph. I don't use fill flash very often, but there are times when it's essential.

Anonymous said...

"what ever DxO says"

well DxO says the high-iso performance of the k-5 is 3.5 stops better than the GF1. But that is all just scientific and objective mumbo-jumbo. Its the seat-of pants vibe based tinkering that matters after all, isn't that right Carl?

Carl said...

"But that is all just scientific and objective mumbo-jumbo."

A straightforward objective test—how many stops above and below center can detail be retained in a textured subject—showed no such advantage. Neither do real-world pictures. Getting anywhere near 14 stops of range requires amounts of highlight and shadow recovery that don't result in convincing tones in a print. If I can't print it with beautiful tonality, then it's just meaningless data.

Anonymous said...

3,5 stops? I get (measured ISO) between 1-2 stops depending on where at the ISO scale you look.


Carl said...

Dan, I'm not sure if I understand your note correctly, but if you are saying you see a 1 to 2 stop improvement in SBR for the K5 over earlier Pentaxes or the GF1, I agree that's a lot more realistic than 3.5 stops. I think it's closer to the lower figure, but I'm judging on the basis of wanting full, rich printing information. That puts a strict limit on how much highlight recovery and shadow lifting I'll accept as being truly usable.

I.B. said...

Interesting post. Personally I upgraded from The gf1 to the gf2 and found it worthwhile. The video performance is much improved, the size is dramatically decreased and thouch screen is actually convinient.

I guess the pro viewfinder modell will come soon

Carl said...

"I guess the pro viewfinder model will come soon "

My "wish list" would be for a greatly improved EVF and a GF3 that returns to the GF1 form factor and similar operational system (the GF2 is way too menu-based for me) with the sensor improvements of the GH2. Of course it might be simpler to bite the bullet and get a GH2--if they ever become available again.

nico-foto said...

Hi Carl, i came across this post from someone who published it on dpreview forums. I'm no Pentax user, but i do have a Canon 7D, which is, to me at least, an incredible shooting machine. But! Everytime I check my GF1 pics on the screen, they just pop in a way I have not seen in my Canon, the images from the 20mm 1.7 are exquisitely sharp, and the bokeh so smooth.

So, I can only say that I relate to what you are saying, there's some "magic" going on with m43 and Panasonic lenses, something that is not on all the charts and tests published out there, but that a photographer can relate to.

Thanks for sharing. At least now i don't feel like i'm going mad.

Anonymous said...

even dxomark doesn't say K-5 has 3.5 stops better lowlight capability (it's on average 1 to 1.5 stops on dxomark graph). And high ISO isn't a requirement for most people who take photos with lights on!

Carl said...

"Thanks for sharing. At least now i don't feel like i'm going mad."

No you're not going mad. A friend who is very picky, and partial to contact prints from large format film negatives, looked at a set of a dozen or more GF1 prints at 20.5" wide and his reaction was "these are much better than they have any right to be!"

Russ Fortson said...

Hi Carl, interesting comments, though I can't say I fully agree, but then that's why they make different cameras.

One thing that I did find curious was your comment on the flash system. I haven't experienced any problems, and haven't really read about any on the Pentax areas I peruse. Any chance you had a lemon?

Any way, thanks for the alternative view. The one thing I take from this is that pretty much any camera can give great results now days, and it boils down to what you need.

Anonymous said...


I find your comments regarding the K5 + lenses quality a little 'disturbing'! I have a K5 + 18-135mm and 43mm and have had some 'softness' issues but this has certainly improved with the 43mm with suitable AF adjustments.

I also have a GH2 because I couldn't make my mind up which I wanted and have mostly used it with the 14-140mm.

I certainly find I get more consistently sharp images from the GH2 but there is something 'odd' about the look - to my mind the images have a more 'clinical' look to them whereas the K5 images just seem 'deeper' or 'richer'. I am not sure how to explain it but there is more to it than just sharpness and non-distortion.

Having said that I am not sure if the GH2 'look' is growing on me ...