[aerodave]

For a long time browsing these forums, I've seen many people mention that they refuse to push the sensitivity on their cameras, sometimes even thinking ISO 400 is too high. Looking back, I realize that most of these people were users of the K10D or K20D. Meanwhile, I was gladly using my K100D at ISO 800 without a second thought, and would gladly move to 1600 (with quite acceptable results) if the situation called for it. I figured there shouldn't be any huge performance differences between, say, the K10D and my camera, especially since the K10D was a much more refined and expensive model. So I just assumed that the people who wouldn't dare use ISO 800 on their higher-pixel count cameras simply had higher standards for image quality than I do.

But recently I picked up a used K10D from here on the marketplace. I wanted to get a SDM-capable body to use with my new DA 17-70mm, but I'm not quite ready to pull the trigger on a K-7 until the price comes down a bit. The K10D will do a fine job of tiding me over until then. I'm enjoying my new camera, and it's a tremendous improvement over that K100D in many ways. But there is one area where it feels like a step back, and that is in low-light performance. I finally understand why so many photographers on here have wanted to stay away from high ISO. The performance at, say, ISO 800 (or especially 1600!) is considerably noisier than the K100D's. I know that higher pixel densities mean more noise, but I wasn't expecting the difference to be as pronounced as I saw. You can push the K100D at least a whole stop farther in ISO before the noise levels look the same.

I'm not here to bash the 10- and 14-MP cameras, especially since I plan on gettig one myself...and I love my new-to-me K10D. But I think the old K100D with its (laughable by today's standards) 6 megapixels, is an underrated low-light performer. I'm not so sure I want to get rid of it anymore, and maybe it would make a great second body to hold on to: my low-light workhorse. I think I'd gladly give up some resolution for a clean pic with reasonable shutter speed in low light.

I'm away from my photo catalog and don't have any pics to prove my point. But if you have any examples of how your K100D has really come through in dim light, feel free to post them. Let's give that old camera the recognition it deserves! =)

[Ash]

Hi Dave.

6 Mp is certainly not laughable - one can print an A4 sized photo with no problems with 6Mp, which is more than most people would need. Cropping is where extra Mp would be handy, though...

On K10D vs K100D noise, this is well known. Lots of discussion occurred about this when K10D first came out, and became quite a point of concern for some. Most people will say that ISO 1600 is unusable on the K10D, but again, if adequately exposed, particularly with extra lighting/flash, then you may be surprised with the results. No-one would argue, though, that high-ISO performance took a step down from the K100D to the K10D.

Where you will find a decent gain is from the K100D to the K20D. The more refined CMOS sensor produces beautiful images at ISO 1600 (but again need to be well-exposed) and even performs OK at ISO 3200. Forget about ISO 6400, though.

And with 14Mp, there is that much more versatility to what you can do, but say you just want a 6Mp image, then if you downsize a 14Mp image to 6Mp, you'll get results at ISO 1600 and 3200 you never thought possible before...

[shaolin95]

Interesting, I am going to compare my K100D to my K20D over the weekend. :-)

[superfuzzy]

I have both - k100 and k20. The k100 is still a nice little camera. AF not so good in low or moderate light, but with a wide lens, such as the da 12-24mm it's shines!

[shaolin95]

I want that 12-24mm or the sigma 10-20mm so bad!
I will try to post some images later this week.

[aerodave]

Ash, you make some very good points. I know that the newer models will have better performance, and I'm still looking forward to seeing what my future K-7 will do at high ISO.

And of course I didn't think I had breaking news about the old models that hadn't been discussed many times before. I think it's just an example of how first-hand experience really puts things in perspective.

So while my observations aren't news to many people on here, I've got a whole new appreciation of my old friend.

[Marc Sabatella]

For a long time browsing these forums, I've seen many people mention that they refuse to push the sensitivity on their cameras, sometimes even thinking ISO 400 is too high. Looking back, I realize that most of these people were users of the K10D or K20D.

Perhaps - especially regarding the K10D, which is the weakest of the Pentax cameras in high ISO performance. Mostly, it really is just a question of how picky you are about the results. But also, a question of the scene and your exposure - an ISO 1600 picture taken of one scene with one exposure might look much better or much worse than an ISO 1600 picture taken with the same camera but of a different scene and/or at a different exposure.

I know that higher pixel densities mean more noise, but I wasn't expecting the difference to be as pronounced as I saw. You can push the K100D at least a whole stop farther in ISO before the noise levels look the same.

Are you comparing at 100%? That's not a fair comparison - you're blowing the image (and its noise) up bigger on the K10D. compare both at the same size - and make sure it's the same subject, same exposure - and see if the difference is really *that* great. No question the K10D is not as good as other cameras at high ISO, but aside cases where the "banding" that the K10D is prone too, it shouldn't be a whole stop of difference when performing a fair comparison.

BTW, aside form the fact that having more pixels (and hence higher pixel densities) means you are tempted to blwo the image up bigger, that doens't in itself cause more noise. That's kind of a myth, and see the thread on noise FF versus APS-C for more on this. According to someone (GordonBGood) who has extensively analyzed the RAW data from different Pentax cameras, as well as the design specs of the cameras themselves, what does the K10D in was the decision not to use an analog amplifier to implement higher ISO but to try to take advantage of the 22 bit ADC and do it all digitally. According to Gordon, this would have worked in theory, but the way they actually went about it was flawed.

But I think the old K100D with its (laughable by today's standards) 6 megapixels, is an underrated low-light performer.

Actually, it's pretty consistently lauded for this, so I wouldn't call it underrated. but I do think if you perform a more carefully controlled experiment, you might find the gap between it and the K10d not quite as great as you think, and in any case, that's unique to the K10D. The other higher-resolution cameras use a traditional analog amplifier and when comparing image at the same size, differences in noise are practically indiscernible. Or, perhpas more accurately stated, you'd have difficulty reaching any sort of consensus in a blind test which was actually better.

[Alfisti]

I have both - k100 and k20. The k100 is still a nice little camera. AF not so good in low or moderate light, but with a wide lens, such as the da 12-24mm it's shines!

Interesting comment, the 12-24 is the one lens I just cannot seem to get to work as well on the 20D as it did on the 100D.

As for noise, reduce the 20D files to 6mp and the mirror the 100D

[aerodave]

Are you comparing at 100%? That's not a fair comparison - you're blowing the image (and its noise) up bigger on the K10D.

I completely agree that 100% view puts the K10D at a disadvantage based on zooming in closer. But from a sensor performance standpoint, that's the real apples-to-apples comparison. You buy higher resolution to let you crop more or blow up your prints bigger. So a higher-resolution sensor should be expected to be zoomed in more. If I'm going to keep them both the same size for everything, I'm wasting 4 million pixels.

Don't get me wrong, Marc; I think you rase a very good point. I think it's definitely an application-dependent thing, and whether you need to take advantage of the higher resolution depends a lot on the shot. For purposes where getting to 100% view (or in the case of print, a size where the pixels start to become noticeable) is not the goal, the differences between the K10D and the K100D start to disappear. And for most of what I do, that's the case. But sometimes I pixel-peep...and that's when I notice.

[Alfisti]

I completely agree that 100% view puts the K10D at a disadvantage based on zooming in closer. But from a sensor performance standpoint, that's the real apples-to-apples comparison. You buy higher resolution to let you crop more or blow up your prints bigger. So a higher-resolution sensor should be expected to be zoomed in more. If I'm going to keep them both the same size for everything, I'm wasting 4 million pixels.

Don't get me wrong, Marc; I think you rase a very good point. I think it's definitely an application-dependent thing, and whether you need to take advantage of the higher resolution depends a lot on the shot. For purposes where getting to 100% view (or in the case of print, a size where the pixels start to become noticeable) is not the goal, the differences between the K10D and the K100D start to disappear. And for most of what I do, that's the case. But sometimes I pixel-peep...and that's when I notice.

Sort of, in reality you should then upsample the 6mp image to 10mp to reflect a print.

[aerodave]

Sort of, in reality you should then upsample the 6mp image to 10mp to reflect a print.

That's pretty unfair to the 6mp camera, since upsampling is far more damaging than downsampling.

Here's an analogy: You're an engineer involved in crash testing automobiles. And for the purposes of your current research, you care about evaluating the crashworthiness of cars at the limits of their expected real-world performance. You know the Ferrari is going to rate poorly because it's meant to be driven fast. But you could just manipulate the results of the Ferrari you're testing by slowing it way down and saying "Surely the owner will keep it under the speed limit!" That's what downsampling the 10mp camera is like. It's giving it a handicap that might not reflect real world use.

Upsampling the 6mp camera, however, is aking to strapping rockets to the Honda Civic to make a more spectacular crash. Not particularly fair to the slower vehicle.

[Alfisti]

Not a fan of analogies myself.

You can't just sample one way, to make an 11*14 print you need to upsample a 6mp image quite aggressively whereas a 14mp image is essentially at 100%.

If both are shot at ISO1600 you'll see very little if any difference but at 100% on screen the 100D will look far better.

The only fair comparison is that the same display size not 100%.

Now DR is another thing altogether.

Well I have the K100d and the K20d....

If I am really doing a night of low-light shots of people... the K100d always gets the nod.

I thought I would get rid of the K100 too... but it is such a great little concert/go anywhere shooter... NEVER! From my cold, dead hands.

[aerodave]

Not a fan of analogies myself.

Well you're no fun.

The only fair comparison is that the same display size not 100%.

Again, let me say that I completely understand that it's an application-dependent thing. Sometimes having the same display size is the fairest comparison. Sometimes it's viewing 100%. It completely depends on what your objectives are and what you want to do with the images.

It's like the DOF/cropped sensor debate raging in another thread here. There's no way to come up with a one-size-fits-all measure of merit for anything related to a tool's performance, because the things that matter change with how that tool is being used.

I'm really not trying to be argumentative. My original point was that, for the way I use my cameras, I find the K100D to have much more satisfying low-light performance. How I am using my equipment is leading me to that conclusion, so it's a valid one...for me at least.

[Marc Sabatella]

I completely agree that 100% view puts the K10D at a disadvantage based on zooming in closer. But from a sensor performance standpoint, that's the real apples-to-apples comparison. You buy higher resolution to let you crop more or blow up your prints bigger. So a higher-resolution sensor should be expected to be zoomed in more. If I'm going to keep them both the same size for everything, I'm wasting 4 million pixels.

But it's not like the K100D is going to look any better when doing larger prints or extreme crops. *All* cameras will have worse noise at larger print sizes than small ones (ditto for crops). It's not fair to say the K10D is worse than the K100D just because it looks worse at a large print size than the K100D looks at a smaller one. If you want to print big, fine - but then, do that for *both* cameras and then compare. and you'll find they *both* look worse than the K100D did at the smaller size.

The point being, increasing pixel count isn't what makes noise worse - going to a larger print size is. While the higher resolution camera might not give you *better* noise performance at a given print size, it isn't making it worse, either. It's basically a non-factor.

Now, of course you are right to ask yourself if you're just "wasting" those extra pixels if they provide no advantage. Well, they *do* provide an advantage. First, in low ISO shots, no question you'll get more resolution from the camera with more pixels. Second in high ISO, what that extra pixels really give you is the ability to do more aggressive NR and still retain detail. That is, the K10D and K100D might be similar when comparing at the same size but not doing any additional NR. But given tha the K10D starts off with more pxiels, and thus more resolution, you can perform some relatively heavy NR if you want on that original file before resizing and still retain as much detail as the K100D. So that by the time you downsize it, it actually has much detail but *less* noise thn the K100D.

At least, all of this would be true if not for the specific flaw in the K10D design I mentioned (which i just learned about yesterday; until then I had no idea why the K10D always came out anomalously worse in noise comparisons). That is, the K10D is sort of unique case, because it really does have some disadvantages in the noise department due to its extremely unusual signal amplification design - not its pixel count. The K200D uses the same sensor as the K10D but the same type of signal amplification design as the K100D, and thus makes a better point of comparison. And indeed, I've found through countless direct comparisons that what I say holds true.