[FullertonImages]

How about the fact that they are both horribly underexposed? A K10d looks better than that at 1600 if it's properly exposed.

[silverflower]

...Both raw files (K7 using DNG, K20D using PEF), imported into iPhoto and screen captured in TIFF. ...

- Just a thought, maybe some raw converters didn't have yet proper profiles for such a new camera and use of Pentax Photo Laboratory 4 for both might give more accurate results.

[er1kksen]

How about the fact that they are both horribly underexposed? A K10d looks better than that at 1600 if it's properly exposed.

that's a good point. One thing that always amuses me about the DCResource night-scene noise tests is that they always expose the Canon and Nikon night shots at least 1 EV higher than the Pentax and Olympus shots... and as a result, the Canikon results have a lot less noise (they're receiving more light, after all). I mean, I'll give him the benefit of the doubt that it's just due to differing lighting conditions from day to day and not brand bias, but still, the methodology is lacking...

[dantekgeek]

A worrying trend is certainly starting to appear, at least from the comparisons I've seen.

I agree that the 1600 example gives the K-7 the edge, with "nicer" noise (although more), and more detail, but at 3200, things really start to go wrong.

I wonder if the issues we're seeing might be addressed in a firmware update, or if we've just seen a run of badly done tests?

[solar1]

My feeling at this point is that the K-7 ISO 1600 really does look much better.

Things are likely not fully minutely finessed through the fine adjustment tuning phase yet. I'm content to wait a while for 1 to 2 firmware updates to really see what the K-7 can do. I'll be ready to buy :-)

Edit: I'm awaiting some really well exposed shots run through Pentax software, i.e. made for the camera specifically.

[benjikan]

Was this shot in RAW? Because if it was...well...WTF???

[krypticide]

Fix the white balance. I've found that changing the white balance can bring in significant blue/purple spots like you see there. Noise itself doesn't bother me; it's the spots that do.

[Ash]

OK, what's the story here?
If someone who shoots RAW with the same settings on both cams can make the comparison, then I'd be keen to see whether the verdict changes...

Sounds like here there's enough preferring each side to say that both sensors give similar results...

[celetron]

I am not that sure about that....

I reduced the noise (not aggressively) at the 1600 ISO sample of k7 in order to get it close to the noise levels of k20 and the result was much more pleasing than the one of k20...

If in fact k20 used NR and k7 not then it is obvious that k7 sensor/(in camera software) combo is superior to the k20 one...

here is the result:

[kenyee]

Interesting...look in the upper left at the yellow area. The K20D has reddish noise. The K7 does not which I like better...

[Ash]

Looks like more of a WB issue.
Noise-wise, hard to distinguish, though I too would agree the K-7 has the upper hand.
Detail retention looks quite similar.
All very subjective, though...

[garyk]

I have both and shoot with a telephoto lens. And i will tell you the 20 is much better.
It is clunky but sharper images by far.
G

[robgo2]

Sorry, but I disagree. K7 has more detail, better colors, better white balance. You have helped persuade me to buy one.

Thanks

I agree completely. The K-7 shows much more detail and appears more natural. Its noise is film-like and relatively unobtrusive. The K20D looks smeared. I presume that we are looking at 100% crops. That is not how images are normally viewed. The proper goal of photography is to produce images that are pleasing and have impact, not images that are completely noise free. To make noise reduction an end in itself is very misguided, IMO.

BTW, no APS-C camera can be said to be excellent at high ISO. For that, you need FF, which carries it own trade-offs.

Rob

[jamesm007]

I have both and shoot with a telephoto lens. And i will tell you the 20 is much better.
It is clunky but sharper images by far.
G

Its an old post when the K-7 first came out. I remember the debates (heated) of the K-7 being better. It had to be, it was new, JohnCPentax said it would have less noise... but we all know the truth now. Even it you take out the slight NR of the K20D, its still cleaner a bit, at high ISO. The K20D used a two channel design and K-7 a 4 channel which probably increased the noise a tad, this 4 channels was for faster FPS. Even Samsung's NX10 (who people thought after one more year of development would be superior in noise at high ISO) is not any better than the K-7 if as good(?)... strange. Take a look at DPR review for some comparisons, or DXO or just about any reliable source.

As a personal note I feel Pentax got a bad deal from Samsung, they were probably expecting right to the end a faster cleaner sensor. It just never quite happened. There close, but the K20D is just better enough to notice. Not that it makes much difference as realistically you may as well say they are the same(?)... well

Now to counter this I bet Pentax made a deal with the devil and got to this day a sensor no one else has for its entry level K-x dSLR. This thing is really good, plain and simple. Does it have great ISO6400, no, but the pics are printable at 10X8, and its arguable the best high ISO APS sensor out; don't try that with the K20D without some serious PP work and luck. Just guessing Pentax had to save face, and did not want to be known as the poor high ISO manufacture as at this time that's the popular seller, mega-pixels, high ISO... and it worked.

[Pentaxor]

Even it you take out the slight NR of the K20D, its still cleaner a bit, at high ISO. The K20D used a two channel design and K-7 a 4 channel which probably increased the noise a tad, this 4 channels was for faster FPS. Take a look at DPR review for some comparisons, or DXO or just about any reliable source.

There close, but the K20D is just better enough to notice. Not that it makes much difference as realistically you may as well say they are the same(?)... well

Now to counter this I bet Pentax made a deal with the devil and got to this day a sensor no one else has for its entry level K-x dSLR. This thing is really good, plain and simple. Does it have great ISO6400, no, but the pics are printable at 10X8, and its arguable the best high ISO APS sensor out; don't try that with the K20D without some serious PP work and luck.

from Dpreview, read the highlighted parts.

Image Quality

No reason to complain at base ISO: the K-7 images show good detail and natural colors straight out the box. Compared to the predecessor in-camera sharpening has been reduced which results in slightly softer but also cleaner looking output. Pixel peepers can revert to shooting RAW which, in combination with some careful sharpening during the conversion process, will result in some visible extra image detail. It's advisable though to use quality lenses if you want to make the most of the camera's 14.6 megapixel nominal resolution.

At higher sensitivities the picture doesn't look quite so good. JPEG noise is visibly and measurably higher than on some of the direct competitors. At default settings the K-7 retains relatively good detail up to the highest ISO settings but shows fairly large amounts of both luminance and chroma noise. Increasing the in-camera reduction will add some pretty heavy detail blurring to the mix. Some rivals, especially the Nikon D300, deliver a better balanced mix of noise reduction and detail retention. However, when shooting in RAW the K-7's RAW noise is pretty much on the same level as the competition. Therefore, for maximum image quality in low light situations your best bet is shooting RAW and applying customized noise reduction in post processing.


All in all, apart from a slightly different tone curve, weaker default sharpening and slightly higher (!) RAW noise the K-7 output is very similar to the K20D. Purely from an image quality point of view there is therefore no urgent need to upgrade for current K20D users.

also try to look for JohnBee's tutorial on NR pp and look at the sample images done at ISO3200 and ISO 6400. really darn clean images with great detail. and it's not that complicated or difficult that you might think.

and Yes, my RAW images beat the hell out of my in-camera jpegs ay any given day.