[Raybo]

In these examples, the top of the bottle shows more detail in the K-7 shot, because it is in perfect focus, while the K-20D shot is not. I don't know if sharp focus makes noise more apparent, but underexposure does. It looks to be more than -0.3EV, but I am not the one doing the measurement.

Rob

Right you are Rob!

I'm going to call "shananigans" on the images.

[nostatic]

If I had to guess I'd say motion blur as I don't see anything particularly sharp (and I have a very good copy of the 50-135). But what I am seeing so far is more "keepers" with the k7 under these bad conditions. While perhaps no sane person shoots that I do regularly. Whether it is just better low light AF or something else I'm not sure. The other thing is that the raw files have a slightly different look to them. Whether that is due to NR or AA filter or sensor/firmware difference I don't know but I think I see a difference in the way the noise is rendered.

No shenanigans. They are what they are. I'll leave it to someone else to do the carefully controlled brick wall photos. I just shoot and move on...

[PentaxPoke]

Edit: pics in this post removed. A much better test is in post 30 below.

I'm just not seeing a lot of difference here folks. Also, I completely agree with static about the number of "keepers."

[kittykat46]

If I had to guess I'd say motion blur as I don't see anything particularly sharp (and I have a very good copy of the 50-135). But what I am seeing so far is more "keepers" with the k7 under these bad conditions. While perhaps no sane person shoots that I do regularly. Whether it is just better low light AF or something else I'm not sure. The other thing is that the raw files have a slightly different look to them. Whether that is due to NR or AA filter or sensor/firmware difference I don't know but I think I see a difference in the way the noise is rendered.

No shenanigans. They are what they are. I'll leave it to someone else to do the carefully controlled brick wall photos. I just shoot and move on...

What I'm seeing is the K-7 definitely has faster, more accurate low-lighf AF, especially under incandescent lighting.
If you shoot hand-held, K-7 also has better SR.

Put all of that together = "More low-light keepers"

[eva2000]

thanks for the comparison results I don't own or ever used K20D, i went from K200D to K-7 and love the improvements for high iso and low light AF

[grishazzz]

PentaxPoke
Thanks for the comparison results!!! Now I know for sure: I will not buy K-7. I'd rather wait for Pentax FF, or I will get another brand.

[frogger]

:-( In those comparison shots the K-7 crop shot does look like a step back from the K20D. I'm looking at the noise on the lens.

[kittykat46]

I don't really agree the K-7 shots show a step back in IQ vs. the K20D.
The two are very similar, IQ-wise, apart from the K-7's much better AWB, which won't matter if you shoot RAW.

I've taken the liberty to pixel-peep with Pentax Poke's shots, with due acknowledgement.

The model farm house above has an incredible amount of texture detail, which is a great test for both cameras. I lined them up in my side-by-side viewer. If anything, the K-7 has retained a "slightly" greater amount of detail, which the K20D has smeared away if you look carefully. The K-7 shows more noise, but the photographer is at liberty to clean it up either in Post-Processing, or under JPEG NR settings. I'd rather the camera sensor retain more detail up front.

The 1st crop is from the K20D, the 2nd crop is from the K-7.

[Ash]

This may be explained by less NR processed on the K-7 image, permitting 'more' detail, whereas also retaining more noise.

I think by the few comparison tests so far, it's fair to say the sensor performance is similar for still images, though K-7's refinement of the sensor was to permit 'longer' live-view, which means video capability...

[nostatic]

This may be explained by less NR processed on the K-7 image, permitting 'more' detail, whereas also retaining more noise.

I think by the few comparison tests so far, it's fair to say the sensor performance is similar for still images, though K-7's refinement of the sensor was to permit 'longer' live-view, which means video capability...

If that is the case then I say, "GREAT!" to Pentax. The photographer should have the choice.

I know there is a lot of call for "high iso/low noise" but why is that? I want high iso as I shoot handheld in low light a lot. But I know that there is no free lunch - a sensor can only do so much. So to get the "low noise" part the manufacturers have to use math trickery. I personally do not want that.

To me what is important is getting "keepers" in low light. And right now the K7 seems to be much better at that than any previous Pentax camera, and I'm guess is on par with any comparably priced Canikon. If a Canikon has a less noisy image then great for them, but from examples I've seen it isn't going to be night and day, and frankly I don't like the way that many cameras render high iso noise.

If the holy grail is to mimic film, then give me the noise at high iso - don't smear it with NR. Or at least give me the choice. I want keepers that I can print 16x24 and have them look good to my eye, not files that I can pixel peep some small area over and over seeing if there is "noise." But different strokes...

[PentaxPoke]

I don't really agree the K-7 shots show a step back in IQ vs. the K20D.
The two are very similar, IQ-wise, apart from the K-7's much better AWB, which won't matter if you shoot RAW.

I've taken the liberty to pixel-peep with Pentax Poke's shots, with due acknowledgement.

The model farm house above has an incredible amount of texture detail, which is a great test for both cameras. I lined them up in my side-by-side viewer. If anything, the K-7 has retained a "slightly" greater amount of detail, which the K20D has smeared away if you look carefully. The K-7 shows more noise, but the photographer is at liberty to clean it up either in Post-Processing, or under JPEG NR settings. I'd rather the camera sensor retain more detail up front.

The 1st crop is from the K20D, the 2nd crop is from the K-7.

I agree kitty. Static shows the same things in his shots, so I think what we are seeing is consistent. K-7 shows more detail, k20d has more built-in NR.

[pcarfan]

To me, your iso 1600 shots with the K7 looks as good as iso 400 with my K10D...not just the noise but the color, contrast and sharpness. K20D is close. This is an excellent improvement over the K10D. If I decide to stay with Pentax, K7 will definitely be a HUGE improvement over the K10D....I can't wait to get my next DSLR....I have a feeling it will be the K-7. Thank you for posting these.

[jstevewhite]

The model farm house above has an incredible amount of texture detail, which is a great test for both cameras. I lined them up in my side-by-side viewer. If anything, the K-7 has retained a "slightly" greater amount of detail, which the K20D has smeared away if you look carefully. The K-7 shows more noise, but the photographer is at liberty to clean it up either in Post-Processing, or under JPEG NR settings. I'd rather the camera sensor retain more detail up front.

The 1st crop is from the K20D, the 2nd crop is from the K-7.

In this comparison it looks to me like Pentax/Samsung has increased the sensitivity of the sensor like, maybe a third of a stop. The K20D looks underexposed. I'd be interested in seeing the same shot with +.3EV on the K20D... I think that might produce essentially identical images. I'd shoot 'em myself, but I don't have a K-7

I think the greater number of "keepers" from the K-7 in low light probably has to do with better SR and better AF rather than anything else, but a sensor that's .5 or .3 stops "faster" probably helps, too. (not positing that as an original thought, but as agreement with some others in the thread.)

[jstevewhite]

Here are three images from my K20D. I don't have a k-7, sorry. These are ISO3200, tripod mounted, at -.3EV, 0EV, and +.3EV. Here's the grid view:



I corrected them in LR to 0EV in PP. They're shot RAW PEF, light from overcast coming from distant south window. I labeled each one. The noise is very clumpy and ugly at -0.3EV, a little better at 0EV, and best at +0.3EV. I think this illustrates what a sensor that was .3 stops more sensitive might mean to high-ISO noise profiles, and how important metering is to noise. I think the EXIF is preserved if you want to snoop it.

Here is - 0.3EV:



Here is 0EV:



and at last +0.3EV:

[PentaxPoke]

OK, I did one more test that you pay prefer. This one has the following conditions:

-f/16 to get max DOF and eliminate any concerns about NR smearing vs lens blur. I also used my best lens on each camera: the DA70 ltd.

-AB800 studio strobe lighting and X-sync on cameras at the exact same power settings to make the lighting consistent.

-Scene selected that has a lot of detail, and a lot of contrast so you can see noise in shadows. The histogram runs from almost saturated, to almost black. There is a lot of detail in both the light and the black (eg, black wood grain)

-all 4 NR settings used on each camera. An iso 100 is also shot for a "noiseless" comparison.

The full set with all user NR settings is in the original folder (feel free to download, and crop, etc.) :

K-7 and K20d Iso tests - a set on Flickr


Teaser pics:

(Note that below i show the factory default NR settings for each camera since the user NR setting is relative. "NR off" for example, does not mean that there is no NR done in the camera image processing.)

K-7 iso 100



K-7 iso 3200 default NR:



k20d iso 3200 default NR:




My conclusion? Dang these cameras have good high iso performance (when properly exposed)