My K-7 vs K-20d High ISO comparisons

PentaxPoke · 07-10-2009, 09:04 PM

I was interested in observing how the two cameras compare at high-iso with a highly detailed scene. I assume they should be close, because the sensor is essentially the same, but I wanted to do a quick comparison just the same. I know many have been doing this, but I thought I would add my tests to the forum.

Now this is not an entirely scientific comparison, but it is indicative of the results I have been seeing. I decided to just use the in-camera default NR settings for each camera. I also elected to output default jpeg files. I did keep the exposure settings identical for each camera since we all know how sensitive noise is to exposure, so I felt it was essential to keep that constant between the two cameras.

I also did another quick test that might interest some. It is 4 shots of the same subject at ISO 3200 with the K-7 at the 4 different NR settings.

UPDATE: jump to post # 30 for a more recent test I did: http://www.pentaxforums.com/forums/664653-post30.html

Here is the full set for the K-7 / K-20d comparisons:
K-7 and K20d Iso tests - a set on Flickr

Here is the full set for the NR variation tests on the K-7:
K-7 ISO 3200 with NR variation - a set on Flickr

Quick summary of my observations:

-White balance of k-7 is clearly superior. Maybe the best of any camera I have used. The tungsten lights in my house are very challenging for photography. I use commercial 130v bulbs instead of residential 120v. (The reason is they last a LOT longer, but as a trade-off, there is a bit less light, and it tends more toward yellow.) The K-7 has not had a single problem nailing the WB. I am quite impressed.

-As I have shown in similar tests here before, high-iso performance on the K20d is very good if you properly expose. When compared with the K-7, there are subtle differences, but nothing to suggest that the K-7 is worse or better. To me, that was to be expected.

-NR settings on the K-7 have an observable effect. Even on "high" there does not appear to be an objectionable loss in detail.

Did I mention that the K-7 is really fast?

For those that don't want to go to the link themselves, here is the k20d at ISO 3200:

And here is the K-7 at ISO 3200:

Raybo · 07-10-2009, 09:51 PM

If the WB was equal how would the images compare?
I'm seeing less noise in the K20 images (IMHO).

Ray

*Ash* · 07-10-2009, 10:02 PM

Thanks for sharing these results.
Nice summary.
Indeed AWB is well refined in the K-7. Not too much of an issue if shooting RAW, but still a great advance for those who like JPEGs out of camera.

Though I agree that in order to make better comparisons in noise, ALL settings should be fixed between the cameras and using the same lens, including WB, particularly when shooting JPEG.
As you've proven, even at ISO 3200 both cameras produce useable images, despite some loss of detail as expected at such high ISO settings.

Who would have dreamed of getting such high ISO results in the film era?

*nostatic* · 07-10-2009, 11:20 PM

another quick and dirty. K7 vs K20d, 50-135* at 50mm. Crappy low light, f2.8, iso 1600, 1/4-1/5s shutter speed. Shot raw and developed in Aperture2. k7 is -0.3ev.

k20 - no correction

k7 - no correction

k20 AWB correction (basically click a button)

k7 AWB correction

k20 crop of AF point

k7 crop of AF point

PentaxPoke · 07-10-2009, 11:35 PM

Thanks for posting those static. They really add to the thread: Mine shows default out-of-the-camera results while holding exposure constant, and yours shows the WB constant processed from RAW. I noticed you said that the K-7 was -0.3ev. Was that to try and equalize the exposure differences between the two sensors?

Your crops are also very interesting. The K-7 shows more noise, but shows significantly more detail, like the grain of the wood, and the ridges of that white cup. I am wondering now if the K20d by default, has more NR? Even if the user NR is "off", there still is noise reduction in the image processing of any camera.

*nostatic* · 07-10-2009, 11:47 PM

I have the k7 at -0.3 because earlier in the day I was noticing that 0ev was a little "hot" for my tastes. It is a little dark in this photo and 0ev would have been closer to the K20d under these conditions I think.

That said I agree with your comments. What I'm seeing is more "noise" and more detail from the K7. For what I shoot and what I'm looking for, I actually like that combination a lot. I tend to go high iso and output b&w. Any sort of NR ends up looking like smears when I work the file. The k7 seems to put out a lot more of a rough grain at high iso.

Here are the crops with some sharpening, contrast increase but no definition increase or highlight/shadow tweak.

Raybo · 07-11-2009, 12:05 AM

Whatever that white thing is looks "whiter" on the K-7.

I'm still wondering if the WB is equal? K-7 image looks to be a bit larger IMHO.....

Raybo · 07-11-2009, 12:07 AM

Originally Posted by nostatic

I have the k7 at -0.3 because earlier in the day I was noticing that 0ev was a little "hot" for my tastes. It is a little dark in this photo and 0ev would have been closer to the K20d under these conditions I think.

That said I agree with your comments. What I'm seeing is more "noise" and more detail from the K7. For what I shoot and what I'm looking for, I actually like that combination a lot. I tend to go high iso and output b&w. Any sort of NR ends up looking like smears when I work the file. The k7 seems to put out a lot more of a rough grain at high iso.

Here are the crops with some sharpening, contrast increase but no definition increase or highlight/shadow tweak.

You used different lenses........................Right?

PentaxPoke · 07-11-2009, 12:22 AM

In my set, the lenses are the same. I used the DA70 ltd.

Raybo · 07-11-2009, 12:33 AM

Originally Posted by PentaxPoke

In my set, the lenses are the same. I used the DA70 ltd.

Something is amiss..................the second set of images is a tad bigger.

Raybo · 07-11-2009, 12:36 AM

My bad PP!

Your images are okee dokee, nostatic's images are a bit off.

Sorry,
Ray

*nostatic* · 07-11-2009, 12:44 AM

same lens. Roughly the same position but it is handheld so there is a little difference. Shot raw and processed with one-click AWB in Aperture. The crops are roughly the same amount of area but not identical.

Sorry, but I don't do the rigorous, scientifically controlled tests as I don't shoot that way in real life. This approximates the type of conditions I'd normally use (handheld, low light). And I'm lazy.

robgo2 · 07-11-2009, 12:56 AM

Originally Posted by PentaxPoke

Your crops are also very interesting. The K-7 shows more noise, but shows significantly more detail, like the grain of the wood, and the ridges of that white cup. I am wondering now if the K20d by default, has more NR? Even if the user NR is "off", there still is noise reduction in the image processing of any camera.

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

*nostatic* · 07-11-2009, 01:10 AM

Originally Posted by robgo2

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

-0.3 is what was set on the camera and shows in exif for the k7. As for AF, both shot with the same admittedly half-assed procedure but that might also say something about the AF capabilities of the two bodies. But it seems to be fairly consistent - the K7 and K20d files have a different "look" at high iso.

jstevewhite · 07-11-2009, 01:12 AM

Originally Posted by robgo2

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

As nostatic said: "another quick and dirty. K7 vs K20d, 50-135* at 50mm. Crappy low light, f2.8, iso 1600, 1/4-1/5s shutter speed. Shot raw and developed in Aperture2. k7 is -0.3ev."

This is the 50-135 DA*. It's *sharp* on the K20D. I think at those shutter speeds we probably have some camera motion involved. The K20D shot is either OOF as you said, or blurred by motion - quite possible at 1/4 -1/5 second.