[audiobomber]

Sorry this post is just what I've come to expect on these boards; unsubstantiated talk. If you can provide detailed tests backing up your claims, lets see them. I have access to two K20D bodies and both exhibit the same noise levels. DxO numbers are just .... numbers. I'm providing test images.

No need to be sorry, we're just discussing your test results. I can't do the comparisons without a 7D. I do have a K20D and it's never produced such ugly ISO 1600 files, even with NR turned off. I've read raw noise tests on the K20D, K-7, D300S and 7D, at DPR and DXOMark, and your results don't correlate. So who should I believe, Some Guy On The Internet trying to show the 7D is superior, or my own observations plus professional testing? I don't blame you for being pissed that Ogl said the K20D was so much better, but everything I read says the K20D and 7D are neck and neck for high ISO performance. That is clearly not what your images show.

Did you not read that I said the NN profiles was set to "0" for luminance and "0" for colour? This is minimal noise reduction, I could apply a lot more to both files to clean them up more; but to repeat myself yet again, this was to show high ISO noise, not the effectiveness of Noise Ninja and not the in-camera noise reduction of the bodies. It is RAW sensor output, so your "I use weak and it looks better" is a moot point as we are not testing in-camera NR.

I read this; "Noise Reduction and sharpening applied in Noise Ninja". There's not much point in doing your own noise reduction if the camera can do a better job. I suggest that just turning off NR in both cameras would be a better representation than a random tweak in Noise Ninja. I have some test photos in my hard drive that were taken at 1600, 3200 and 6400 with the K20D and NR off. Your ISO 1600 file looks worse than my ISO 3200 file. I can post them for you this evening if you like (I'm at work now).

In looking at your results, I suspect you missed focus with the K20D. As everyone knows, the K20 AF does not do well under tungsten light. There has to be some explanation for the way these files look.

[Big G]

This is getting insane.

JUST FOR YOU, I will repeat the tests and allow the camera bodies to do the noise reduction.

Oh, incase you didn't read my post, focus was done manually using live view to bypass any suggestions that one camera didn't focus properly so once again another moot point.

[pingflood]

This is getting insane.

JUST FOR YOU, I will repeat the tests and allow the camera bodies to do the noise reduction.

Well, now you're screwed. You will need to post every conceivable combination of in-camera reduction and suffer endless arguments about whether "medium" on one equals "low" on another and so on and so forth.

I will share ISO 12800 from a 7D, that I posted on another thread.



And 100% crop to avoid the "well Canon obliterates detail" argument.

[audiobomber]

This is getting insane.

JUST FOR YOU, I will repeat the tests and allow the camera bodies to do the noise reduction.

Why not just turn NR off? I did my test during daylight in a darkened room, using MF and AF with a lens I know focusses accurately on a tripod, with 2s timer to disable SR. There were some misses. I suspect it was because 2s wasn't enough to let the tripod settle down on the carpet. I shot a few at each ISO and took the best one of each set.

[audiobomber]

I will share ISO 12800 from a 7D, that I posted on another thread.
And 100% crop to avoid the "well Canon obliterates detail" argument.

I'm impressed. I don't think the K20 could do this, even if it had a 12800 setting.

[pingflood]

I'm impressed. I don't think the K20 could do this, even if it had a 12800 setting.

Well, the K20D is also quite a bit older. Looking through my K20D files I would probably be ok with shooting at 1600 where needed, but past that I would not want to go.

The 7D I will shoot at 3200 all day and 6400 is not a major worry. But it is a lot newer than a K20D!

[Big G]

Well, now you're screwed. You will need to post every conceivable combination of in-camera reduction and suffer endless arguments about whether "medium" on one equals "low" on another and so on and so forth.

Yeah exactly, it's not going to be possible and the "ahh, but you shouldn't use weak, you need to use strong" arguments will go on for pages ad nauseum.

Why not just turn NR off?

Dude, you really aren't reading any posts. NR was turned OFF in the initial tests with a very minimal noise reduction / sharpening profile applied in Noise Ninja since they were straight RAW to TIFF files.

EDIT: Just to stress and reiterate, this is not a K20D bashing thread. I am in constant praise of the K20D given it's age, build quality and the impressive IQ it produces on landscapes at ISO 100. What this thread is for is to debunk the utter drivel on the other thread that the 7D sucks at high ISO. Even of the face of utterly simple RAW to TIFF image tests with virtually identical settings / lenses, the disputes roll on.

[youky63]

I support you Big G on this thread. I also have a K20D, that I also use with NR off. More than 15.000 shots later, I can tell your picture has the very typical pattern of this camera. And sure 7D offers better high iso results.

[JohnBee]

I'm impressed. I don't think the K20 could do this, even if it had a 12800 setting.

You are right it couldn't.
In fact... I'd go as far as saying a K20D most likely wouldn't have the DR to manage the highlights in this image, at ISO6400.

[str8talk83]

People are getting a little defensive here..... I would be interested to see the low ISO noise comparison. One of the weaknesses of the 7D that I have been hearing about is how it sacrifices low ISO noise performance for better high ISO results. I have seen real world examples of noise in skies for example in ISO 100 shots. Can easily be fixed with some minor noise reduction (and still providing better results than my 40D's due to higher resolution) but interesting nonetheless. I think it is pretty obvious that the 7D is better than the K20D in this respect which is no surprise really. Newer tech (generally) equals better performance.

The post about the D7000 being equal to the noise performance of the D3 made me laugh a little bit. I think that people are dreaming in general to think that an APS-C sized sensor will equal a FF sensor at this time. If a company stuck with a certain MP count (kind of like Nikon with 12MP) and perfected it over 3 or 4 years they MIGHT be able to equal the D3 noise performance. Now the D3s....forget about it! While the 7D is great for a crop sensor body, it still is around the same performance (from what I've seen) as the Canon 5D which is a much older design, but has the larger sensor.

[er1kksen]

I've seen noise in the skies with my K20D at ISO 100 and with my 40D at ISO 100 as well, for that matter. You can't really sacrifice low ISO quality to boost the high ISO (except by designing for a higher base ISO). That's just not really how sensors work, though a lot of folks on the internet like to imply that it happens... it's sort of like the pixel density=noise myth (the principle is basically correct, but it doesn't show up in real life at DSLR pixel densities). The 7D might "see" more noise in, for example, the sky at base ISO, but that's because it samples both the signal and the noise on the sensor at a higher rate. A sensor that samples that signal and noise at a lower rate will have some of the noise averaged out and look smoother, even though at equal output sizes they have the same noise levels.

[dylansalt]

The 7D is a decent compromise though.

For anything better at high iso, one will have to fork out for the d3s.