[Alfisti]

Looking at more images now, anyone saying the 20D is better noise wise than the 100D from ISO800 up is on crack.

[farfisa]

I have both--the K20 images are excellent, like the K100 sensor but just more pixels!

One thing with the AF--EVERY one of my lenses has required a focus adjustment on the K20. Different AF system. Try some of that--go through every step from -10 to +10 if you have to.

I haven't been shooting RAW lately (switched to Jpeg on vacation and never switched back), so I can't supply any images right now, but I can shoot something Monday if you want. Also haven't investigated the noise, but the K20 has more pleasing noise than the K10 did (not a typo--had a K10 in between).

Try some manual focusing zoomed in with Live View on the K20. It starts to look really pixellated, and that's when you know it's in focus. Should work for the bookshelf test.

This is one of my favorite ISO 1600 shots from the K100d and the Tammy 28-75 at 75mm and F2.8 I think this shows the sensor and lens at their best...
From a New Years Eve Rave in SF. Look at the eyelash on here right eye... come on that is good!



And here was a freakish test at ISO 6400 with the K20 and the Helios 44M just to see what ISO 6400 looked like.

Is it poster-able...? not really, but in a pinch... maybe?
P.S. B&W is the friend of upper ISO...

[farfisa]

Gus, we're talking about boring images, not amazing images, k?

Okay back to boring images in crappy light (it's pissing out there!) because that's where we see the noise/detail problems:

I did a quick set up with the K100 and the K20, both shot RAW, 25% sharpening, just imported into Photoshop CS3 and exported for web at 80%. Shot at 100 and 800 ISO with the K20, and 200 and 800 with the K100.

So it looks like I missed focus on the K100, but this was the best of the shots I got, trying both auto and manual focus. The important thing is your NEW camera, right?

I used the 43 ltd at f/4 (don't have the 35/2).

Full test images--K100D, followed by K20D:




100% crops, 25% sharpening. Moved the tripod in between because I had to redo the K100 pics when I forgot to shoot RAW--d'oh!

K100 ISO 200:


K100 ISO 800:


K20 ISO 100:


K20 ISO 800:


I gotta say it'd be hard to beat that K20 performance at ISO 100.

Looking at the K20 images, I'm wondering if there's some sharpening in there after all, but there'd be a white outline around the text if I did...

I just added some edge crops--you can see the image plane was a little out, I guess, because the K100 seemed to have the text more in focus on the edge.
This also illustrates how maddening tests like this can be But at least I got a good result for the K20, which is what all this was trying to prove.

K100 edge:


K20 edge:

---

Last edited by farfisa; 05-16-2009 at 11:41 AM.

[OregonJim]

Full test images--K100D, followed by K20D:

Here's my take, FWIW:

The first two images look identical, and represent real-world viewing distances.

Everything else is just pixel-peeping. The K100D shots are a little OOF - other than that, I see no practical, qualitative differences at all.

To me, pixel-level comparisons on-screen are like looking at a supermodel's skin pores under a microscope - it has little meaning compared to the whole.

[farfisa]

Here's my take, FWIW:

The first two images look identical, and represent real-world viewing distances.

Everything else is just pixel-peeping. The K100D shots are a little OOF - other than that, I see no practical, qualitative differences at all.

To me, pixel-level comparisons on-screen are like looking at a supermodel's skin pores under a microscope - it has little meaning compared to the whole.

Yep, but the OP wants to pixel peep:

Just wondering, is your 100D sharper (as in WAY sharper) [than the K20D] at pixel level?

[Raybo]

I was seriously thinking about upgrading to a K20 (currently have the K100DS), but I just might wait to see how the K-7 turns out.

[Alfisti]

Here's my take, FWIW:

To me, pixel-level comparisons on-screen are like looking at a supermodel's skin pores under a microscope - it has little meaning compared to the whole.

That's absurd, it's the only way to check for focus accuracy and given how awful my first crop is it's hardly 'little meaning'. I use labs that demand 300ppi so if I print 11*14 or larger these 1:1's at 14mp start to count. No tto mention these 14mp files are slower to work with and the camera significantly larger and heavier than the 100D so i'd hope to get the extra pixels I paid for.

[Alfisti]

Thanks Farfisa, I feel better now because even though you've downsampled the 20D image is STILL looks noiser than the 100D which is what I am experiencing. Why is it then that i keep seeing "the 20D is better than the 100D at high ISO"?

The focus issues have been discovered by accident as I was testing for noise, i set the 35/2 to -4 and it is better but it is still VERY VERY unpredictable, sometimes bang on, sometimes slightly BF and sometimes it seems to well ... miss everything.

What a farking battle this is.

[farfisa]

Thanks Farfisa, I feel better now because even though you've downsampled the 20D image is STILL looks noiser than the 100D which is what I am experiencing. Why is it then that i keep seeing "the 20D is better than the 100D at high ISO"?

The focus issues have been discovered by accident as I was testing for noise, i set the 35/2 to -4 and it is better but it is still VERY VERY unpredictable, sometimes bang on, sometimes slightly BF and sometimes it seems to well ... miss everything.

What a farking battle this is.

Hey Alfisti, I hear you. On the one hand I can't believe I survived without focus adjustment, but on the other hand it's a whole other can of worms!

I've been back and forth with some of the adjustments too--like I'm convinced a certain lens needs a +3, then it's a +5 all of a sudden.

btw, I didn't downsample anything but the first two--the rest are 100% crops from the center (and edge for the last two).

I think the K20 sensor's great, but maybe the extra pixels mean that the noise is harsher because noise has to be per pixel, so you'll see a finer grain which might be less pleasing. Does that make sense?

Looking at the two ISO 800 crops though, I think the grain is a little softer on the K100 even at pixel level, but my image is softer to begin with, so this might not help.

However, I'm sure you'll get some great results out of the K20, and see that the extra pixels aren't just a marketing gimmick!--I think it was definitely worth the upgrade for me, but I still like to have the K100 around for its portability.

[HermanLee]

Thanks Farfisa, I feel better now because even though you've downsampled the 20D image is STILL looks noiser than the 100D which is what I am experiencing. Why is it then that i keep seeing "the 20D is better than the 100D at high ISO"?

The focus issues have been discovered by accident as I was testing for noise, i set the 35/2 to -4 and it is better but it is still VERY VERY unpredictable, sometimes bang on, sometimes slightly BF and sometimes it seems to well ... miss everything.

What a farking battle this is.

Alfisti,
Do you experience the inconsistent focus when shooting those books?
I'm sure you've known this but I just want to remind you about the size of the AF sensor is bigger than the red dot that we see in the VF.
And I think their position in every camera is also different slightly.
Maybe you want to check where your K20D's AF sensor is located first.

Last time somebody did the test with a blank paper and a black dot to locate the sensor size and area but I couldn't find that thread anymore. Maybe somebody knows where that thread is and can post the link.

[Alfisti]

Hey Alfisti, I hear you. On the one hand I can't believe I survived without focus adjustment, but on the other hand it's a whole other can of worms!

I've been back and forth with some of the adjustments too--like I'm convinced a certain lens needs a +3, then it's a +5 all of a sudden.

It's just farking SCHITZO. I'll point it at something and NOTHING is in focus then i do it again and it's bang on. It's stark raving mad i swear it.

[Alfisti]

Alfisti,
Do you experience the inconsistent focus when shooting those books?
I'm sure you've known this but I just want to remind you about the size of the AF sensor is bigger than the red dot that we see in the VF.

I have heard this and is there a reason for it? It's maddening, i'd happily MF but at wide angle i just can't do it.

[Marc Sabatella]

I'd assume the reason is that AF sensors as small as the actual red dots would be close to useless - way to often they'd be on an object without any edges or contrast. the bigger the sensor, the more likely they'll hit on something. Too big and they're hard to control. So a compromise has to be reached. That's my guess, anyhow.

BTW, regarding the nosie issue: I've looked at enough images from the K100D, K200D, and K20D to be convicned there really isn't that much differences in noise performance in any of them. So much affects the amount and appearance of noise in a shot - the color of the light, the exposure, the dynamic range of the scene, whether the subject is in the lights or the shadows or the midtones, the detail in the scene, etc. And of cours,e viewing 100% is unfair to the camera with more pixels. It's really tough to make apples-apples comparisons, but when you do, differences are usually in the you-have-to-pixel-peed-to-see-them category, and even then, you end up with half the viewers picking one as better and half the viewers picking the other.