[ManuH]

"Mehlsack: I talked to a staff member who confirmed that thethering will be delivered later by firmware (final or later)"

Talk is cheap. Until there is an official commitment I'd rather put this in the rumors category. Same with other possible firmware improvements (CD-AF in video). I've seen enough products in my life making promises they didn't held.

[er1kksen]

Regardless of possible support issues, if someone could process that RAW file through ACR6 (either CS5 or LR3, whichever) and post a full-size jpeg of the result, that would be fantastic. I note that someone already tried it but only posted web-size samples...

It may not be a full indication of the K-5's ability due to preproduction firmware and incomplete RAW file support, but it'd be nice to see a "the K-5 will be at least this good if not better" sample.

[alessandro63]

I just tried to develop the dng (6400iso) file with RPP (Raw Photo Processor).
The "grain" is finer than LR example, and the detail in the halftone screen in the sheet even bigger.
Impressive. Really similar to a 1600iso with the K-7.
(RPP used without any noise removal nor sharpening gives usually a finer noise with any kind of camera, more similar to C1 than LR or Photoshop).
original crop

RPP

[tcdk]

Just tried to add two stops of exposure in LR3 and add +50/+50 NR.


Pentax K-5 High ISO +2 Exp

I must say that I'm impressed.

6400 looks as good as my K20D at 800 and quite usable (I use my K20D at 800 when I really need it, but never go much beyond that)

25600 cleans up as well as 1600 from my K20D.

If the AF is also better, it needs to be really expensive to not end up under my xmas tree

---

Last edited by tcdk; 09-22-2010 at 08:27 AM. Reason: Added preview

[raz]

Aren't we exaggerating a little ? I don't remember seeing iso 800 images with that much noise in the K7, K20D, K10D or K100D.

Not saying this image is bad, "au contraire", I'm amazed how much detail is preserved and how small the noise is (very pleasant) for iso 6400. But still, there is lots of noise - maybe similar to Iso 3200 (or 1600) from my K100D.

[tcdk]

raz: Just had a quick look at my K20D high iso shots and I do think the K5/6400 are comparable to K20D/800. The chroma noise is a bit more dominant on the K20D, while the K5 is a bit more luminance (which also cleans up easier).

Well, to my eyes - you don't have to agree

[rparmar]

Thanks Falk for your good work.

[rustynail925]

i'd like to see a higher resolution of that

[Col]

If the AF is also better, it needs to be really expensive to not end up under my xmas tree

I'm not going to waste any money on a dmnd Xmas tree Besides I think I'm convinced that its time to move the K20D into the reserves, before then.

[kevinschoenmakers]

Oh man, useable ISO25600, think of the possibilities!

[audiobomber]

raz: Just had a quick look at my K20D high iso shots and I do think the K5/6400 are comparable to K20D/800. The chroma noise is a bit more dominant on the K20D, while the K5 is a bit more luminance (which also cleans up easier).

Well, to my eyes - you don't have to agree

And I don't.

[alessandro63]

Aren't we exaggerating a little ?

Actually, yes. This is the 6400 image with two small crops of the K-7 at 1600 and 3200.
All RPP processed. The "grain" of ther K-5 at 6400 seems to be somewhere between the two. Which would be a very good result, IMHO.

[olenl]

Ok, so I wanted to see if new sensor is any better than K-x's.

I loaded the ISO 6400 sample into lr3 and applied 25 color reduction (default for any new images).

Then I loaded freshly made images at ISO 800, 1600, 3200 and 6400 (I turned off ANY NR turnable off in camera menu btw), in order to compare luminance noise (since color noise is completely removed by lr3 with no visible side effects - only luminance noise makes difference since when you reduce in, it blurs the picture which is really bad for IQ).

So my conclusions are as such ..
first off the K-5 noise looks "larger" but I guess it's because of the higher resolution (really strange though, maybe results of in-camera NR?), and perhaps because of JPEG compression.
I made conclusion that ISO 800 and 1600 or K-x are definitely better than presented image. ISO3200 is also seems a bit better , and 6400 same or little worse..

as the result, I am not sure if it's worth it to upgrade to K-5 from my K-x .. sure the extra wheel and other stuff are good but only when it falls to 900 dollars and below .. I don't really think they improved high ISO comparing to K-x much.

But still it might be better at resolution at high ISO, we'll see more tests.

also maybe it's better at lower ISO's (3200, 1600 and 800) than K-x.

P.S. this is just my opinion, I don't pretend to conduct serious tests or something. anyone who owns K-x is welcome to share own results too ..

[janneman]

Actually, yes. This is the 6400 image with two small crops of the K-7 at 1600 and 3200.
All RPP processed. The "grain" of ther K-5 at 6400 seems to be somewhere between the two. Which would be a very good result, IMHO.

As a trial i printed a K-7 3200 picture on A3+ paper. Not very good but hang it higher than eye level and no-one will notice. I am quite sure after seeing this that 3200 can now be printed on A3+ without too much problems ( if you can prevent people form looking at it with magnifying glass. )
However, Pixelpeepers will be bashing ths performance as they are likely to look only at 100%.

[tmtke]

I've made this for you, pixel peepers

I'm pretty much excited about this quality at ISO 6400. Wow.

Processing done with RawTherapee.
demosaicing: Amaze
sharpening: none (!)
chroma denoising: 50 (this removes color noise, not luminance noise, so most of the details are kept - just look at the rasters...)

I just don't do much pixel peeping, for I'm not printing ISO 3200+ shots, but this sensor does the same precious job like I saw on high-end Nikons (D3/D700 are really good examples ). And don't forget that they are full-frame, 12mp cameras...



fullsize: click here