[rawr]

Just in case anyone missed it, after studying the K-x's RAW files, dpreview forum's pentax guru GordonBGood says:

4) This is the very best high ISO performance of any camera Pentax has produced to date, matching or exceeding any of the Nikon models including the D300 and D300s. This was never really a problem with even the K20D and K-7 in that they likely match most competitors high ISO performance with just a little touch of Noise Reduction (NR), but this K-x appears to match or exceed other competitors without any NR or with very very little, just as those competitors may apply.

Plus

With the K-x and its new source of CMOS sensor, that [poor dynamic range of the K20D and K7] problem has gone away, with the K-x having about a 12 to 13 stops of DR at base ISO of ISO 200 as per the way DR is measured by DxOMark, and still has a 10 stop DR at ISO 1600 and a 8 stop DR even at ISO 12800..

Re: RawTherapee shots are up: Pentax SLR Talk Forum: Digital Photography Review

That's good news. Very good news to have the wide high ISO DR confirmed too.

[Ash]

Wow. Even more impressive...

Beating the D300 in DR?!?
Brilliant.

A FF sensor emulated from K-x's would just be a charm!

[cbaytan]

Yeah but isn't a high iso photo harder to "fix-up" in photoshop than a low iso photo. Besides I think the K-x has expanded iso range down to 100.

Now can K-x can take photos at 100 ISO on hardware level or not? Or is it just a software emulation.

[rawr]

Now can K-x can take photos at 100 ISO on hardware level or not? Or is it just a software emulation.

You have to enable expanded sensitivity in the menu to get it down to 100 ISO from it's default floor of 200 ISO, so I'd guess the 'hardware level' ISO is 200.

[cbaytan]

You have to enable expanded sensitivity in the menu to get it down to 100 ISO from it's default floor of 200 ISO, so I'd guess the 'hardware level' ISO is 200.

Suppose we are both right that raises another question, if hardware level is 200 ISO minimum and some software gobbledygook makes it 100, what does it do to picture makes it look like 100 ISO then?

[rawr]

I'm not sure how it is achieved, and what it's real meaning is. Maybe some experts will know.

What I can say is that the distinctions between 'hardware' and 'software' in complex systems like cameras are becoming harder and harder to distinguish, since the 'hardware' needs software to do anything, including microcode running within the chip level hardware itself, including the sensor circuitry, let alone software running external to the core CPU and DSP hardware, like any user loadable camera firmware. There is no simple 'bare metal' hardware doing anything by itself anymore in most cameras.

[falconeye]

Suppose we are both right that raises another question, if hardware level is 200 ISO minimum and some software gobbledygook makes it 100, what does it do to picture makes it look like 100 ISO then?

If you read my DxO style lab test I just published for the K-x, you'll see that the gray level noise at ISO 200 and ISO 100 are about the same. So, native ISO is 200.

But at ISO 100 some "magic" kicks in and this results in less noisy shadows, below about 1% luminosity. I don't know how this is achieved. But the resulting dynamic range at ISO 100 is huge!

[cbaytan]

If you read my DxO style lab test I just published for the K-x, you'll see that the gray level noise at ISO 200 and ISO 100 are about the same. So, native ISO is 200.

But at ISO 100 some "magic" kicks in and this results in less noisy shadows, below about 1% luminosity. I don't know how this is achieved. But the resulting dynamic range at ISO 100 is huge!

Yes I did read, actually your article raised the question, I knew K-x as only ISO 100, nothing else.

But that raises another question, if a firmware can raise the DR tremendously, why Pentax don't do it the same thing say on K7? Like ISO 50 or so? I guess we have lots'of ambiguity here.

[falconeye]

Yes I did read, actually your article raised the question, I knew K-x as only ISO 100, nothing else.

But that raises another question, if a firmware can raise the DR tremendously, why Pentax don't do it the same thing say on K7? Like ISO 50 or so? I guess we have lots'of ambiguity here.

No, you got me wrong.

The huge dynamic range is from the sensor, first of all.

The possible bump making true ISO 100 in the shadows maybe adds another 0.5 EV to the DR. But even without the DR is excellent. I actually added a detail section discussing the bump to my blog article.

If the bump isn't just a sign of poor measurement from my side then it could be applied to native ISO 100 sensors as well. But this isn't a firmware trick. It would be something in the sensor hardware.


The question is how to provide lower than native ISO at all. Naively, this should result in blown highlights where the sensor cells for bright light flow over. Because a sensor with a higher native ISO has a smaller full well capacity. One way to really provide lower ISO would be to create less electrons per photon. But I have no idea how a cell could adaptively change this.

If anybody has an idea, please come over to my lumolabs thread to discuss it.

[cbaytan]

The huge dynamic range is from the sensor, first of all.

K-x DR 13.3 stops? A digital sensor finally reached to a T-MAX 400 DR ?

Thats great news, now the sharpness issue, wish K-x or K7 are sharper cameras. I wish there were Hot-Rod version of the both cameras.

The link for who doesn't know hot rod modification yet.:

Hot Rod Visible

[falconeye]

Thats great news, now the sharpness issue, wish K-x or K7 are sharper cameras. I wish there were Hot-Rod version of the both cameras.
The link for who doesn't know hot rod modification yet.:
Hot Rod Visible

You are a funny guy ...

Go to my article again and look at my resolution analysis for K-x (really, open the test charts in "Original" size on SmugMug to see; I pay their service for a reason ...). Then look up the zone plate chart, near Nyquist. Then click your own link and inspect "NikonD200HR Charts.jpg", 3rd row, with the star plot. You see the same ugly color fringing and moiré artifacts coming from a too weak or missing AA filter.

A would even say that my test doesn't rule out a possibility the K-x has no AA filter at all ... Did anybody check it out already?

And still you complain about a lack of sharpness? How about a decent lens and proper focussing technique?

You are a funny guy ...
Really.
Personally, I prefer sharpening headroom in post processing over color and fringing artifacts.

[cbaytan]

You are a funny guy ...

Personally, I prefer sharpening headroom in post processing over color and fringing artifacts.

LOL, I am being told being one of a kind and funny, but you tell me about being funny after suggesting to me a good lens and focusing technique, and especially once after asking me out of one threads just because I don't own a K-7. We really enjoy your technical knowledge but now please don't "proper focus" me. And please keep up with the meticulous technical work.

You gotta understand that sharpness will be never enough, at least for 5-10 years from now on. Comparing only three Pentax models does not mean much when sensor technology is out there continuously developing. and sharpening with software is whole different story.

I'll post the link somehow showing k-7 is softer than K20D as soon as I re-find it, there was a colorful toy-cow and some books in those testing pictures. Basically (simply speaking) moire and sharpness are the both ends of the problem and I believe K10D's huge success came from weak AA filter. I believe that so strong maybe without K10D's huge success with weak AA filter, most of us wouldn't be here as Pentaxians. I mean K10D's success made people return to Pentax or converted to Pentax from Canikon and such. was it about 5.5 million K10D units sold? BTW apart from that sample; only recently I am convinced that sharpness between K20 and K-7 is not much different. But I prefer to stay paranoid and pixel peeping about the sharpness issue.

[Aristophanes]

If you read my DxO style lab test I just published for the K-x, you'll see that the gray level noise at ISO 200 and ISO 100 are about the same. So, native ISO is 200.

But at ISO 100 some "magic" kicks in and this results in less noisy shadows, below about 1% luminosity. I don't know how this is achieved. But the resulting dynamic range at ISO 100 is huge!

So why have the default for the K-x set at ISO 200? With such a DR advantage at ISO 100, why not just set it there and be done with it?

There must some noticeable downside to this "magic".

[falconeye]

There must some noticeable downside to this "magic".

I found ISO 100 to clip at about 80% linear luminosity, compared to where ISO 200 clips. So, ISO 100 is more like ISO 125 and should be used with -1/3 EV compensation.

Moreover, I am not sure if the tone response curve is as linear as at ISO 200. I didn't check it though. Most people will be just happy with the DR at ISO 200 which is very good as well.

[falconeye]

after asking me out of one threads just because I don't own a K-7.
[...]
You gotta understand that sharpness will be never enough
[...]
BTW apart from that sample; only recently I am convinced that sharpness between K20 and K-7 is not much different. But I prefer to stay paranoid and pixel peeping about the sharpness issue.

@cbaytan, please don't take personal offense. Not intended by my part. This other thread was run by owners all having the same issue in an attempt to help each other out. You had been the only non owner joining the thread. So, please don't be too surprised by a reaction similiar to a non drinker joining an Alcoholics Anonymous group

Back on topic.

I share your passion for resolution. This is why I was surprised by your comment. My question about lens and focussing applied to tests you were referring to. It wasn't targeted at you as you actually didn't contribute test shots.

But even w/o contributing test shots, you can help. In my blog, I posted resolution chart test shots for K-x (and K-7). The K-x looks suspiciously similiar to your Hot Rod test shots without an AA filter. Maybe, you can have a look at my Original size SmugMug version and provide your own interpretation of the results.

In particular, do you think that the K-x actually IS a Hot rod Pentax w/o an AA filter? And if so, how then can't it be sharp enough?

BTW, K20D and K-7 have equal resolution, according to my findings.