noisepeepers strike again

deejjjaaaa · 07-10-2009, 02:14 PM

usual suspects : GordonBGood (today) and Oleg_V (yesterday)

Production K-7 Dynamic Range vs. K20D...: Pentax SLR Talk Forum: Digital Photography Review

K7 - preprocessed RAW files - noise is worse than seemed [Page 1]: Pentax SLR Talk Forum: Digital Photography Review

*Ash* · 07-10-2009, 03:06 PM

Said boldly "the K-7's image quality in this one respect is not up to the standards of the rest of the camera" then in the same breath "those who were happy with the K20D as to image quality will likely also be happy with the image quality from the K-7".

Gordon had no images to back his numerous comments of a similar vain up...

A lot of talk in Oleg's thread but not much substance - I'd take the posted K-7 'findings' with a grain of salt. Lots of bashing, not much real-life shooting happening...

GaryM · 07-10-2009, 03:44 PM

Originally Posted by Ash

Said boldly "the K-7's image quality in this one respect is not up to the standards of the rest of the camera" then in the same breath "those who were happy with the K20D as to image quality will likely also be happy with the image quality from the K-7".

Gordon had no images to back his numerous comments of a similar vain up...

A lot of talk in Oleg's thread but not much substance - I'd take the posted K-7 'findings' with a grain of salt. Lots of bashing, not much real-life shooting happening...

Gordon's a good guy, and the key phrase is "in this one respect". He was referring to dark shadow noise at low ISOs, and the inability to boost the shadows without introducing more noise than, for example, a Nikon D90. And by "the standards of the rest of the camera" he was praising the exceptional handling, build quality and speed of the K-7. His was a long ways from a Gloom And Doom post, and more of a desire for future improvements from Pentax sensors. Seems reasonable to me. Newsflash: the K-7 is not a perfect camera. Still seems like a pretty damn good body to me.

*Ash* · 07-10-2009, 03:51 PM

I guess it's just human nature... we will never be satisfied.
Granted, developments occur from consumer demand and discontentment, though the comparisons can be far too miniscule in detail, going well beyond what one can see in real life.

When we start splitting atoms, let alone splitting hairs, we can cause nuclear reactions...

kenyee · 07-10-2009, 05:05 PM

Ash: Gordon is one of the most technical people I've read and really knows his stuff from what I can see. He's not out to bash stuff, but he will try to analyze things. He's the first person who mentioned the chroma noise in the K20D's shadows and was correct (verified by Herb Chong on FM).

*pawzitiv* · 07-10-2009, 05:53 PM

GordonBGood is known to be an extremely technical person and his efforts to debunk all the subjective babble is commendable. Still, when a terms like "standard deviation" is being applied to a tool used for making photographic images, my eyes gloss over. I must have missed the moment in time when photography evolved to include statistical analyses.

There is much science and technology involved in the development and manufacture of a new digital camera but I feel there is very little of either involved in using that digital camera to best advantage.

I applaud Benjikan for his lack of participation in threads relating to K-7 vs. everything on the planet. Instead, he marches onward using his reported inferior equipment to create gorgeous images that impress even the toughest critics (magazine editors) and contributes to the forum by way of very informative tutorials about studio lighting.

Please, grab your nearest camera and go make some great images .. or learn something from your images that aren't so great.

rawr · 07-11-2009, 01:55 AM

Despite everything, 'noisepeeping' is useful. Hopefully it keeps the pressure on the Samsung/Pentax engineers to keep paying attention to this.

Video, live view etc are nice new engineering marvels in the K7, but continuously striving to improve camera image quality should remain job number 1 for Pentax DSLR engineers. This includes noise, dynamic range etc etc.

RMabo · 07-11-2009, 05:26 AM

Neither Oleg nor Gordon has translated their technical findings into real prints.
The big question is - does Oleg theory about RAW noise reduction killing resolution, and as thus having less image quality than 12Mp cameras, shows up when printing - if it shows up, at what sizes does it shows up compared to the 12Mp sensors?

The other big question is, if Gordons findings shows up when printing and if it does shows up - at what print sizes does it shows up?

My experience is that noise looks so much bigger when viewed on screen at 100% than when printed in the standard sizes most people uses for their family albums. Therefore, those postings are more academic theories than about real results when printing.

Mr Benjamin K has showned that the K20D has enough resolution for big poster printing. It is likely that the K-7 is as suitable as the K20D for those big prints.

It would be interresting to know if Mr Oleg and Mr Gordon makes prints as large as Benjamin does... If not, then why bother?

regken · 07-11-2009, 06:18 AM

Originally Posted by RMabo

It would be interresting to know if Mr Oleg and Mr Gordon makes prints as large as Benjamin does... If not, then why bother?

It's great we are blessed with both Ben and Gordon. I admire both for what they bring us. As you said Gordon's analysis shows us the sensor is very good with slight limitations which I am happy to know about.

Now where is the guru of AF so we can decide if it's worth upgrading from the K10D?

deejjjaaaa · 07-11-2009, 08:36 AM

good thing that at least RMabo stopped his nonsense about magic improvements that were supposed to appear out of nowhere in productions sensors vs hastily invented so-called "beta sensors" with beta firmware intended to tame the people wondering where is the promised increase in sensor iq in raw.

now he is back to his old tricks about printing... virtue out of necessity, when you can't beat it - hide it so that nobody sees it...

w/ a proper skill you can always find a way to put the different equipment into such conditions that you will see not any difference between two pieces...

Kidding

vinzer · 07-11-2009, 08:43 AM

Originally Posted by RMabo

It would be interresting to know if Mr Oleg and Mr Gordon makes prints as large as Benjamin does... If not, then why bother?

I'm actually thankful for the contributions of Gordon. He does it for free, he does it fairly, and he shares his result with the rest of us. Gordon is not assailing the K-7 wholly, guys, and I think that point needs repeating.

I'm an admirer of Ben's superb work, but why do we need to question the work done by Gordon (a technical guy who probably only does photography as a hobby) and compare it to Ben (an established pro)? There are many other ways to enjoy the photography, and it's not just by making prints. Some are more technically-inclined, like Gordon and Falk, and that's fine, too.

Do we always have to assail the photographic prowess of anyone who simply finds it interesting to make technical examinations of cameras? They are discovering information of a technical nature that camera makers would not admit to us, and they're doing us a favor for doing the things they do.

You can dismiss or shrug off their findings, but I don't see why the need to call them out if they make prints or not. Photography is so much more than final prints, especially in this digital age.

Nubi · 07-11-2009, 03:51 PM

I have had my k-7 for about a week now, and I have to tell you that the camera is starting to grow on me a bit, but the noise issue is..... well, it is an issue. The noise is not bad, but it is there, and just enough that, you know, we have to talk about it. I do feel that it is not a huge improvement from k20, not at all like the viewfinder. Pentax will have to address the issue, I think. I do think that the longer the camera has been in use, more noticeable in certain colors. I wonder if mag alloy makes the camera retain more heat than they anticipated.... I don't know.

*nostatic* · 07-11-2009, 04:43 PM

This is being discussed in the dslr forum here. I am firmly in the "let the noise happen" camp as I'd rather have a choice (and I think certain noise is a feature, not a bug). I don't believe in a free lunch. I did not expect better iq going from K20d to K7. No one should. I did expect more features and more keepers. And this has been proven true so far.

There will always be measurebators. And that's...ok. It has it's place. But it has little or nothing to do with taking pictures and/or making art. Different strokes. Seems to me that measurebators are much better off buying Canikon. If you want to actually shoot pictures, most anything will do the job.

*falconeye* · 07-11-2009, 07:01 PM

Originally Posted by Ash

Gordon had no images to back his numerous comments of a similar vain up...
A lot of talk in Oleg's thread but not much substance

-1 on Gordon
+1 on Oleg

Applying NR (at highest ISO) to RAW can be ok. It is neither good nor bad in the first place. Esp. binning the green channel looks like a good idea...

Originally Posted by GaryM

Gordon's a good guy

+1

Originally Posted by pawzitiv

I must have missed the moment in time when photography evolved to include statistical analyses.

Well ... the thread title contains the word noise and noise is a statistical phenomenon. If you want to understand what's going on, then this is the prize you'll have to pay.

On Gordon's findings ... he already published this for the pre-series sensor and now finds it confirmed for the series sensor. Which means that pre-series sensors have been very close to series already.

What Gordon studied is the read-out noise. This is one factor in overall noise, full well capacity being the other. However, APS-C CMOS sensors are all similiar in their full well capacity.

IMHO, finding that K-7 and K20D have the same noise performance is not that bad. There may be differences in banding and uniformity in darkest areas, though. This awaits to be studied with a series camera.

Nubi · 07-11-2009, 09:13 PM

Originally Posted by falconeye

-1 on Gordon
+1 on Oleg

Applying NR (at highest ISO) to RAW can be ok. It is neither good nor bad in the first place. Esp. binning the green channel looks like a good idea...

+1

Well ... the thread title contains the word noise and noise is a statistical phenomenon. If you want to understand what's going on, then this is the prize you'll have to pay.

On Gordon's findings ... he already published this for the pre-series sensor and now finds it confirmed for the series sensor. Which means that pre-series sensors have been very close to series already.

What Gordon studied is the read-out noise. This is one factor in overall noise, full well capacity being the other. However, APS-C CMOS sensors are all similiar in their full well capacity.

IMHO, finding that K-7 and K20D have the same noise performance is not that bad. There may be differences in banding and uniformity in darkest areas, though. This awaits to be studied with a series camera.

I agree with you that the noise is not bad, and that it is similar to that of k20. To me, it is a bit inconsistent from one shot to another. In some pics I can hardly notice, but in some instances I see it more than I expect it. I keep going back to the heat issue. Apparently the engineers struggled with the heat issue in regards to LV too. LV is such an improvement that I keep using it for an extended period of time, and for a couple of occasion, it shut off because of the heat, I presume.