K-5 outperforms Phase One P65+ and Nikon D3s in Dynamic Range...how is this possible?

[996GT2]

What kind of sensor trickery is going on here? How can an APS-C sensor beat both a full-frame sensor (D3s) and a medium format sensor (Phase One P65+) for dynamic range?

I'm pretty skeptical about these results. K-5 is beating out a P65+ for dynamic range when the latter is 1) Medium format and 2) using 16 bit processing?

http://dxomark.com/index.php/en/Camera-Sensor/Compare-sensors/%28appareil1%29/676|0/%28appareil2%29/579|0/%28appareil3%29/628|0/%28onglet%29/0/%28brand%29/Pentax/%28brand2%29/Phase%20One/%28brand3%29/Nikon

 

[arrfep]

That's really interesting. At first glance of the title I assumed someone was considering in-cam HDR as a true measure of dynamic range, but then checked the DxO link. The info appears to be corroborated by other reviews too. I would be interested in seeing what DPR has to say when their review is done. I can't say that I'm too surprised. After the High-ISO showing the K-x put on, improving DR seemed the next natural step. Imagine if this sensor was scaled up to FF but stayed at 16ish MP. Yowza. I've always wished Pentax was a bigger player and this just reinforces that.

Now if only I can get a full-frame Foveon sensor with K-5 DR and D3s high ISO in a D700 body with dual cards slots and compatible with EF lenses...I would never buy a camera again.

 

[ElFenix]

dxomark also has it being the nex 5, when it's the same sensor.

 

[996GT2]

dxomark also has it being the nex 5, when it's the same sensor.

NEX-5 is 14 megapixel
K-5 is 16 megapixel

Same sensor? I think not.

 

[ElFenix]

huh, i thought the nex 5 shared the sensor with the alpha 55, a580, and d7000. i guess not.

 

[IronWing]

I thought dynamic range is a function of individual sensor pixel design so the total number of pixels has no bearing.

 

[Cattykit]

I've been asking the same question. Pentax must have developed some crazy image processing algorithms. Check out real user samples below, the notion of "expose to the right" is just invalid for K5. Not only ISO 512,000! looks damn freaking good but you can also extract a high quality image from a severely underexposed raw file. And, by saying severely underexposed, I mean it's pitch black underexposed.

http://forums.dpreview.com/forums/read.asp?forum=1036&message=36806297
Dowload the file here: http://www.multiupload.com/4TQCIKGKQS

http://www.pentaxeros.com/forum/index.php?topic=37916
http://www.pentaxeros.com/images/IMGP9982.rar

http://forums.dpreview.com/forums/read.asp?forum=1036&message=36691159

 

[996GT2]

I've been asking the same question. Pentax must have developed some crazy image processing algorithms. Check out real user samples below, the notion of "expose to the right" is just invalid for K5. Not only ISO 512,000! looks damn freaking good but you can also extract a high quality image from a severely underexposed raw file. And, by saying severely underexposed, I mean it's pitch black underexposed.

http://forums.dpreview.com/forums/read.asp?forum=1036&message=36806297
Dowload the file here: http://www.multiupload.com/4TQCIKGKQS

http://www.pentaxeros.com/forum/index.php?topic=37916
http://www.pentaxeros.com/images/IMGP9982.rar

http://forums.dpreview.com/forums/read.asp?forum=1036&message=36691159

I just played with that RAW file in CR 6.0.

Amazing!

Original:

Edited:

 

[Cattykit]

At first, before playing with the RAW file myself, I did not believe what I was seeing. How can there be no banding or pattern noise in that situation? In that regard, my FF 5d mk2 can't even do half. 1/3, or 1/4 of what K5 is doing there.
It's crazy.

 

[996GT2]

At first, before playing with the RAW file myself, I did not believe what I was seeing.
It's crazy.

What they've managed to do is absolutely insane. When I opened that RAW file I was like, wtf...it's pitch black. Then I played around in camera raw and was totally amazed by the results.

 

[Gooberlx2]

Yeah, I was floored when people started posting super under-exposed RAWs on dpreview and pentaxforums. They increased to 14-bit from 12-bit RAW. That's one piece (of many I'm sure) that helps.

I'd like to see how well recovery works on over exposed images. Nonetheless, these sensors that Pentax is using seem pretty incredible. Hopefully it drums up business and/or better exposure for them, especially when combined with the massively improved AF and usual excellent build quality.

 

[extra]

Pretty cool, I saw that--definately one low read noise sensor. A sign of what's to come! Hopefully this will give pentax a bunch more sales--I'm a canon shooter and canon needs competition because they (and nikon) tend to stagnate on features when not pressed. The other nice thing about pentax is weather resistance. I've always loved Pentax and I hope this gives them the success they deserve. They have some awesome features (moveable sensor baby, maybe they'll implement tilt/shift that's sensor based someday, drool).

However, dxomark is still full of crap and has absolutely no bearing on how your real world image quality will be. The site should have close to zero credibility at this point. (there's some really lol-worthy overall ratings, go check it out if you haven't). Peeps are delusional (and will be disappointed) if they think this will give you landscape photo results in the real world that are better than a new mfdb or a new full frame cam. However, it's good to see that for once dxomark will be giving Pentax sales, instead of making people who would have considered them look to other brands because of a meaningless rating being lower.

 

[kalster]

wow, ridiculous DR . good on you pentax. time to upgrade my k10d to the k-5

 

[996GT2]

It looks like the Nikon D7000 performs similarly to the K5, which makes sense since both are using very similar (if not identical) Sony sensors.

http://dxomark.com/index.php/en/Cam...76|0/(onglet)/0/(brand)/Nikon/(brand2)/Pentax

 

[gar655]

Too bad (in the case of the K-5) the camera is a crap shoot. Pentax has poor QC.

I ordered one and 2 expensive Pentax lenses (55/1.4 and 50-135/2.8). The whole kit was crap. Either the camera was junk, the lenses were junk or all 3 were junk.

Couldn't focus, TONS of PF and CA with the 55/1.4, when the 50-135 managed to get somewhere near focus or as good as it could it was as dull as a putty knife, poor contrast.

The camera in "matrix" metering would blow highlights like crazy.

So the sensor may be pretty good or even really awesome, but when the rest of the system is junk, then who cares.

Sent it all back, got a D700, 28-300VR and 50/1.8- TEN times the camera!

Gene

 

[kalster]

Too bad (in the case of the K-5) the camera is a crap shoot. Pentax has poor QC.

I ordered one and 2 expensive Pentax lenses (55/1.4 and 50-135/2.8). The whole kit was crap. Either the camera was junk, the lenses were junk or all 3 were junk.

Couldn't focus, TONS of PF and CA with the 55/1.4, when the 50-135 managed to get somewhere near focus or as good as it could it was as dull as a putty knife, poor contrast.

The camera in "matrix" metering would blow highlights like crazy.

So the sensor may be pretty good or even really awesome, but when the rest of the system is junk, then who cares.

Sent it all back, got a D700, 28-300VR and 50/1.8- TEN times the camera!

Gene

really, thats very surprising. i have been using the k10d for over 3 years and have never had any QC issues. focus is not the strong point of the k10d but metering and picture quality is amazing. the build quality is great too, used a rebel xti prior to this and that felt like a toy compared to the pentax

 

[magomago]

are any of the differences significant at all according to how DXO benchmarks ?

IMO, DXOMark is junk science. I'm not sure how much it truly tells me about any real usage of the camera when producing pictures. But its a great PR tool for the company.

 

[Gooberlx2]

are any of the differences significant at all according to how DXO benchmarks ?

IMO, DXOMark is junk science. I'm not sure how much it truly tells me about any real usage of the camera when producing pictures. But its a great PR tool for the company.

In any case, you can download some RAWs yourself and play with them. It's pretty damn impressive.