[JohnBee]

There have been quite a few discussions with reference to the K-5's high ISO abilities lately, one of which I believed the K-5(and variants) would embark in FF territory. To test this theory, I took ISO6400 RAW samples from each camera(unprocessed) and extracted 200% from regions that I felt shared comparable tones.

Before, getting to the samples, I'd wanted to add that the K-5 required a + .75 Ev adjustment in ACR in order to effectively match the D700 shot. Therefore... there is good reason to conclude that the K-5 RAW would have lower SN than what is actually seen in these samples.*

K-5, ISO6400 (left) - D700 ISO6400 (middle) - A55 (right)



K-5, ISO6400 RAW


A-55, ISO6400 RAW


D700, ISO6400 RAW


Considering the K-5 RAW needed a push in exposure, I think it's safe to say that the D700 has finally met it's match.

---

Last edited by JohnBee; 09-28-2010 at 07:00 AM.

[Jyrkira]

Unfortunately these images tell nothing about noise/signal ratio (noise/detail). Real life photo would be more useful. I also suppose the K-5 noise images have been taken with 0.x firmware.

[JohnBee]

Unfortunately these images tell nothing about noise/signal ratio (noise/detail). Real life photo would be more useful. I also suppose the K-5 noise images have been taken with 0.x firmware.

That's true, unfortunately... we don't yet have controlled samples to work with as of yet.
This is nothing more than a noise level SN assessment at 200%.

Having said this, from what I've been able to see, the K-5 has more detail per pixel than the D700 at ISO6400. Mainly because of the increase in resolution and where: the K-5 has less low level NR than the D700 at this sensitivity.

I have since confirmed that the D700 does not apply low level NR at ISO6400!

So from my perspective, the K-5 effectively beats the D700 at ISO6400 in both noise and at the detail level of things.

The jury is still out on DR and color space. But my hope is that it will follow suite to the characteristics seen here.

[er1kksen]

Alright, so their noise characteristics at a certain grey level look similar, but I wonder which will have more usable detail in the shadows, for instance? That's the main difference for me between the K20D and K-x; my K20D looked good at base exposure at high enough ISOs, but the K-x gives me an actual manipulatable file at high ISOs, which is why I'm willing to use it at as much as two stops above where I'd shoot my K20D.

I'm not trying to imply the D700 will be better, as we clearly don't know yet. I'm just much more curious about that particular characteristic than this.

Also: you extracted 200% for these crops. Does that not mean that the K-5 crop is under greater magnification? Equal-area crops would be more relevant imo.

[janneman]

There have been quite a few discussions with reference to the K-5's high ISO abilities lately, one of which I believed the K-5(and variants) would embark in FF territory. To test this theory, I took ISO6400 RAW samples from each camera(unprocessed) and extracted 200% from regions that I felt shared comparable tones.

Ouch, some bad blotches on the K-5 image...

...

Oh.. Oops.... Geee......

....

Reminder to myself: Need to clean laptopscreen

[JohnBee]

Also: you extracted 200% for these crops. Does that not mean that the K-5 crop is under greater magnification? Equal-area crops would be more relevant imo.

Here are two 100% crops(no adjustements).
Note: Though both samples fall into different color regions, they were taken in very close tolerances within the histogram(almost exact).

FTR. the K-5 sample was taken in a dimly lit scene while the D700 crop was taken under studio lighting. With this in mind, I'd expect a slight improvement in favor the K-5. However, even at this level, I'd be more than happy to have such performances from an APS-C sensor.


RAW samples extracted at: 0,0,0,0,0 ACR values.


Considering the K-5 has a higher MP rating, I'd say it is showing excellent SNR against its FF counterpart

[robgo2]

There have been quite a few discussions with reference to the K-5's high ISO abilities lately, one of which I believed the K-5(and variants) would embark in FF territory. To test this theory, I took ISO6400 RAW samples from each camera(unprocessed) and extracted 200% from regions that I felt shared comparable tones.

Considering the K-5 RAW needed a push in exposure, I think it's safe to say that the D700 has finally met it's match.

No doubt, the succesor to the D700 will blow the K-5 out of the water, but it is still very impressive how far APS-C sensors have come in such a short time. I, for one, was doubtful that the K-5 would be a quantum leap above the K-7, but seeing is believing. Hopefully, when the final production version of the K-5 is released, we will not be disappointed. Meanwhile, I am feeling increasingly confident that my decision to resist the attraction of FF was a wise one. Why endure the added size, weight and expense when such excellent IQ is possible with an APS-C camera?

Rob

[er1kksen]

Here are two 100% crops(no adjustements).
Note: Though both samples fall into different color regions, they were taken in very close tolerances within the histogram(almost exact).

FTR. the K-5 sample was taken in a dimly lit scene while the D700 crop was taken under studio lighting. With this in mind, I'd expect a slight improvement in favor the K-5. However, even at this level, I'd be more than happy to have such performances from an APS-C sensor.


RAW samples extracted at: 0,0,0,0,0 ACR values.


Considering the K-5 has a higher MP rating, I'd say it is showing excellent SNR against its FF counterpart

Thanks, that's a little more helpful. The K-5 still appears to have a slightly wider noise "pattern" so to speak but the difference is very small.

[JohnBee]

The K-5 still appears to have a slightly wider noise "pattern" so to speak but the difference is very small.

I agree.
CHROMA noise seems very similar:



My guess is that the difference might be resolution dependent. Though on a SN scale(other than pitch), they seems nearly identical.

[Pentaxor]

There have been quite a few discussions with reference to the K-5's high ISO abilities lately, one of which I believed the K-5(and variants) would embark in FF territory. To test this theory, I took ISO6400 RAW samples from each camera(unprocessed) and extracted 200% from regions that I felt shared comparable tones.
K-5, ISO6400 (left) - D700 ISO6400 (right)



K-5, ISO6400 RAW


D700, ISO6400 RAW


Considering the K-5 RAW needed a push in exposure, I think it's safe to say that the D700 has finally met it's match.

thanks for your effort JB. it does look very promising, but it would be much more interesting if you were able to provide ISO comparisons at 12,800. that would be a great leap if the K-5 perform not good, but very good at that ISO level.

[Big G]

Johnbee, with all due respect, these shots tell us nothing.

Without seeing how detail, DR and colour are rendered with proper sample images taken in controlled conditions, then we can't tell how good the K-5 is. If you compare the D700 with the Sony a55 over on Imaging Resource at ISO 6400, the "blank" sections look pretty similar for noise levels, however the actual detail retained in the D700 samples are significantly better.

I will wait for lab tests .

[ogl]

There have been quite a few discussions with reference to the K-5's high ISO abilities lately, one of which I believed the K-5(and variants) would embark in FF territory. To test this theory, I took ISO6400 RAW samples from each camera(unprocessed) and extracted 200% from regions that I felt shared comparable tones.

K-5 has NR in RAW at ISO6400.

[JohnBee]

K-5 has NR in RAW at ISO6400.

I think that may be right actually...

I was pixel peeping some A55 RAW files last night and realized that those most likely didn't have NR at ISO6400. Mainly given the grain pitch difference between that camera and the K-5. However, for what it's worth, I think the NR applied at this level seems less than what we would find in a K-x ISO3200(for example).

The interesting things is... despite that, I think the K-5 manages to retains more detail at that sensitivity than the A55. However, we honestly won't know for sure until the K-5 is fully documented.

Another interesting tidbit with respect to the K-5 is where it really seems to have aprox. 1 stop over the A55(same sensor?). Which I thought was really interesting. I'm guessing(based on hearsay) that the issue there is with the A55's translucent membrane or something along those lines.

[JohnBee]

Johnbee, with all due respect, these shots tell us nothing.

Without seeing how detail, DR and colour are rendered with proper sample images taken in controlled conditions, then we can't tell how good the K-5 is. If you compare the D700 with the Sony a55 over on Imaging Resource at ISO 6400, the "blank" sections look pretty similar for noise levels, however the actual detail retained in the D700 samples are significantly better.

I will wait for lab tests .

That's an excellent idea(the A55 I meant )!
I'm going to add A55 samples to the tests and see how they compare.

PS. I know this is not controlled, but there's nothing wrong with trying to get a peek at what's to come right?

[JohnBee]

I have updated the thread with the A55.
Let me know what you think.