K-5, ISO51200 Ownage!

JohnBee · 10-21-2010, 12:59 PM - **1 Like**

Originally posted by telfish

I am still trying to work on the original image to get close to your results. I am closer but not really there. Here is a 100% crop from the original (tiff) worked in LR and Denoise 5.

Without looking, I'd look toward the base file.
Here's what mine looks like(with good reason):

Without explaining why(thus ruining the party).
That should get you well on your way with Denoise.

Last edited by JohnBee; 05-01-2011 at 11:02 AM.

JohnBee · 10-21-2010, 01:05 PM

And the other part:

Should do the trick.

JohnBee · 10-21-2010, 01:09 PM - **1 Like**

And of course, you'll have to resaturate the file afterward

telfish · 10-21-2010, 01:30 PM

Originally posted by JohnBee

And of course, you'll have to resaturate the file afterward

I am pretty nearly there. I am amazed what you can do with a little time on these files. Now I need to work on some of my own K5 files and see if I can get close to these results. Spot on exposure seems to make a huge difference, if you are under even slightly the noise increases dramatically.

Thanks for your time and patience.

Terry

Last edited by telfish; 10-21-2010 at 01:45 PM.

falconeye · 10-21-2010, 01:50 PM

I tried out a development snapshot from my own denoiser (color option is currently disabled) and started from the much heavier LR2 (no denoising at all) rather than pre-denoised LR3. This is what I got:

(click to enlarge)

I should mention that our denoiser currently has a single slider only which may go into advanced settings. The idea is to have automated denoising...

I don't think I already matched up to the Topaz Denoise result (although it would be interesting to see what Topaz does to the non-denoised LR2 output). But it doesn't look too bad compared to LR3 output in https://www.pentaxforums.com/forums/pentax-dslr-discussion/119157-last-one-no...o-51200-a.html. The darker tone came from the B&W conversion.

Big G · 10-21-2010, 02:06 PM

Oh - for the record - Topaz Denoise is absolutely fantastic, I highly recommend everyone picks up a copy of it regardless of what you shoot.

It even tidies up ISO 100 shots of blue skies very nicely when you add just a very tiny amount of noise reduction. I like to add a little grain using the grain slider (I notice Johnbee has added grain to his shots too

) which makes look terrifically non-digital and more akin to film.

Jeff Charles · 10-23-2010, 07:45 AM

Originally posted by Xatnep_CJ

Hey guys,

this time you played a little trick, I must say.
The RAW file is about one f-stop overexposed which makes it an ISO 25600. As we have learnt during the last days the K-5 produces high ISOs by pushing ISO 1600 underexposed images. So by overexposing an ISO 51200 by one f-stop the result is more or less the same an exposing an ISO 25600 correctly. This should also have an influence on the noise level.

I would be verry happy if you convince I am wrong.

It you are basing your conclusion that the shot was overexposed on the +1 EV exposure compensation, I have to disagree with you. Exposure compensation is used to modify the exposure that the meter chose, resulting in an adjustment to shutter speed or aperture, or both. If the resulting exposure, which is determined only by shutter speed and aperture, does not lose significant highlights, the photo is not overexposed. The photographer could just as well have used manual exposure to choose the same settings.

In high-ISO photos, the problem is not the high ISO. It is too little light. Increasing exposure to get as much light to the sensor as possible, i.e., exposing to the right, while avoiding blown highlights, is the right strategy, because it increases the signal-to-noise ratio.

dexmus · 10-23-2010, 12:02 PM

I found some more support for this great performance by K-5. Steve Bingham in dpreview forums compared D7000, D300 and D700. This is the summary:

Quote:

My findings: The D7000 really needs to be compared carefully with the D700. It pretty much whipped my D300 in every IQ catagory.
1- The D7000 has better resolution then the D700 (expected) Considerably better (not expected).
2- The noise level of the D3000 at 3200 was equal to the D700 WHEN BOTH FILES WERE MADE THE SAME SIZE (either by upresing the D700 or down rezing the D7000 in PS CS5. Comparing the two totally different file sizes at 200% gave a 1 stop edge to the D700. But we DON'T print that way. We print a certain size. The same scene printed at 16 x 24 inches would display equal noise from these two different cameras at ISO 3200. Wow. This was a shocker.
3- Dynamic range is more difficult to measure. First you need two files that have identical levels and curves. This is very difficult to do. Based on my attempting to "stretch" the DR in PS CS5, I would say the D7000 might well have an edge here also. (totally unexpected) This awaits real tests.

Seems even on Nikon side, this sensor is knocking over the big guns...

JohnBee · 10-23-2010, 12:50 PM

Originally posted by dexmus

I found some more support for this great performance by K-5. Steve Bingham in dpreview forums compared D7000, D300 and D700. This is the summary:

Seems even on Nikon side, this sensor is knocking over the big guns...

I wonder if anyone shyould tell them the K-5 has better IQ than the D7000?
Must be some serious trolling following this statement

Raylon · 10-23-2010, 12:53 PM

Originally posted by JohnBee

I wonder if anyone shyould tell them the K-5 has better IQ than the D7000?
Must be some serious trolling following this statement

Just a quick question. Because the 7000 and k5 have the same sensor, does that mean that the RAW files should be nearly identical? Or did Pentax do something to the sensor to modify it more than Nikon did. I assumed that the difference would be in how bodies themselves handled the image file after it was taken that made a difference between the K-5 and 7000, not the sensor. I am interested in seeing how different they actually are.

JohnBee · 10-23-2010, 12:58 PM

Originally posted by Raylon

Just a quick question. Because the 7000 and k5 have the same sensor, does that mean that the RAW files should be nearly identical? Or did Pentax do something to the sensor to modify it more than Nikon did. I assumed that the difference would be in how bodies themselves handled the image file after it was taken that made a difference between the K-5 and 7000, not the sensor. I am interested in seeing how different they actually are.

There is definitately a difference between the two at the grain and noise pattern level. Though I don't think it's apparent at all until sensitivities reach ISO6400 and above.

My suspicion is that manufacturers have their own unique way of handling data off the sensor(including noise reduction methods). Which may not seem like a big deal to some(claiming it is simple NR etc) but I think the results clearly show otherwise.

Raylon · 10-23-2010, 01:00 PM

Originally posted by JohnBee

There is definitately a difference between the two at the grain and noise pattern level. Though I don't think it's apparent at all until sensitivities reach ISO6400 and above.

My suspicion is that manufacturers have their own unique way of handling data off the sensor(including noise reduction methods). Which may not seem like a big deal to some(claiming it is simple NR etc) but I think the results clearly show otherwise.

Ok that's what I suspected also. I imagine a low ISO test between the two would have nearly identical results, but like it's been shown before, I doubt the Nikon will be able to match how Pentax handles high ISO's.

jani80k · 10-23-2010, 01:12 PM

I am really looking forward to this great camera.

Hide · 11-07-2010, 06:47 AM

(Please pardon me for such a low level question)

Does this example show the greatness of the software in NR or the greatness of K5?

What if it wasnt the K5 that did the shot?

(I'm a fan of pentax, don't shoot me =P )

dgaies · 11-07-2010, 06:58 AM

Originally posted by Hide

(Please pardon me for such a low level question)

Does this example show the greatness of the software in NR or the greatness of K5?

What if it wasnt the K5 that did the shot?

(I'm a fan of pentax, don't shoot me =P )

Mostly the "greatness" of the K-5, IMO. While NR software does a fantastic job of removing a lot of the noise present in the ISO 51200 file, the amount of detail that is there to begin with is quite impressive.

10-21-2010, 01:05 PM	#47
JohnBee Veteran Member Join Date: Sep 2008 Location: Newrfoundland Photos: Albums Posts: 4,667 Original Poster	And the other part: Should do the trick. Last edited by JohnBee; 05-01-2011 at 11:02 AM.

10-21-2010, 01:30 PM	#49
telfish Veteran Member Join Date: Aug 2009 Location: NY Posts: 407	Thanks again Originally posted by JohnBee And of course, you'll have to resaturate the file afterward I am pretty nearly there. I am amazed what you can do with a little time on these files. Now I need to work on some of my own K5 files and see if I can get close to these results. Spot on exposure seems to make a huge difference, if you are under even slightly the noise increases dramatically. Thanks for your time and patience. Terry Last edited by telfish; 10-21-2010 at 01:45 PM.

10-21-2010, 01:50 PM	#50
falconeye Veteran Member Join Date: Jan 2008 Location: Munich, Alps, Germany Photos: Gallery Posts: 6,871	I tried out a development snapshot from my own denoiser (color option is currently disabled) and started from the much heavier LR2 (no denoising at all) rather than pre-denoised LR3. This is what I got: (click to enlarge) I should mention that our denoiser currently has a single slider only which may go into advanced settings. The idea is to have automated denoising... I don't think I already matched up to the Topaz Denoise result (although it would be interesting to see what Topaz does to the non-denoised LR2 output). But it doesn't look too bad compared to LR3 output in https://www.pentaxforums.com/forums/pentax-dslr-discussion/119157-last-one-no...o-51200-a.html. The darker tone came from the B&W conversion. Last edited by falconeye; 10-21-2010 at 02:11 PM.

10-23-2010, 12:02 PM	#53
dexmus Veteran Member Join Date: Oct 2009 Location: United States Photos: Gallery \| Albums Posts: 392	some more support for this performance I found some more support for this great performance by K-5. Steve Bingham in dpreview forums compared D7000, D300 and D700. This is the summary: Quote: My findings: The D7000 really needs to be compared carefully with the D700. It pretty much whipped my D300 in every IQ catagory. 1- The D7000 has better resolution then the D700 (expected) Considerably better (not expected). 2- The noise level of the D3000 at 3200 was equal to the D700 WHEN BOTH FILES WERE MADE THE SAME SIZE (either by upresing the D700 or down rezing the D7000 in PS CS5. Comparing the two totally different file sizes at 200% gave a 1 stop edge to the D700. But we DON'T print that way. We print a certain size. The same scene printed at 16 x 24 inches would display equal noise from these two different cameras at ISO 3200. Wow. This was a shocker. 3- Dynamic range is more difficult to measure. First you need two files that have identical levels and curves. This is very difficult to do. Based on my attempting to "stretch" the DR in PS CS5, I would say the D7000 might well have an edge here also. (totally unexpected) This awaits real tests. Seems even on Nikon side, this sensor is knocking over the big guns...

10-21-2010, 12:59 PM - 1 Like	#46
JohnBee Veteran Member Join Date: Sep 2008 Location: Newrfoundland Photos: Albums Posts: 4,667 Original Poster	Originally posted by telfish I am still trying to work on the original image to get close to your results. I am closer but not really there. Here is a 100% crop from the original (tiff) worked in LR and Denoise 5. Without looking, I'd look toward the base file. Here's what mine looks like(with good reason): Without explaining why(thus ruining the party). That should get you well on your way with Denoise. Last edited by JohnBee; 05-01-2011 at 11:02 AM.

10-21-2010, 01:09 PM - 1 Like	#48
JohnBee Veteran Member Join Date: Sep 2008 Location: Newrfoundland Photos: Albums Posts: 4,667 Original Poster	And of course, you'll have to resaturate the file afterward

10-21-2010, 02:06 PM	#51
Big G Veteran Member Join Date: May 2009 Posts: 581	Oh - for the record - Topaz Denoise is absolutely fantastic, I highly recommend everyone picks up a copy of it regardless of what you shoot. It even tidies up ISO 100 shots of blue skies very nicely when you add just a very tiny amount of noise reduction. I like to add a little grain using the grain slider (I notice Johnbee has added grain to his shots too ) which makes look terrifically non-digital and more akin to film.

10-23-2010, 07:45 AM	#52
Jeff Charles Senior Member Join Date: Nov 2006 Location: Oaxaca, Mexico; Shohola PA, USA Posts: 219	Originally posted by Xatnep_CJ Hey guys, this time you played a little trick, I must say. The RAW file is about one f-stop overexposed which makes it an ISO 25600. As we have learnt during the last days the K-5 produces high ISOs by pushing ISO 1600 underexposed images. So by overexposing an ISO 51200 by one f-stop the result is more or less the same an exposing an ISO 25600 correctly. This should also have an influence on the noise level. I would be verry happy if you convince I am wrong. It you are basing your conclusion that the shot was overexposed on the +1 EV exposure compensation, I have to disagree with you. Exposure compensation is used to modify the exposure that the meter chose, resulting in an adjustment to shutter speed or aperture, or both. If the resulting exposure, which is determined only by shutter speed and aperture, does not lose significant highlights, the photo is not overexposed. The photographer could just as well have used manual exposure to choose the same settings. In high-ISO photos, the problem is not the high ISO. It is too little light. Increasing exposure to get as much light to the sensor as possible, i.e., exposing to the right, while avoiding blown highlights, is the right strategy, because it increases the signal-to-noise ratio.

10-23-2010, 12:50 PM	#54
JohnBee Veteran Member Join Date: Sep 2008 Location: Newrfoundland Photos: Albums Posts: 4,667 Original Poster	Originally posted by dexmus I found some more support for this great performance by K-5. Steve Bingham in dpreview forums compared D7000, D300 and D700. This is the summary: Seems even on Nikon side, this sensor is knocking over the big guns... I wonder if anyone shyould tell them the K-5 has better IQ than the D7000? Must be some serious trolling following this statement

10-23-2010, 12:53 PM	#55
Raylon Guest	Originally posted by JohnBee I wonder if anyone shyould tell them the K-5 has better IQ than the D7000? Must be some serious trolling following this statement Just a quick question. Because the 7000 and k5 have the same sensor, does that mean that the RAW files should be nearly identical? Or did Pentax do something to the sensor to modify it more than Nikon did. I assumed that the difference would be in how bodies themselves handled the image file after it was taken that made a difference between the K-5 and 7000, not the sensor. I am interested in seeing how different they actually are.

10-23-2010, 12:58 PM	#56
JohnBee Veteran Member Join Date: Sep 2008 Location: Newrfoundland Photos: Albums Posts: 4,667 Original Poster	Originally posted by Raylon Just a quick question. Because the 7000 and k5 have the same sensor, does that mean that the RAW files should be nearly identical? Or did Pentax do something to the sensor to modify it more than Nikon did. I assumed that the difference would be in how bodies themselves handled the image file after it was taken that made a difference between the K-5 and 7000, not the sensor. I am interested in seeing how different they actually are. There is definitately a difference between the two at the grain and noise pattern level. Though I don't think it's apparent at all until sensitivities reach ISO6400 and above. My suspicion is that manufacturers have their own unique way of handling data off the sensor(including noise reduction methods). Which may not seem like a big deal to some(claiming it is simple NR etc) but I think the results clearly show otherwise.

10-23-2010, 01:00 PM	#57
Raylon Guest	Originally posted by JohnBee There is definitately a difference between the two at the grain and noise pattern level. Though I don't think it's apparent at all until sensitivities reach ISO6400 and above. My suspicion is that manufacturers have their own unique way of handling data off the sensor(including noise reduction methods). Which may not seem like a big deal to some(claiming it is simple NR etc) but I think the results clearly show otherwise. Ok that's what I suspected also. I imagine a low ISO test between the two would have nearly identical results, but like it's been shown before, I doubt the Nikon will be able to match how Pentax handles high ISO's.

10-23-2010, 01:12 PM	#58
jani80k New Member Join Date: Mar 2010 Location: North Germany Posts: 21	I am really looking forward to this great camera.

11-07-2010, 06:47 AM	#59
Hide Forum Member Join Date: Apr 2010 Location: Hong Kong/Singapore/Melbourne Posts: 54	(Please pardon me for such a low level question) Does this example show the greatness of the software in NR or the greatness of K5? What if it wasnt the K5 that did the shot? (I'm a fan of pentax, don't shoot me =P )

Thread Tools	Search this Thread
	Search this Thread: Advanced Search

11-07-2010, 06:58 AM	#60
dgaies Veteran Member Join Date: Oct 2009 Location: Maryland / Washington DC Posts: 3,917	Originally posted by Hide (Please pardon me for such a low level question) Does this example show the greatness of the software in NR or the greatness of K5? What if it wasnt the K5 that did the shot? (I'm a fan of pentax, don't shoot me =P ) Mostly the "greatness" of the K-5, IMO. While NR software does a fantastic job of removing a lot of the noise present in the ISO 51200 file, the amount of detail that is there to begin with is quite impressive.