K-7 high ISO Performance RAW

Edmund · 07-01-2010, 07:41 AM

Originally posted by simico

On this picture which photo is from which camera?

(one of them is the K-7, of course)
Same one-click NR done on the 3 original photos, resized to same size.

The upper left seems to be the best quality. OK, so tell us which cameras took which pictures.

Marc Sabatella · 07-01-2010, 12:43 PM

Originally posted by shang

One thing I don't understand is that if d300 and k-7 get equally well high ISO RAW images, then why K-7 gets worse JPEGs?

Because better/worse is subjective. As I've said before. Some people prefer heavy NR that reduces noise but also smears away detail, others prefer lighter touch with NR that preserves detail but leaves noise around. Pentax generally does the latter, which many prefer, but those who don't will cal the results "worse".

shang · 07-01-2010, 02:24 PM

Originally posted by Marc Sabatella

Because better/worse is subjective. As I've said before. Some people prefer heavy NR that reduces noise but also smears away detail, others prefer lighter touch with NR that preserves detail but leaves noise around. Pentax generally does the latter, which many prefer, but those who don't will cal the results "worse".

thanks a lot, Marc!

generally speaking, k-x has better high ISO results(let's suppose we define "good" as less noise). so if pentax replaces the k-7 RAW-->JPEG algorithm by the one used in K-x, then the k-7 would get better results, is this true?

CWyatt · 07-01-2010, 03:00 PM

Originally posted by shang

thanks a lot, Marc!

generally speaking, k-x has better high ISO results(let's suppose we define "good" as less noise). so if pentax replaces the k-7 RAW-->JPEG algorithm by the one used in K-x, then the k-7 would get better results, is this true?

I don't think so, it depends on the sensor too. K-x has a newer, slightly improved sensor.

Marc Sabatella · 07-01-2010, 03:09 PM

Originally posted by shang

generally speaking, k-x has better high ISO results(let's suppose we define "good" as less noise). so if pentax replaces the k-7 RAW-->JPEG algorithm by the one used in K-x, then the k-7 would get better results, is this true?

It's probably true that if this were technically possible (the algorithm used on the K-x might only work for that sensor), then same people who like that type of NR processing will like it better on the K-7. It's also probably true that the people who prefer the type of NR the K-7 does will think using the K-x algorithm would be worse. As I said tried to say clearly in my previous post, preference in NR is *subjective*. That means two different people might look at the same two images and one might prefer image A and the other might prefer mage B. That's why it's often impossible to talk in general terms about "better" and "worse".i

**Rondec** · 07-01-2010, 06:03 PM

The kx is really nice, but I have to say that the jpeg engine doesn't deal well with red colors at high iso. Turns it into a smeary mess. However, seems like you can do a better job with RAW at fixing that. Just saying that RAW can make a difference even with the kx sensor.

kittykat46 · 07-01-2010, 06:33 PM

Originally posted by Rondec

The kx is really nice, but I have to say that the jpeg engine doesn't deal well with red colors at high iso. Turns it into a smeary mess. However, seems like you can do a better job with RAW at fixing that. Just saying that RAW can make a difference even with the kx sensor.

That's a weakness common to the K-7 and my earlier K100D as well. Even in RAW to a certain extent, the interpretation of strongly saturated Reds seems to be behind the competition, especially at high ISO.
Shoot at a scene with someone wearing Bright Scarlet clothing, or a Bright Red flower, and you'll notice the loss of detail.
In the actual case that I had, the red cloth had fine micro-detail patterns and texture, but the sensor simply failed to capture much of that, even with a DA Limited lens.

Anyway, the K-7's RAW output is really very good in general, just some weak points here and there.

m8o · 07-01-2010, 06:56 PM

Originally posted by simico

On this picture which photo is from which camera?

(one of them is the K-7, of course)
Same one-click NR done on the 3 original photos, resized to same size.

I don't know but I want the camera on the bottom if it's the same ISO as the two above it. I'd guess the top right is the K-7.
(what does "one-click NR" mean?)

simico · 07-02-2010, 01:37 AM

Originally posted by Edmund

The upper left seems to be the best quality. OK, so tell us which cameras took which pictures.

Originally posted by m8o

I don't know but I want the camera on the bottom if it's the same ISO as the two above it. I'd guess the top right is the K-7.
(what does "one-click NR" mean?)

Top left is K-m, top right is K-7, bottom is K-x. ISO1600 studio test RAW images from DPreview. "One-click NR" is just that, with one mouse-click I enabled the built-in Noise Ninja (using the same default strength and smooth setting for all three images).
Did similar comparisons with Canon 7D and Nikon D300 raw files. My conclusion: even with just a lazy default NR the differences aren't really that much even at 100% pixel peeping, with a quick 1-2 minute NR tweaking differences can become really small. So I, personally, wouldn't worry that much about noise anymore. As always, YMMV

kari · 07-02-2010, 02:21 AM

Originally posted by simico

Top left is K-m, top right is K-7, bottom is K-x. ISO1600 studio test RAW images from DPreview. "One-click NR" is just that, with one mouse-click I enabled the built-in Noise Ninja (using the same default strength and smooth setting for all three images).
Did similar comparisons with Canon 7D and Nikon D300 raw files. My conclusion: even with just a lazy default NR the differences aren't really that much even at 100% pixel peeping, with a quick 1-2 minute NR tweaking differences can become really small. So I, personally, wouldn't worry that much about noise anymore. As always, YMMV

Wow, K-m looks best to me!

m8o · 07-02-2010, 07:29 AM

Originally posted by simico

...My conclusion: even with just a lazy default NR the differences aren't really that much even at 100% pixel peeping, with a quick 1-2 minute NR tweaking differences can become really small. So I, personally, wouldn't worry that much about noise anymore. As always, YMMV

Szia Simico, I can't necessarily agree completely. I looked at the photos and selected the one that had the most distasteful noise signature, and chose the K-7 correctly. After spending years of time fighting noise, often without success, there's something to be said about starting with the best input one can start with. If the replacement to the K-7 doesn't have a noise signature like the K-x, I think I may have to get the K-x as my next purchase, which I've been delaying for over a year.

Marc Sabatella · 07-02-2010, 11:35 AM

Originally posted by kari

Wow, K-m looks best to me!

Me too, although that doesn't really surprise me - I had pixel peeped similar samples (the still life from imaging resource, and the bottles in particular) before and come to the same conclusion. I looked at the K200D instead of the K-m, but they have the same basic sensor.

Not to say I think either of these cameras outperform the the K-x at high ISO *in general*, and there could simply be a focus difference here, but this particular K-m image retains detail best to my eye, with the other two roughly similar in that regard. The K-x image definitely shows less noise for the same (lack of) detail than the K-7 image does. The K-x also shows less noise than the K-m, but appears to pay for it with less detail. Which kind of surprises me, because supposedly the in-camera RAW NR on the K-x doesn't kick in until ISO 3200, whereas it's already present on the K-m at ISO 1600. That's why I suspect a slight difference in focus point (or lens used here may be explaining some of the difference. But it could also have to do with the specific textures involved here being ones the 10MP sensor deals with better; skin textures seem to be a particular strength of the K-x sensor in this respect.

Anyhow, I do think the basic point of the comparison is made - the differences are small enough that it takes a 100% view to see differences, and even then people might disagree on which "wins" (although I suspect few would pick the K-7). However, the same comparison performed at ISO 3200 or 6400 may show a clearer advantage to the K-x.

soccerjoe5 · 07-04-2010, 11:38 AM

Originally posted by Urkeldaedalus

Hi Diego, nice to see you back on the board. What JPEG settings do you usually use?

Hey buddy, it's been a while huh? Hmm.. it's set to "Bright", the High-ISO NR to Low. I guess I didn't tweak it that much!

I shoot with it on Auto-ISO up to ISO2200 regularly. As long as I expose properly, I'm not bothered with the noise.

RonHendriks1966 · 07-04-2010, 12:27 PM

I use the K-7 on several occasions up to iso3200 (even iso5000 is usable but iso6400 is just for emergency's) and with Noise Ninja cleaning them up afer the RAW developed in PSE. Works very good to me.

Raw at iso2500

Even jpeg direct from the K-7 in black and white at iso3200 are great:

07-02-2010, 11:35 AM	#27
Marc Sabatella Veteran Member Join Date: Sep 2006 Location: Denver, CO Photos: Gallery Posts: 10,685	Originally posted by kari Wow, K-m looks best to me! Me too, although that doesn't really surprise me - I had pixel peeped similar samples (the still life from imaging resource, and the bottles in particular) before and come to the same conclusion. I looked at the K200D instead of the K-m, but they have the same basic sensor. Not to say I think either of these cameras outperform the the K-x at high ISO in general, and there could simply be a focus difference here, but this particular K-m image retains detail best to my eye, with the other two roughly similar in that regard. The K-x image definitely shows less noise for the same (lack of) detail than the K-7 image does. The K-x also shows less noise than the K-m, but appears to pay for it with less detail. Which kind of surprises me, because supposedly the in-camera RAW NR on the K-x doesn't kick in until ISO 3200, whereas it's already present on the K-m at ISO 1600. That's why I suspect a slight difference in focus point (or lens used here may be explaining some of the difference. But it could also have to do with the specific textures involved here being ones the 10MP sensor deals with better; skin textures seem to be a particular strength of the K-x sensor in this respect. Anyhow, I do think the basic point of the comparison is made - the differences are small enough that it takes a 100% view to see differences, and even then people might disagree on which "wins" (although I suspect few would pick the K-7). However, the same comparison performed at ISO 3200 or 6400 may show a clearer advantage to the K-x. Last edited by Marc Sabatella; 07-02-2010 at 11:40 AM.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Boosting High ISO Performance	Jewelltrail	Pentax DSLR Discussion	30	10-10-2010 10:26 PM
K-7 HIGH ISO NR In RAW	Christopher M.W.T	Pentax DSLR Discussion	2	06-06-2010 05:19 PM
K20D High ISO Performance	joelovotti	Pentax DSLR Discussion	22	03-17-2009 06:47 PM
K200D low light/high iso performance	indytax	Pentax DSLR Discussion	22	02-07-2009 07:49 PM
K20D raw files at high ISO	Martynas	Pentax News and Rumors	21	02-01-2008 05:45 PM

07-01-2010, 07:41 AM	#16
Edmund Forum Member Join Date: May 2010 Location: Kentucky Posts: 68	Originally posted by simico On this picture which photo is from which camera? (one of them is the K-7, of course) Same one-click NR done on the 3 original photos, resized to same size. The upper left seems to be the best quality. OK, so tell us which cameras took which pictures.

07-01-2010, 12:43 PM	#17
Marc Sabatella Veteran Member Join Date: Sep 2006 Location: Denver, CO Photos: Gallery Posts: 10,685	Originally posted by shang One thing I don't understand is that if d300 and k-7 get equally well high ISO RAW images, then why K-7 gets worse JPEGs? Because better/worse is subjective. As I've said before. Some people prefer heavy NR that reduces noise but also smears away detail, others prefer lighter touch with NR that preserves detail but leaves noise around. Pentax generally does the latter, which many prefer, but those who don't will cal the results "worse".

07-01-2010, 02:24 PM	#18
shang Senior Member Join Date: May 2010 Location: long island Posts: 135 Original Poster	Originally posted by Marc Sabatella Because better/worse is subjective. As I've said before. Some people prefer heavy NR that reduces noise but also smears away detail, others prefer lighter touch with NR that preserves detail but leaves noise around. Pentax generally does the latter, which many prefer, but those who don't will cal the results "worse". thanks a lot, Marc! generally speaking, k-x has better high ISO results(let's suppose we define "good" as less noise). so if pentax replaces the k-7 RAW-->JPEG algorithm by the one used in K-x, then the k-7 would get better results, is this true?

07-01-2010, 03:00 PM	#19
CWyatt Veteran Member Join Date: Jul 2009 Posts: 1,291	Originally posted by shang thanks a lot, Marc! generally speaking, k-x has better high ISO results(let's suppose we define "good" as less noise). so if pentax replaces the k-7 RAW-->JPEG algorithm by the one used in K-x, then the k-7 would get better results, is this true? I don't think so, it depends on the sensor too. K-x has a newer, slightly improved sensor.

07-01-2010, 03:09 PM	#20
Marc Sabatella Veteran Member Join Date: Sep 2006 Location: Denver, CO Photos: Gallery Posts: 10,685	Originally posted by shang generally speaking, k-x has better high ISO results(let's suppose we define "good" as less noise). so if pentax replaces the k-7 RAW-->JPEG algorithm by the one used in K-x, then the k-7 would get better results, is this true? It's probably true that if this were technically possible (the algorithm used on the K-x might only work for that sensor), then same people who like that type of NR processing will like it better on the K-7. It's also probably true that the people who prefer the type of NR the K-7 does will think using the K-x algorithm would be worse. As I said tried to say clearly in my previous post, preference in NR is subjective. That means two different people might look at the same two images and one might prefer image A and the other might prefer mage B. That's why it's often impossible to talk in general terms about "better" and "worse".i

07-01-2010, 06:03 PM	#21
Rondec Loyal Site Supporter Join Date: Mar 2009 Location: Gladys, Virginia Photos: Gallery Posts: 27,653	The kx is really nice, but I have to say that the jpeg engine doesn't deal well with red colors at high iso. Turns it into a smeary mess. However, seems like you can do a better job with RAW at fixing that. Just saying that RAW can make a difference even with the kx sensor.

07-01-2010, 06:33 PM	#22
kittykat46 Veteran Member Join Date: Dec 2007 Posts: 923	Originally posted by Rondec The kx is really nice, but I have to say that the jpeg engine doesn't deal well with red colors at high iso. Turns it into a smeary mess. However, seems like you can do a better job with RAW at fixing that. Just saying that RAW can make a difference even with the kx sensor. That's a weakness common to the K-7 and my earlier K100D as well. Even in RAW to a certain extent, the interpretation of strongly saturated Reds seems to be behind the competition, especially at high ISO. Shoot at a scene with someone wearing Bright Scarlet clothing, or a Bright Red flower, and you'll notice the loss of detail. In the actual case that I had, the red cloth had fine micro-detail patterns and texture, but the sensor simply failed to capture much of that, even with a DA Limited lens. Anyway, the K-7's RAW output is really very good in general, just some weak points here and there.

07-01-2010, 06:56 PM	#23
m8o Veteran Member Join Date: Jul 2007 Location: 40°-55'-44" N / 73°-24'-07" W [on LI] Photos: Gallery \| Albums Posts: 3,092	Originally posted by simico On this picture which photo is from which camera? (one of them is the K-7, of course) Same one-click NR done on the 3 original photos, resized to same size. I don't know but I want the camera on the bottom if it's the same ISO as the two above it. I'd guess the top right is the K-7. (what does "one-click NR" mean?)

07-02-2010, 01:37 AM	#24
simico Veteran Member Join Date: Jan 2009 Location: Budapest Posts: 821	Originally posted by Edmund The upper left seems to be the best quality. OK, so tell us which cameras took which pictures. Originally posted by m8o I don't know but I want the camera on the bottom if it's the same ISO as the two above it. I'd guess the top right is the K-7. (what does "one-click NR" mean?) Top left is K-m, top right is K-7, bottom is K-x. ISO1600 studio test RAW images from DPreview. "One-click NR" is just that, with one mouse-click I enabled the built-in Noise Ninja (using the same default strength and smooth setting for all three images). Did similar comparisons with Canon 7D and Nikon D300 raw files. My conclusion: even with just a lazy default NR the differences aren't really that much even at 100% pixel peeping, with a quick 1-2 minute NR tweaking differences can become really small. So I, personally, wouldn't worry that much about noise anymore. As always, YMMV

07-02-2010, 02:21 AM	#25
kari Senior Member Join Date: Feb 2010 Photos: Albums Posts: 142	Originally posted by simico Top left is K-m, top right is K-7, bottom is K-x. ISO1600 studio test RAW images from DPreview. "One-click NR" is just that, with one mouse-click I enabled the built-in Noise Ninja (using the same default strength and smooth setting for all three images). Did similar comparisons with Canon 7D and Nikon D300 raw files. My conclusion: even with just a lazy default NR the differences aren't really that much even at 100% pixel peeping, with a quick 1-2 minute NR tweaking differences can become really small. So I, personally, wouldn't worry that much about noise anymore. As always, YMMV Wow, K-m looks best to me!

07-02-2010, 07:29 AM	#26
m8o Veteran Member Join Date: Jul 2007 Location: 40°-55'-44" N / 73°-24'-07" W [on LI] Photos: Gallery \| Albums Posts: 3,092	Originally posted by simico ...My conclusion: even with just a lazy default NR the differences aren't really that much even at 100% pixel peeping, with a quick 1-2 minute NR tweaking differences can become really small. So I, personally, wouldn't worry that much about noise anymore. As always, YMMV Szia Simico, I can't necessarily agree completely. I looked at the photos and selected the one that had the most distasteful noise signature, and chose the K-7 correctly. After spending years of time fighting noise, often without success, there's something to be said about starting with the best input one can start with. If the replacement to the K-7 doesn't have a noise signature like the K-x, I think I may have to get the K-x as my next purchase, which I've been delaying for over a year.

07-04-2010, 11:38 AM	#28
soccerjoe5 Veteran Member Join Date: May 2008 Location: Philippines Photos: Gallery Posts: 1,343	Originally posted by Urkeldaedalus Hi Diego, nice to see you back on the board. What JPEG settings do you usually use? Hey buddy, it's been a while huh? Hmm.. it's set to "Bright", the High-ISO NR to Low. I guess I didn't tweak it that much! I shoot with it on Auto-ISO up to ISO2200 regularly. As long as I expose properly, I'm not bothered with the noise.

07-04-2010, 12:27 PM	#29
RonHendriks1966 Banned Join Date: Jan 2009 Photos: Gallery Posts: 9,675	I use the K-7 on several occasions up to iso3200 (even iso5000 is usable but iso6400 is just for emergency's) and with Noise Ninja cleaning them up afer the RAW developed in PSE. Works very good to me. Raw at iso2500 Even jpeg direct from the K-7 in black and white at iso3200 are great: