K-1 Full res JPGs versus D810 ISO 12800, ISO 51200

KDAFA · 03-19-2016, 08:20 AM

Originally posted by Giklab

The K-1 samples are obviously OOC JPEGs at default settings.
What's also painfully obvious is that the NR algorithm hasn't changed much from the K-5...

The OP stated no noise reduction ....?

Last edited by KDAFA; 03-19-2016 at 08:56 AM.

mee · 03-19-2016, 08:23 AM

Originally posted by Bromberger

Looks good.

Weird though: The 810 has a lot of colour noise, while the K1 has none. And colour noise can be corrected quite easily. It's almost as someone ran colour noise reduction over only one of the two pictures.

Also, what's with the large horizontal lines (over the bottles, or the yellow area around the clock) in the 51k-iso-picture of the K1? It looks almost as if the scene was lit through window blinds.

Not trying to bash the K1, i'm confident it's gonna be an awesome camera.

Honesty, online, sometimes gets misconstrued as bashing sadly. But, for one, I agree with you -- the K-1 high ISO shots are much more mushy than the comparable D810 shots. And there is less contrast. That said, looking at the EXIF data, its still a beta firmware so perhaps this can be improved upon before release? PENTAX K-1 Ver. 0.38

**jpzk** · 03-19-2016, 08:54 AM

Originally posted by ogl

K-1 is much better, but blurred. Nikon is sharp.

How can K-1 is "better" if "blurred" ?

That is what I call an oxymoron
https://en.wikipedia.org/wiki/Oxymoron

JP

ogl · 03-19-2016, 09:08 AM

Originally posted by jpzk

How can K-1 is "better" if "blurred" ?

That is what I call an oxymoron
https://en.wikipedia.org/wiki/Oxymoron

JP

Easy. It's two different sense.
Better is terms of noise. Blurred as picture.

Giklab · 03-19-2016, 09:59 AM

Originally posted by micromacro

Forgive my ignorance, but what exactly wrong with NR algorithm? Nikon algorithm is better?
I love my K5iis, ISO is acceptable up to 10000, imo. Seriously, how often people use extremely high ISO.

It's mushy as hell. It's an objectively bad algorithm, compared to other current ones. And it makes this comparison very difficult, because it looks like the Nikon had it turned off...

---------- Post added 19-03-16 at 18:01 ----------

Originally posted by KDAFA

The OP stated no noise reduction ....?

Then the OP is wrong. The in-camera NR is clearly being used.
If that's without NR, then god help this camera.

mee · 03-19-2016, 10:11 AM

Well.. like several of us pointed out, those images are using a very early beta firmware. This is probably exactly why Ricoh didn't want any reviews using full sized images so soon...

Giklab · 03-19-2016, 10:13 AM

Originally posted by mee

Well.. like several of us pointed out, those images are using a very early beta firmware. This is probably exactly why Ricoh didn't want any reviews using full sized images so soon...

Exactly.
Again, no point in comparing cause we'll just be disappointed

melander · 03-19-2016, 11:10 AM

Originally posted by Giklab

It's mushy as hell. It's an objectively bad algorithm, compared to other current ones. And it makes this comparison very difficult, because it looks like the Nikon had it turned off...

---------- Post added 19-03-16 at 18:01 ----------

Then the OP is wrong. The in-camera NR is clearly being used.
If that's without NR, then god help this camera.

Go here

Pentax K-1 hus digital systemkamera
Press the "testbilder" tab
At the bottom you have the iso series, the top iso series is no NR
The bottom is with NR

Giklab · 03-19-2016, 12:31 PM

Originally posted by melander

Go here

Pentax K-1 hus digital systemkamera
Press the "testbilder" tab
At the bottom you have the iso series, the top iso series is no NR
The bottom is with NR

I admit I may be wrong, but I'm still very skeptical. The first series looks remarkably similar to what the K-5 or K-3 models produce using in-camera NR.

Na Horuk · 03-19-2016, 12:44 PM - **3 Likes**

lol the replies in this thread show exactly why Pentax doesn't allow people to post full size preproduction photos.
Its a nonscientific test and one camera is a finished product, while the other is still far from it.
Noise is something that many cameras deal with automatically, even if you shoot raw - and this is brand specific, not sensor specific. And pretty much no raw looks good at 100% at high ISO (or even at moderate ISO).

I think ultimately the K-1's sensor performance will be very similar to D800E, D810.

**Rondec** · 03-19-2016, 01:09 PM - **3 Likes**

Originally posted by Giklab

I admit I may be wrong, but I'm still very skeptical. The first series looks remarkably similar to what the K-5 or K-3 models produce using in-camera NR.

I really wouldn't make too much of it. You are looking at preproduction firmware. These are jpegs, but even if you used RAW, you would be using software (ACR or whatever) that isn't designed to develop files from this camera, yet. Version 1.0 will certainly be coming out in two to three weeks and then will be the time to be upset if things aren't where they should be.

But honestly, I don't know anyone who shoots jpegs in really high iso situations. The best out of camera jpeg engine is going to be worse than a dedicated noise reduction software engine like DXO or Noise Ninja. And RAW files should be fine, regardless of what the jpeg engine looks like at 51K.

Bromberger · 03-19-2016, 01:56 PM

Originally posted by mee

I agree with you -- the K-1 high ISO shots are much more mushy than the comparable D810 shots. And there is less contrast.

This is really weird. That is not what i wrote at all.

I do NOT find the K1 shots mushy, and i do NOT find they have less contrast either.

npc · 03-19-2016, 04:46 PM

Originally posted by Rondec

I really wouldn't make too much of it. You are looking at preproduction firmware. These are jpegs, but even if you used RAW, you would be using software (ACR or whatever) that isn't designed to develop files from this camera, yet. Version 1.0 will certainly be coming out in two to three weeks and then will be the time to be upset if things aren't where they should be.

.

Idon't expect any major changes between this and the final firmware. It takes years to develop a camera and usually some of the firmware code is already borrowed from the previous modes. By the time those "preproduction" models are out to early reviewers all major software development is frozen and only bugfixes would get in. Some features that are known to have bugs anyway might be disabled initially, some features might actually get dropped in the official release and of course if early reviewers stumble on some major bugs these will be addressed - but that's it. There's no time to develop and test major features between the anouncement and the shipping date (and of course the production lines are already running).
So "preproduction firmware" in reality does not mean "still under development" but more like "we reserve the right to change some features in the final release".

Glorfindelrb · 03-19-2016, 04:55 PM

0.38 is really old, we saw a lot of 0.46 but never a so old version. I don't see why the sensor specific part would not evolve, it's obviously not shared with any existing product.

Kunzite · 03-19-2016, 04:58 PM

Actually I expect image quality tweaks to be performed (also) at the later stages, rather than adding or changing features.
There are usually two significant "freeze" points in software development: feature freeze, when you decide that you won't incorporate any new feature and instead you'll put efforts into making what was already included work as it should, and code freeze, after which only critical defects will be fixed (this prevents the QA from testing an ever changing software)
I suspect firmware 0.38 was before feature freeze, given that we could see differences in the menus from other/later versions.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Pentax K-30. ISO 400...12800.	Mr. A.S.	Pentax K-30 & K-50	9	08-11-2012 06:01 AM
Misc The fun of shooting at ISO 51200 (K-5)...	igor	Post Your Photos!	10	07-14-2011 01:43 PM
K-5, DNG/RAW ISO Case Study (6400-51200)...	JohnBee	Pentax K-5 & K-5 II	47	10-29-2010 09:14 PM
K-5 Vs K-7 ISO 6400 & 12800	eigelb	Pentax News and Rumors	7	09-28-2010 03:19 AM
Another ISO 12800 sample from D7000, good news for K-5?	leeak	Pentax News and Rumors	26	09-18-2010 03:57 PM

03-19-2016, 08:20 AM	#31
KDAFA Veteran Member Join Date: Dec 2012 Posts: 528	Originally posted by Giklab The K-1 samples are obviously OOC JPEGs at default settings. What's also painfully obvious is that the NR algorithm hasn't changed much from the K-5... The OP stated no noise reduction ....? Last edited by KDAFA; 03-19-2016 at 08:56 AM.

03-19-2016, 08:23 AM	#32
mee Veteran Member Join Date: May 2010 Posts: 7,403	Originally posted by Bromberger Looks good. Weird though: The 810 has a lot of colour noise, while the K1 has none. And colour noise can be corrected quite easily. It's almost as someone ran colour noise reduction over only one of the two pictures. Also, what's with the large horizontal lines (over the bottles, or the yellow area around the clock) in the 51k-iso-picture of the K1? It looks almost as if the scene was lit through window blinds. Not trying to bash the K1, i'm confident it's gonna be an awesome camera. Honesty, online, sometimes gets misconstrued as bashing sadly. But, for one, I agree with you -- the K-1 high ISO shots are much more mushy than the comparable D810 shots. And there is less contrast. That said, looking at the EXIF data, its still a beta firmware so perhaps this can be improved upon before release? PENTAX K-1 Ver. 0.38

03-19-2016, 08:54 AM	#33
jpzk Site Supporter Join Date: Jan 2009 Location: Québec Photos: Gallery \| Albums Posts: 8,251	Originally posted by ogl K-1 is much better, but blurred. Nikon is sharp. How can K-1 is "better" if "blurred" ? That is what I call an oxymoron https://en.wikipedia.org/wiki/Oxymoron JP

03-19-2016, 09:08 AM	#34
ogl Banned Join Date: Dec 2007 Location: Sankt Peterburg Photos: Gallery \| Albums Posts: 8,382	Originally posted by jpzk How can K-1 is "better" if "blurred" ? That is what I call an oxymoron https://en.wikipedia.org/wiki/Oxymoron JP Easy. It's two different sense. Better is terms of noise. Blurred as picture.

03-19-2016, 09:59 AM	#35
Giklab Veteran Member Join Date: Jan 2012 Location: Slovenia Photos: Gallery Posts: 2,182	Originally posted by micromacro Forgive my ignorance, but what exactly wrong with NR algorithm? Nikon algorithm is better? I love my K5iis, ISO is acceptable up to 10000, imo. Seriously, how often people use extremely high ISO. It's mushy as hell. It's an objectively bad algorithm, compared to other current ones. And it makes this comparison very difficult, because it looks like the Nikon had it turned off... ---------- Post added 19-03-16 at 18:01 ---------- Originally posted by KDAFA The OP stated no noise reduction ....? Then the OP is wrong. The in-camera NR is clearly being used. If that's without NR, then god help this camera.

03-19-2016, 10:11 AM	#36
mee Veteran Member Join Date: May 2010 Posts: 7,403	Well.. like several of us pointed out, those images are using a very early beta firmware. This is probably exactly why Ricoh didn't want any reviews using full sized images so soon...

03-19-2016, 10:13 AM	#37
Giklab Veteran Member Join Date: Jan 2012 Location: Slovenia Photos: Gallery Posts: 2,182	Originally posted by mee Well.. like several of us pointed out, those images are using a very early beta firmware. This is probably exactly why Ricoh didn't want any reviews using full sized images so soon... Exactly. Again, no point in comparing cause we'll just be disappointed

03-19-2016, 11:10 AM	#38
melander Veteran Member Join Date: Jul 2008 Location: Sweden, Umea Posts: 876	Originally posted by Giklab It's mushy as hell. It's an objectively bad algorithm, compared to other current ones. And it makes this comparison very difficult, because it looks like the Nikon had it turned off... ---------- Post added 19-03-16 at 18:01 ---------- Then the OP is wrong. The in-camera NR is clearly being used. If that's without NR, then god help this camera. Go here Pentax K-1 hus digital systemkamera Press the "testbilder" tab At the bottom you have the iso series, the top iso series is no NR The bottom is with NR

03-19-2016, 12:31 PM	#39
Giklab Veteran Member Join Date: Jan 2012 Location: Slovenia Photos: Gallery Posts: 2,182	Originally posted by melander Go here Pentax K-1 hus digital systemkamera Press the "testbilder" tab At the bottom you have the iso series, the top iso series is no NR The bottom is with NR I admit I may be wrong, but I'm still very skeptical. The first series looks remarkably similar to what the K-5 or K-3 models produce using in-camera NR.

03-19-2016, 12:44 PM - 3 Likes	#40
Na Horuk Veteran Member Join Date: Feb 2012 Location: Slovenia, probably Photos: Gallery \| Albums Posts: 11,186	lol the replies in this thread show exactly why Pentax doesn't allow people to post full size preproduction photos. Its a nonscientific test and one camera is a finished product, while the other is still far from it. Noise is something that many cameras deal with automatically, even if you shoot raw - and this is brand specific, not sensor specific. And pretty much no raw looks good at 100% at high ISO (or even at moderate ISO). I think ultimately the K-1's sensor performance will be very similar to D800E, D810.

03-19-2016, 01:09 PM - 3 Likes	#41
Rondec Loyal Site Supporter Join Date: Mar 2009 Location: Gladys, Virginia Photos: Gallery Posts: 27,663	Originally posted by Giklab I admit I may be wrong, but I'm still very skeptical. The first series looks remarkably similar to what the K-5 or K-3 models produce using in-camera NR. I really wouldn't make too much of it. You are looking at preproduction firmware. These are jpegs, but even if you used RAW, you would be using software (ACR or whatever) that isn't designed to develop files from this camera, yet. Version 1.0 will certainly be coming out in two to three weeks and then will be the time to be upset if things aren't where they should be. But honestly, I don't know anyone who shoots jpegs in really high iso situations. The best out of camera jpeg engine is going to be worse than a dedicated noise reduction software engine like DXO or Noise Ninja. And RAW files should be fine, regardless of what the jpeg engine looks like at 51K.

03-19-2016, 01:56 PM	#42
Bromberger Senior Member Join Date: Feb 2015 Location: Stuttgart, Germany Posts: 186	Originally posted by mee I agree with you -- the K-1 high ISO shots are much more mushy than the comparable D810 shots. And there is less contrast. This is really weird. That is not what i wrote at all. I do NOT find the K1 shots mushy, and i do NOT find they have less contrast either.

03-19-2016, 04:46 PM	#43
npc Veteran Member Join Date: Apr 2013 Posts: 313	Originally posted by Rondec I really wouldn't make too much of it. You are looking at preproduction firmware. These are jpegs, but even if you used RAW, you would be using software (ACR or whatever) that isn't designed to develop files from this camera, yet. Version 1.0 will certainly be coming out in two to three weeks and then will be the time to be upset if things aren't where they should be. . Idon't expect any major changes between this and the final firmware. It takes years to develop a camera and usually some of the firmware code is already borrowed from the previous modes. By the time those "preproduction" models are out to early reviewers all major software development is frozen and only bugfixes would get in. Some features that are known to have bugs anyway might be disabled initially, some features might actually get dropped in the official release and of course if early reviewers stumble on some major bugs these will be addressed - but that's it. There's no time to develop and test major features between the anouncement and the shipping date (and of course the production lines are already running). So "preproduction firmware" in reality does not mean "still under development" but more like "we reserve the right to change some features in the final release".

03-19-2016, 04:55 PM	#44
Glorfindelrb Veteran Member Join Date: Feb 2012 Posts: 364	0.38 is really old, we saw a lot of 0.46 but never a so old version. I don't see why the sensor specific part would not evolve, it's obviously not shared with any existing product.

03-19-2016, 04:58 PM	#45
Kunzite Pentaxian Join Date: Apr 2007 Location: Romania Posts: 15,132	Actually I expect image quality tweaks to be performed (also) at the later stages, rather than adding or changing features. There are usually two significant "freeze" points in software development: feature freeze, when you decide that you won't incorporate any new feature and instead you'll put efforts into making what was already included work as it should, and code freeze, after which only critical defects will be fixed (this prevents the QA from testing an ever changing software) I suspect firmware 0.38 was before feature freeze, given that we could see differences in the menus from other/later versions.