K7 - preprocessed RAW files - noise is worse than seemed

Started Jul 9, 2009 | Discussions
Oleg_V Contributing Member • Posts: 627
K7 - preprocessed RAW files - noise is worse than seemed

I've being waiting k7 samples to find out either pentax will continue to apply
noise reduction to RAW files at high ISO or choose to improve sensor instead.
Previously there were some study on this regard started here
http://forums.dpreview.com/forums/readflat.asp?forum=1036&message=27206266 .
Recently the preprocessing of RAW files were admitted even by DXO mark. Now
looking at the IR review samples I can say that nothing is changed. The ISO 6400
samples demonstrate very strong preprocessing that reveal themselves by strong
correlation of noise. The NR procedure involves averaging of the adjacent bayer
cells as well as averaging of the green sensors inside the single cell. Needless
to say that it kills details irreversibly.

I've applied the same detection algorithm to nikon NEF files available at IR
cite. Surprisingly, I've failed to see any signs of hidden NR applied to them
for D300 and D90 at ISO6400 as well as D700 at ISO25600. Since I also have nikon
lenses accompanied by my old D70 camera, I ultimately decided to stop any
investment in pentax.

sfa1966
sfa1966 Veteran Member • Posts: 3,474
Re: K7 - preprocessed RAW files - noise is worse than seemed

Oleg_V wrote:

The NR procedure involves averaging of the adjacent bayer
cells as well as averaging of the green sensors inside the single cell. Needless
to say that it kills details irreversibly.

You wouldn't happen to be a virologist by profession would you?

I've applied the same detection algorithm to nikon NEF files available at IR cite. Surprisingly, I've failed to see any signs of hidden NR applied to them [snip] ... I ultimately decided to stop any investment in pentax.

Good call. So many of my ISO6400 shots are totally and irreversibly ruined by the 'Pentax single-cell averaging green sensor syndrome' that I, and surely many others, will make a similar decision soon.

-- hide signature --

Cheers,
sfa

A very limited photographer ...

 sfa1966's gear list:sfa1966's gear list
Sigma DP1 Sigma DP2 Merrill Pentax K-3 Pentax K-1 Pentax smc DA 15mm F4 ED AL Limited +11 more
jonny1976 Veteran Member • Posts: 4,277
Re: K7 - preprocessed RAW files - noise is worse than seemed

how big do you print a iso 3200 and above file?
do u use htese files to make a living?

Oleg_V wrote:

I've being waiting k7 samples to find out either pentax will continue to apply
noise reduction to RAW files at high ISO or choose to improve sensor instead.
Previously there were some study on this regard started here
http://forums.dpreview.com/forums/readflat.asp?forum=1036&message=27206266 .
Recently the preprocessing of RAW files were admitted even by DXO mark. Now
looking at the IR review samples I can say that nothing is changed. The ISO 6400
samples demonstrate very strong preprocessing that reveal themselves by strong
correlation of noise. The NR procedure involves averaging of the adjacent bayer
cells as well as averaging of the green sensors inside the single cell. Needless
to say that it kills details irreversibly.

I've applied the same detection algorithm to nikon NEF files available at IR
cite. Surprisingly, I've failed to see any signs of hidden NR applied to them
for D300 and D90 at ISO6400 as well as D700 at ISO25600. Since I also have nikon
lenses accompanied by my old D70 camera, I ultimately decided to stop any
investment in pentax.

-- hide signature --
 jonny1976's gear list:jonny1976's gear list
Nikon D2X
jonny1976 Veteran Member • Posts: 4,277
Re: K7 - preprocessed RAW files - noise is worse than seemed

i have sold many iso 3200 6400 of k10d aat newspaper that were heavily underexposed and printed well ....i don't need iso 6400 to print larger than 4*6 or 3200 for 8*10...the same do my friend with d3 or d700.

all this discussion about high is are for those who don'tknow how to use their time well.

lastly..some of the best shots i have seen on national geo and other paper are really grainy but they were great shots...far better to have detail and fine grain than plastic images.

sfa1966 wrote:

Oleg_V wrote:

The NR procedure involves averaging of the adjacent bayer
cells as well as averaging of the green sensors inside the single cell. Needless
to say that it kills details irreversibly.

You wouldn't happen to be a virologist by profession would you?

I've applied the same detection algorithm to nikon NEF files available at IR cite. Surprisingly, I've failed to see any signs of hidden NR applied to them [snip] ... I ultimately decided to stop any investment in pentax.

Good call. So many of my ISO6400 shots are totally and irreversibly ruined by the 'Pentax single-cell averaging green sensor syndrome' that I, and surely many others, will make a similar decision soon.

-- hide signature --
 jonny1976's gear list:jonny1976's gear list
Nikon D2X
Jimbob Productions Veteran Member • Posts: 3,776
Pre-prod

Oleg aren't the test shots at IR done with a pre-prod body?

Roland Mabo Forum Pro • Posts: 12,462
Re: K7 - preprocessed RAW files - noise is worse than seemed

Why would this be of interrest?

Isn't the final look of the image most important than the technology underneath it?
I'm not interrested in technology, I'm interrested in the images.

And I'm interresed in how the images looks when prined, not 100% crops and not of any signal analysis becauset that is like measuring speakers instead of listening to them. I prefer listening, I prefer using my eyes. That is how I look at images.

-- hide signature --
LuzArt Veteran Member • Posts: 6,643
Stop Making Sense!

Roland, you know this is not the forum for making sense!

Incidently, the remastered version of Talking Heads' Stop Making Sense is superb . I know this because I listened to it through my nice speakers, instead of measuring and analysing them

Ben
--
http://www.luzartphotography.com
http://luzart.blogspot.com

 LuzArt's gear list:LuzArt's gear list
Nikon D700 Nikon AF-S Nikkor 70-200mm f/2.8G ED VR II Nikon AF-S Teleconverter TC-20E III Tamron AF 28-75mm F/2.8 XR Di LD Aspherical (IF) +2 more
gaddigad Senior Member • Posts: 1,906
Re: K7 - preprocessed RAW files - noise is worse than seemed

Oleg I'm interested to know whether a raw conversion software can do single-cell averaging on the several photosites comprising one pixel (including the two green ones)?

Vaards Senior Member • Posts: 1,328
Thanks for post!

Oleg_V wrote:

I've being waiting k7 samples to find out either pentax will continue to apply
noise reduction to RAW files at high ISO or choose to improve sensor instead.
Previously there were some study on this regard started here
http://forums.dpreview.com/forums/readflat.asp?forum=1036&message=27206266 .
Recently the preprocessing of RAW files were admitted even by DXO mark. Now
looking at the IR review samples I can say that nothing is changed. The ISO 6400
samples demonstrate very strong preprocessing that reveal themselves by strong
correlation of noise. The NR procedure involves averaging of the adjacent bayer
cells as well as averaging of the green sensors inside the single cell. Needless
to say that it kills details irreversibly.

I've applied the same detection algorithm to nikon NEF files available at IR
cite. Surprisingly, I've failed to see any signs of hidden NR applied to them
for D300 and D90 at ISO6400 as well as D700 at ISO25600. Since I also have nikon
lenses accompanied by my old D70 camera, I ultimately decided to stop any
investment in pentax.

Sad if you really want to leave Pentax camp. But it seems inevitable - all valuable Pentaxians sooner or later grows up and leaves Pentax. And the most sad fact - average concentration of ignorant users increases.

exdeejjjaaaa
exdeejjjaaaa Veteran Member • Posts: 8,263
Re: K7 - preprocessed RAW files - noise is worse than seemed

Roland Mabo wrote:

Why would this be of interrest?

Isn't the final look of the image most important than the technology underneath it?
I'm not interrested in technology, I'm interrested in the images.

And I'm interresed in how the images looks when prined, not 100% crops and not of any signal analysis becauset that is like measuring speakers instead of listening to them. I prefer listening, I prefer using my eyes. That is how I look at images.

Roland, who cares how you look at your images ? this topic is a purely technical discussion about the sensor...

-- hide signature --

 exdeejjjaaaa's gear list:exdeejjjaaaa's gear list
Sony a7R II Sony FE 55mm F1.8 Phase One Capture One Pro +25 more
Roland Mabo Forum Pro • Posts: 12,462
Re: Stop Making Sense!

LuzArt wrote:

Roland, you know this is not the forum for making sense!

You know, I never learn..

Incidently, the remastered version of Talking Heads' Stop Making Sense is > superb . I know this because I listened to it through my nice speakers, instead > of measuring and analysing them

Thank's for the tip. I will look, sorry - listening, into it.

Speaking about sound, I haven't listened to a CD for more than two months!

Why? Well, I don't know really. I bought a new Goldring pick-up for my Thorens disc spinner and then I replaced my ageing NAD PP-1 with a Thorens MM-01, and suddenly I was happy. Can't explaint it really, but feels no need for CD's at the moment.

-- hide signature --
Roland Mabo Forum Pro • Posts: 12,462
Re: Thanks for post!

I see it more like those that measures images rather than shooting and enjoying images for the art in them, are laving the Pentax camp.

Left are those who shoot images, prints them, looks at them, and gaze in wonder "aaah" or "oooh".
I don't see any disadvantage with this, honestly.

I remember, ten years ago when everyone was shooting film, we looked at images and we discussed images. Not measured resolution and technical charts.

-- hide signature --
exdeejjjaaaa
exdeejjjaaaa Veteran Member • Posts: 8,263
Re: Thanks for post!

Roland Mabo wrote:

I see it more like those that measures images rather than shooting and enjoying images for the art in them, are laving the Pentax camp.

Left are those who shoot images, prints them, looks at them, and gaze in wonder "aaah" or "oooh".
I don't see any disadvantage with this, honestly.

I remember, ten years ago when everyone was shooting film, we looked at images and we discussed images. Not measured resolution and technical charts.

yet, you all "those who shoot images" are upgrading to new cameras... why don't you stay w/ your K10Ds/K20Ds or ol' good *istD ? oh... there are some things apparently better in K-7 ? seems it is the case... so please accept the fact that for some people, who do not print, the sensor is important too...

-- hide signature --

 exdeejjjaaaa's gear list:exdeejjjaaaa's gear list
Sony a7R II Sony FE 55mm F1.8 Phase One Capture One Pro +25 more
awaldram
awaldram Forum Pro • Posts: 13,271
Re: K7 - preprocessed RAW files - noise is worse than seemed

exdeejjjaaaa wrote:

Roland Mabo wrote:

Why would this be of interrest?

Isn't the final look of the image most important than the technology underneath it?
I'm not interrested in technology, I'm interrested in the images.

And I'm interresed in how the images looks when prined, not 100% crops and not of any signal analysis becauset that is like measuring speakers instead of listening to them. I prefer listening, I prefer using my eyes. That is how I look at images.

Roland, who cares how you look at your images ? this topic is a purely technical discussion about the sensor...

Don't think this could be classed as technical.

Cmos sensor have on chip hardware noise reduction
CCD doesn't

therefore CMOS raw has Noise reduction as that's what the sensor gives
CCD doesn't

Nothing Technical at all.

Noise reduction at sensor = RAW data with noise reduction.

Roland is right if you like what you see buy if not don't .

-- hide signature --

 awaldram's gear list:awaldram's gear list
Pentax K-x Pentax Q Olympus PEN E-PM2 Pentax Q7 Pentax K-3 +17 more
ddugan New Member • Posts: 14
Re: K7 - preprocessed RAW files - noise is worse than seemed

As funny as it sounds, I sorta like the noise on the ISO 6400 shots. I think you could do a lot of cool things with it depending on the subject matter and what you might to post-processing.

As an owner of K20 and now a K7, I can definitely tell some differences between the two cameras. There is a lot to like about the new K7, but there is some improvement room for it too. But heck, I'm just going to shoot some pictures and not worry about it. I'll adjust to whatever camera I have, and that will be that.

Thanks for the thoughts,

daren

Roland Karlsson Forum Pro • Posts: 29,124
Re: K7 - preprocessed RAW files - noise is worse than seemed

Roland Mabo wrote:

Isn't the final look of the image most important than the technology underneath it?

It is perfectly possible to make measurements that correlate to the final look. And if that turns out to be totally impossible - then maybe the final look is just pure imagination and myth.

I'm not interrested in technology, I'm interrested in the images.

OK - what are you doing at DPReview then? Why are you in a technology forum? Why bother about the K-7? Why not buy just any camera and make images? If technology is totally irrelevant.

And I'm interresed in how the images looks when prined, not 100% crops and not of any signal analysis

Still - there is a correlation. And if you dont care for signal analysis - then maybe someone else do.

becauset that is like measuring speakers instead of listening to them. I prefer listening,

Good speakers is all about technology. Without good technological knowledge you cannot make good speakers.

I prefer using my eyes. That is how I look at images.

Me too.

But still - why are you here? This is not an image forum.

-- hide signature --
 Roland Karlsson's gear list:Roland Karlsson's gear list
Sigma DP3 Merrill Sigma dp2 Quattro Sony RX100 III Pentax K-3 Pentax K-1 +14 more
OP Oleg_V Contributing Member • Posts: 627
Re: K7 - preprocessed RAW files - noise is worse than seemed

awaldram wrote:

Cmos sensor have on chip hardware noise reduction
CCD doesn't

Nothing more than common misconception. CMOS does have additional in-pixel circuits, sometimes they help to reduce readout noise. I'm talking about processing involving more than one pixel that reduce signal bandwidth in spacial frequency domain so killing details and reducing noise. No one except pentax and sony are using such nasty technology. Moreover - as pointed out by DXO mark pentax is using it since K10 which is not CMOS at all. After all you are all excited about increasing resolution (otherwise you keeps shooting with 6Mp models) but NR is just killing that resolution without any option to turn it off.

Roland Mabo Forum Pro • Posts: 12,462
Re: K7 - preprocessed RAW files - noise is worse than seemed

Who cares about what you think about my posts?

I have the right to express my opinion that if one selects camera after technical analysis of the signal produced by the sensor, instead of how the images looks when printed - then this is a bit odd when we talk about photography beacuse photography is a form of art, and art is about subjectivity and expression. Not technical analysis.
What counts is the images.

-- hide signature --
*isteve Veteran Member • Posts: 9,509
Re: Thanks for post!

exdeejjjaaaa wrote:

yet, you all "those who shoot images" are upgrading to new cameras... why don't you stay w/ your K10Ds/K20Ds or ol' good *istD ? oh... there are some things apparently better in K-7 ? seems it is the case... so please accept the fact that for some people, who do not print, the sensor is important too...

People who own a DSLR and do not print their photographs are very hard to take seriously even on a "technical" forum.
--
Steve

Any fool can take a picture OF something. Its much harder to take a picture ABOUT something.

Roland Mabo Forum Pro • Posts: 12,462
Re: Thanks for post!

There are things better with the K-7 yes, but saying that one shouldn't buy it because of an analysis based upon the signal from the sensor - instead of how the images looks when printed - is something that I can't agree with.

Aren't cameras supposed to take photographs? Isn't this the sole idea?

Now, there is a difference between looking at images - looking at photographs - and analysing signal and saying that one shouldn't buy the K-7 beacuse it applies NR to the signal.

It doesn't matter if the K-7 applies NR or not, if the final images looks good.
And this is what counts in the end - the image, the photograph.

And if you don't like the images, then fine - don't buy the K-7.
But not buying it because an analyse of the signal from the sensor shows up x...
Strange, very strange.

If the images looks good, and we all can see in this forum many great images from the K-7 already, so why should it matter what is done to the signal or not - as long as the end result is good?

Do you mean that consumers should analyse the signal data instead of looking at the images? It is like telling consumers that they shouldn't listen to loudspeakers, only measure them.

-- hide signature --
Keyboard shortcuts:
FForum MMy threads