Just for New Users.. A700 High ISO NR Why Adobe Tests are not valid.

headofdestiny · Feb 3, 2008

Jeez, I just spent the last day on another more "professional" forum, and this RAW NR was a topic of conversation. The difference was, they were talking about the difference in NR between Nikon and Canon. They all acknowledge that both cameras do NR on RAW, and most of them described the Nikon RAW in the same way folks are describing our Sony's "cooked" RAW. In fact, some of the Canon owners thought that Canon should start using more/different NR to keep up with Nikon, and that the D3 has a significant noise advantage because of what Nikon's doing to process the files.

Bottom line: Nikon, Sony, and Canon "cook" their RAW. Some may have more NR than others, or some may be more pleasing to one person than the other, but they all do it to a degree. Apparently Canon does it less than Sony and Nikon, but it is there.

I fully support people saying that they do not prefer the Sony A700 high ISO RAW file to it's competition, but I don't support people saying that Sony doesn't have real RAW. I can't imagine what torment these same people had to deal with when choosing a type of film! Let's see, how Provia 400...no wait....Velvia 50....no wait.....Astia 100.....argh! Portra 160!!!!

headofdestiny · Feb 3, 2008

p.s. I was also surprised how open to the idea of Sony so many pro shooters using 1Ds' and digital backs are on some forums. I'm not saying that I think they'll switch, but they wouldn't be against it if the product is good. It sure is funny how much brand loyalty amateurs and semi-pros have compared to a lot of high end pros I read about or meet.

David Kilpatrick · Feb 3, 2008

Ken_5D said:
But most of the problems being blamed on the Sony are the result of
Abobe being what people assume to be the standard.

(---)

Ken_5D said:
Nothing Sony does in the sensor cooks the RAW, Adobe burns the RAW
when it comes to A700 high ISO.. shame on Adobe.

Ken, I've got high ISO raw files from D300, D3, E3, 40D (not so much, I was lax on shooting these) and A700.

The mystery to me is this - RAW from the Nikon should, surely, be much like RAW from the Sony. Same number of pixel locations, same size file, same Bayer pattern.

Yet ACR 4.3.x has consisently produced a tight, regular grain pattern from Nikon and a loose, random structure with a different type of sharpening from A700 - no matter what settings. Both conversions set to zero everything and the A700 remains with tighter, defined patches of colour blurring and similar sized patches of luminance blur. The Nikon has larger, softer edged fusion of colour blurring and much less luminance blur.

The question is - why if the files should be so similar is one algorithm used for Sony, and another for Nikon? Are the raw files somehow instructing ACR to vary its presets? Or are they genuinely different?

David

--
Publishing & Editing Photoworld (photoclubalpha.com) and Master Photo Digital
Currently writing for f2 and the BJP

Ken_5D · Feb 3, 2008

headofdestiny said:
p.s. I was also surprised how open to the idea of Sony so many pro
shooters using 1Ds' and digital backs are on some forums. I'm not
saying that I think they'll switch, but they wouldn't be against it
if the product is good. It sure is funny how much brand loyalty
amateurs and semi-pros have compared to a lot of high end pros I read
about or meet.

I'm not.. REAL pros want the best equpment brand be damned. Replacing $15000 in body and glass is like buying a new car. Do it and write if off when it helps the business.

no one secure in thier own ability latches unconditionally to any brand. period.

I was 2-3 features and/or bad IQ from buying a D300.. I am loyal to what does the job best.
------------
Ken - Happy A700 Owner
http://www.cascadephotoworks.com

Ken_5D · Feb 3, 2008

David Kilpatrick said:
Ken_5D said:

But most of the problems being blamed on the Sony are the result of
Abobe being what people assume to be the standard.

Click to expand...

(---)

David Kilpatrick said:

Nothing Sony does in the sensor cooks the RAW, Adobe burns the RAW
when it comes to A700 high ISO.. shame on Adobe.

Click to expand...

Ken, I've got high ISO raw files from D300, D3, E3, 40D (not so much,
I was lax on shooting these) and A700.

The mystery to me is this - RAW from the Nikon should, surely, be
much like RAW from the Sony. Same number of pixel locations, same
size file, same Bayer pattern.

Yet ACR 4.3.x has consisently produced a tight, regular grain pattern
from Nikon and a loose, random structure with a different type of
sharpening from A700 - no matter what settings. Both conversions set
to zero everything and the A700 remains with tighter, defined patches
of colour blurring and similar sized patches of luminance blur. The
Nikon has larger, softer edged fusion of colour blurring and much
less luminance blur.

The question is - why if the files should be so similar is one
algorithm used for Sony, and another for Nikon? Are the raw files
somehow instructing ACR to vary its presets? Or are they genuinely
different?

David

--
Publishing & Editing Photoworld (photoclubalpha.com) and Master Photo
Digital
Currently writing for f2 and the BJP

I have heard 2nd hand from other posters that Adobe is adjusting thier raw process to fit the camera.. odds are they didn't spend much time with the Sony and to be fair Sony IDC is soft on the raw too.

Bibble / Raw therapee are camera agnostic.

I am quite sure the Sony sensor output is different from the D300...if it is the same base Sony didn't deliver all the NR or Nikon turns it off. I am not questioning that Sony RAW looks a bit different, but it is not the over processed blob fest Adobe.. has lead us to belive..but even the DXO tests on DWIA labs site says the grain is 30-50% larger than the Canon.

The key is comparing the output in an agnostic RAW system vs one that thinkgs it knows what NR you want on your RAW based on the camera.
------------
Ken - Happy A700 Owner
http://www.cascadephotoworks.com

David Kilpatrick · Feb 3, 2008

Ken_5D said:
But now we have tests, and we have some more info form PMA vs Phils
guess based on an unproven assumption that the D300 and A700 use the
very same sensor and use the very same output of that sensor.

Ken, the sensors are both IMX021-based Sony sensors, there is no 'unproven' about that at all. Nikon has released photographs and so has Sony.

Nikon sensor

Sony sensor

The aspects Nikon can change are:

The filter values
The microlens design
The distance and strength of the low-pass (AA) filter
The degree of UV/IR filtering
The firmware parameters for onboard A to D, presuming these are addressable
The actual onboard A to D processors themselves

'The very same sensor' is pretty much proven as far as the silicon in the middle goes. 'The very same output' is definitely not true, as Nikon states that the built-in A to D converter on the sensor offers 12 or 14 bit conversion. Sony only state that it allows 12 bit. Presumption: Nikon has different A to D converters.

Even so, it's very hard to grasp why both running at 12 bit should not produce nearly identical RAW files.

David

--
Publishing & Editing Photoworld (photoclubalpha.com) and Master Photo Digital
Currently writing for f2 and the BJP

headofdestiny · Feb 3, 2008

Ken_5D said:
headofdestiny said:

p.s. I was also surprised how open to the idea of Sony so many pro
shooters using 1Ds' and digital backs are on some forums. I'm not
saying that I think they'll switch, but they wouldn't be against it
if the product is good. It sure is funny how much brand loyalty
amateurs and semi-pros have compared to a lot of high end pros I read
about or meet.

Click to expand...

I'm not.. REAL pros want the best equpment brand be damned.
Replacing $15000 in body and glass is like buying a new car. Do it
and write if off when it helps the business.

no one secure in thier own ability latches unconditionally to any
brand. period.

I was 2-3 features and/or bad IQ from buying a D300.. I am loyal to
what does the job best.
------------
Ken - Happy A700 Owner
http://www.cascadephotoworks.com

Sorry Ken, I wasn't referring to you or anybody specifically. I spend a lot of time on the 1Ds forum, and when I went to another professional, Canon-centric site, I was surprised how much more willing to discuss things people were. I'm fully in support of your position that RAW is not cooked in the A700...unless one thinks that Canon and Nikon RAWs are considered cooked, too. We may not know the recipes, but it seems to me that all of the companies are doin' a little cookin'. It looks like in the A700's case, it's the analog NR + digital NR, and I imagine the digital NR is the one that is adjustable.

Ehrik · Feb 3, 2008

headofdestiny said:
Jeez, I just spent the last day on another more "professional" forum,
and this RAW NR was a topic of conversation. The difference was,
they were talking about the difference in NR between Nikon and Canon.
They all acknowledge that both cameras do NR on RAW

I think they confuse NR in the raw converter they are using with NR on the
raw files. A number of people have analysed Nikon and Canon raws (prior to
conversion) and found no NR smearing on them.

I think also that the scene in the OP is not of the kind that is prone to
exhibit NR artefacts. A scene with deep shadows and fine low-contrast
detail would be more suitable.

Just my two oere
Erik from Sweden

kiklop_2 · Feb 3, 2008

Ken_5D said:
I have heard 2nd hand from other posters that Adobe is adjusting
thier raw process to fit the camera.. odds are they didn't spend much
time with the Sony and to be fair Sony IDC is soft on the raw too.

That's misleading to say at least. Adobe can't influence the noise pattern found on raw files; blaming adobe is just plain wrong !

Both d300 and a700 process raw files (for good sake, processing raw files is one advantage of CMOS technology) but on d300 this process can be lowered much more than with a700 and Nikon is doing better job in doing it (finer noise .. more pleasing to the eye and more conservative on destroying details).

Ken_5D said:
The key is comparing the output in an agnostic RAW system vs one that
thinkgs it knows what NR you want on your RAW based on the camera.

David may provide you with raw comparisons files he taken with both d300 and a700. Convert them with DCRAW and explore; what you will find (again) is th same thing David is mentioning .. the only difference with ACR is that ACR makes this differences more visible .. that's all.

Happy shooting !

--
-- http://www.dyxum.com - the world of the Minolta/Alpha mount digital SLR

headofdestiny · Feb 3, 2008

Ehrik said:
headofdestiny said:

Jeez, I just spent the last day on another more "professional" forum,
and this RAW NR was a topic of conversation. The difference was,
they were talking about the difference in NR between Nikon and Canon.
They all acknowledge that both cameras do NR on RAW

Click to expand...

I think they confuse NR in the raw converter they are using with NR
on the
raw files. A number of people have analysed Nikon and Canon raws
(prior to
conversion) and found no NR smearing on them.

I think also that the scene in the OP is not of the kind that is
prone to
exhibit NR artefacts. A scene with deep shadows and fine low-contrast
detail would be more suitable.

Just my two oere
Erik from Sweden

Nope, definitely not. It was all about RAW NR in camera.

dkloi · Feb 3, 2008

David Kilpatrick said:
The aspects Nikon can change are:

The filter values
The microlens design
The distance and strength of the low-pass (AA) filter
The degree of UV/IR filtering
The firmware parameters for onboard A to D, presuming these are
addressable
The actual onboard A to D processors themselves

[SNIP]

David Kilpatrick said:
Even so, it's very hard to grasp why both running at 12 bit should
not produce nearly identical RAW files.

The filter values could have a large impact in the low SNR limit. If Sony are using denser filters (for better colour accuracy?) this could impact upon the results. You said that the A700 has better colour rendition than the D300, this could be a possible reason.

I still am a bit puzzled by what sort of non-local RAW filtering one could directly do (without having to interpolate). One possibility is to exploit the fact that read-noise has spatial frequencies higher than the cut-off of the AA filter. Still, I'm still of the opinion that any digital NR is being performed off the sensor.

Cheers,
Daniel.

RJSanchez · Feb 3, 2008

Ken_5D said:
I have heard 2nd hand from other posters that Adobe is adjusting
thier raw process to fit the camera.. odds are they didn't spend much
time with the Sony and to be fair Sony IDC is soft on the raw too.

Bibble / Raw therapee are camera agnostic.

I am quite sure the Sony sensor output is different from the
D300...if it is the same base Sony didn't deliver all the NR or Nikon
turns it off. I am not questioning that Sony RAW looks a bit
different, but it is not the over processed blob fest Adobe.. has
lead us to belive..but even the DXO tests on DWIA labs site says the
grain is 30-50% larger than the Canon.

The key is comparing the output in an agnostic RAW system vs one that
thinkgs it knows what NR you want on your RAW based on the camera.
------------
Ken - Happy A700 Owner
http://www.cascadephotoworks.com

Ken, I am also trying to become enlightened on this subject since I shoot RAW almost exclusively on my A100 and an considering upgrading to the A700. You stated that you have heard second hand reports that Adobe has not spent much time with the A700 RAW. Have you spoken with anyone at Adobe to verify that they are applying NR to the A700 RAW even when the NR slider is set to zero? What I don't understand is why of all the cameras that use RAW, only the A700 seems to have this problem with the Adobe RAW conversion. Why would they single out the A700? If they indeed are doing uncalled for NR, then they should be made aware of the fact that it may well cost them a good number of sales.

lonewolf69 · Feb 3, 2008

Dear All,

With all due respect, is anyone here disputing that the IQ of the Bibble files are much more "pleasing" to the eyes than the Adobe? To my eyes, there is something going on with Bibble and the results are very NICE...

It's fine and dandy to banter back on forth about the dynamics of what NR is being done on the processor and if it's being done POST RAW or PRE RAW, or if D300 is LESS NR artifacts than the A700, but BOTTOMLINE the Bibble images are great looking...

Why are we still arguing about this, what's wrong with using Bibble as the RAW processor/converter, and/or why CAN'T Adobe perform like Bibble? Surely if an independent company like Bibble can produce results like these, it should be a no brainer for Adobe...?!?

I think THAT would be a much more interesting question, don't you agree?
--
-Alex

From the minds of Minolta to the imagination of Sony, Alpha, like no other.

http://www.pbase.com/lonewolf69

Ken_5D · Feb 3, 2008

kiklop_2 said:
Ken_5D said:

I have heard 2nd hand from other posters that Adobe is adjusting
thier raw process to fit the camera.. odds are they didn't spend much
time with the Sony and to be fair Sony IDC is soft on the raw too.

Click to expand...

That's misleading to say at least. Adobe can't influence the noise
pattern found on raw files; blaming adobe is just plain wrong !

Really software can't have an effect in how it demosiaces and process it.. tell me why they same file with both NR and sharpening zeroed has a finer noise pattern in Bibble then Adobe.. Magic? or did Sony put a bug in the RAW to make Adobe smooth out the detail.. please that is one of the silliest things I have heard RAW processor can't affect noise.. hell I want a refund for my Noise Ninja!!! I can do a darn thing according to your.

I am really surprised to here you even try to say something like this.. since we all know that there are lots of difference in RAW converstion including several different algorythms for doing it. And clearly Sony is making calls about the finer noise that Bibble and RAW Therapee and even ACR 3.4 leave more to the user.

kiklop_2 said:
Both d300 and a700 process raw files (for good sake, processing raw
files is one advantage of CMOS technology) but on d300 this process
can be lowered much more than with a700 and Nikon is doing better job
in doing it (finer noise .. more pleasing to the eye and more
conservative on destroying details).

kiklop_2 said:

The key is comparing the output in an agnostic RAW system vs one that
thinkgs it knows what NR you want on your RAW based on the camera.

Click to expand...

David may provide you with raw comparisons files he taken with both
d300 and a700. Convert them with DCRAW and explore; what you will
find (again) is th same thing David is mentioning .. the only
difference with ACR is that ACR makes this differences more visible
.. that's all.

Happy shooting !

--
-- http://www.dyxum.com - the world of the Minolta/Alpha mount digital SLR

--
------------
Ken - Happy A700 Owner
http://www.cascadephotoworks.com

bobmax · Feb 3, 2008

Did you post your D300/A700 comparison images on either your site or here on DPreview (Sony or Nikon forums)?

Would like to see them.... I recall some E-3 images, but must have spaced on the D300 comparison images (comparison w/the A700).

Thanks, Bob
--
Bob

Ken_5D · Feb 3, 2008

David Kilpatrick said:
Ken_5D said:

But now we have tests, and we have some more info form PMA vs Phils
guess based on an unproven assumption that the D300 and A700 use the
very same sensor and use the very same output of that sensor.

Click to expand...

Ken, the sensors are both IMX021-based Sony sensors, there is no
'unproven' about that at all. Nikon has released photographs and so
has Sony.

Nikon sensor

Sony sensor

The aspects Nikon can change are:

The filter values
The microlens design
The distance and strength of the low-pass (AA) filter
The degree of UV/IR filtering
The firmware parameters for onboard A to D, presuming these are
addressable
The actual onboard A to D processors themselves

SO? then it is not same sensor it is a sub-set of a sensor technology once you start changing things.. Like a car is not the same even if it has the same engine if you change out the transmission. Nikon clearly is taking the info from the sensor sites and dealing with differently from Sony probably as you say by changing one more aspects of AD.. since that is where Sony asplaces some of its NR. I can show you some matchin images of 1980;s CPUs and one of them wouldn't run any of the code from the other. Pictures with a standard output package do not tell you what the logic is in the chip..

David Kilpatrick said:
'The very same sensor' is pretty much proven as far as the silicon in
the middle goes. 'The very same output' is definitely not true, as
Nikon states that the built-in A to D converter on the sensor offers
12 or 14 bit conversion. Sony only state that it allows 12 bit.
Presumption: Nikon has different A to D converters.

Again if you change out key parts it is not the same senor any more it is part of a sensor family ...

David Kilpatrick said:
Even so, it's very hard to grasp why both running at 12 bit should
not produce nearly identical RAW files.

David

--
Publishing & Editing Photoworld (photoclubalpha.com) and Master Photo
Digital
Currently writing for f2 and the BJP

--
------------
Ken - Happy A700 Owner
http://www.cascadephotoworks.com

billw54 · Feb 3, 2008

kiklop_2 said:
Both d300 and a700 process raw files (for good sake, processing raw
files is one advantage of CMOS technology) but on d300 this process
can be lowered much more than with a700 and Nikon is doing better job
in doing it (finer noise .. more pleasing to the eye and more
conservative on destroying details).

The "noise reduction" / processing that happens in the sensor chip (Exmor) is reset-noise subtraction. Every CMOS chip does this, because otherwise they would be horribly noisy. This is done on a per pixel basis and cannot result in splotching or loss of detail. There is really nothing to turn down or up, nor any algorithm to tweak.

The Bionz chip is responsible for "traditional" noise reduction. As Walt has pointed out, doing Noise-ninja-style NR on the Bayer pattern data is tricky, and considerably different from applying the same technique to the RGB data after de-mosaicing.

I am of the opinion that we will never really know why the Nikon output looks different from the Sony output.

What is important is if you are satisfied with the Sony output or not. And how important ISO 1600 - 3200 is to you, and what you intend to use it for.

-- Bill
http://billw.smugmug.com

bobmax · Feb 3, 2008

On the D300 forum - reference to your PBase.

http://www.pbase.com/davidkilpatrick/d300a700a1007d_test&page=all

Thanks, Bob
--
Bob

Ken_5D · Feb 3, 2008

headofdestiny said:
Ken_5D said:

headofdestiny said:

p.s. I was also surprised how open to the idea of Sony so many pro
shooters using 1Ds' and digital backs are on some forums. I'm not
saying that I think they'll switch, but they wouldn't be against it
if the product is good. It sure is funny how much brand loyalty
amateurs and semi-pros have compared to a lot of high end pros I read
about or meet.

Click to expand...

I'm not.. REAL pros want the best equpment brand be damned.
Replacing $15000 in body and glass is like buying a new car. Do it
and write if off when it helps the business.

no one secure in thier own ability latches unconditionally to any
brand. period.

I was 2-3 features and/or bad IQ from buying a D300.. I am loyal to
what does the job best.
------------
Ken - Happy A700 Owner
http://www.cascadephotoworks.com

Click to expand...

Sorry Ken, I wasn't referring to you or anybody specifically. I
spend a lot of time on the 1Ds forum, and when I went to another
professional, Canon-centric site, I was surprised how much more
willing to discuss things people were. I'm fully in support of your
position that RAW is not cooked in the A700...unless one thinks that
Canon and Nikon RAWs are considered cooked, too. We may not know the
recipes, but it seems to me that all of the companies are doin' a
little cookin'. It looks like in the A700's case, it's the analog NR
+ digital NR, and I imagine the digital NR is the one that is
adjustable.

No apology needed,, I meant " I am not surprised" the pros are more open than some on this forum.
------------
Ken - Happy A700 Owner
http://www.cascadephotoworks.com

lonewolf69 · Feb 3, 2008

Ken_5D said:
Really software can't have an effect in how it demosiaces and process
it.. tell me why they same file with both NR and sharpening zeroed
has a finer noise pattern in Bibble then Adobe.. Magic? or did Sony
put a bug in the RAW to make Adobe smooth out the detail..

Kiklop, I'm with Ken on this one, because the Bibble results are simply beautiful...? Maybe it's like Ken says it's magic?

Bottomline, IF Bibble can do this, Adobe should be able to as well, but since it's NOT happening.

One might draw the conclusion that Adobe doesn't care, or that they simply did not put enough effort to optimize the A700 Raw conversion. What's your take, and please do reply with your view of the Bibble files? Please??

Thank you,

--
-Alex

From the minds of Minolta to the imagination of Sony, Alpha, like no other.

http://www.pbase.com/lonewolf69

Just for New Users.. A700 High ISO NR Why Adobe Tests are not valid.

Veteran Member

Veteran Member

Veteran Member

Forum Pro

Forum Pro

Veteran Member

Veteran Member

Veteran Member

Senior Member

Veteran Member

Senior Member

Forum Enthusiast

Senior Member

Forum Pro

Veteran Member

Forum Pro

Forum Enthusiast

Veteran Member

Forum Pro

Senior Member

Similar threads

Keyboard shortcuts