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

Were the samples you have on Pbase (D300 vs A700):
1) Shot w/the latest software version in the A700?
2) If not - have you re-shot them?

Thanks, Bob
--
Bob
 
Were the samples you have on Pbase (D300 vs A700):
1) Shot w/the latest software version in the A700?
2) If not - have you re-shot them?
I'll have to check. The originals were definitely not with the v2 software. However, I did a second test, which may be v2 or not - I still have the setup more or less in place, and I could repeat the test yet again using v3. It might not be exactly the same as the lights have been moved a few times, but I can rebuild a nearly identical framed image.

David

--
Publishing & Editing Photoworld (photoclubalpha.com) and Master Photo Digital
Currently writing for f2 and the BJP
 
I just downloaded the marketing pdf for the A700..

I quote....

Exmor™ + BIONZ™ multi-stage
noise reduction

"The A700 uses advanced technologies at every stage
of image acquisition and processing to reduce
electrical signal noise that can degrade image quality,
especially with long exposures or at high ISO
sensitivity settings. During shooting, the Exmor™ CMOS
sensor’s on-chip circuitry suppresses noise both
before and after analogue-to-digital signal conversion"

"Within the BIONZ™ image processing engine, noise
reduction is also applied during both RAW data
conversion and image processing to ensure an
absolute minimum of noise in the fi nal image data.
In addition, high ISO noise reduction is automatically
applied when shooting at ISO 1600 or more"

Thats good enough for me!

--



Clint is on holiday! Soon to return! ;-)
 
Barry (That's good enough for me...),

You still haven't yet answered my direct question to you, regarding Bibble and the A700.

Do you still have a problem with the images, raw files, etc, now with the results that we're seeing with Bibble? I really want to understand if this is a stand on principal with you, despite the nice/decent results (no smearing/splotches) we're getting with Bibble?

-Alex

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

http://www.pbase.com/lonewolf69
 
To me, that's all that matters. The final image, not how I got there. If Bibble gives the image I want, I don't really care how Sony processes the data for RAW.
--

'Dodging and burning are steps to take care of mistakes God made in establishing tonal relationships.'
 
I just downloaded the marketing pdf for the A700..

I quote....

Exmor™ + BIONZ™ multi-stage
noise reduction

"The A700 uses advanced technologies at every stage
of image acquisition and processing to reduce
electrical signal noise that can degrade image quality,
especially with long exposures or at high ISO
sensitivity settings. During shooting, the Exmor™ CMOS
sensor’s on-chip circuitry suppresses noise both
before and after analogue-to-digital signal conversion"

"Within the BIONZ™ image processing engine, noise
reduction is also applied during both RAW data
conversion and image processing to ensure an
absolute minimum of noise in the fi nal image data.
In addition, high ISO noise reduction is automatically
applied when shooting at ISO 1600 or more"

Thats good enough for me!
This is what has been being said. The key wording is "conversion" when RAW is converter to JPG it has more NR applied , this is why often the JPG is better than the RAW because some RAW converters suck comparied to Sony's processor. Then the JPG gets a final pass of image processing as a jpg.

This is what Mark told me and what Walt has been telling you for a long time. I am glad you finally found a source you trust.

-
------------
Ken - Happy A700 Owner
http://www.cascadephotoworks.com
 
If you look closely at the images you posted you will see that each one of the files contains the EXACT same noise splotches in the EXACT same places. The only difference is that it looks like Bible is sharpening the images more than Adobe and also there is an obvious difference in color correction between the two.

If you look at the numbers 1 and 7 in the number MCA217 on the bottle and if you examine the tips of both of these numbers both of the numbers have a big red splotch of noise. It is there in every image from both adobe and bible, I have circled it in red in all three images so you can see the area I am talking about.

You have to realize that no matter what RAW converter you are using it is still processing the exact same RAW file pixel for pixel. If the RAW file has a splotch of red noise that is 4 pixels big at location 120,232 that same splotch is going to be there no matter what RAW converter you use on that file. It may no longer be red but the splotch is going to be there no matter what. So if and I say if the RAW file is cooked it is cooked no matter what you process it with, the detail is still not there. I am not saying that I don't think the final Bible image is more visually appealing what I am saying is that file contains the exact same amount of detail "maybe less because the color noise is gone" as the non color reduced bible image and the adobe image.

I have said this numerous times in the past, I hope that you can see and uderstand what I and others have been talking about from looking at these images and my explanation. Everytime in the past that I try to explain this I get bashed for being a Sony hater bla bla bla. I am not a Sony hater but I am tired of everyone being told they are wrong when they try to explain this.

First the adobe image



Then the Bible image



Now the NR bible image. The splotch is still there it just contains no color now.


Below is the proof. Both are from the SAME A700 RAW file ISO 3200.
With defau;t settings, and sharpening, and NR settings set to ZERO.
Adobe applies way more NR without user option to change than Sony
ever has. Still I see little outrage from some posters nor from Phil
on this.

Before you look at these. These shots are made to SHOW nose
patterns. These are not how good the camera takes images. This is to
show how much detail and noise at high ISO Sony still leaves to user
choice if the user is smart enough to choose a good RAW tool.

This is Adobe.. and what Phil and others think is OK to use for
testing cameras.

This is bibble which leaves the choices on noise to the user.

And for those who for artistic reasons like grain this is an image
processed to remove color noise leave the REAL Sony Luminance grain
pattern in.. It has been sharpened and had only color NR applied..
all grain you see is from the A700 none added by software. This is
Bibble...

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

------------
Ken - Happy A700 Owner
http://www.cascadephotoworks.com
--
Thanks,
Chadd
http://www.pbase.com/chadd
 
The grand-standing needs to stop.

1. At ISO 1600 and above Sony is doing something to the RAW data.
2. Different RAW readers/processors yield different results.

A. Exmor does something that seems harmless.
B. Bionz does something that we would like to have a on/off switch for.
I just downloaded the marketing pdf for the A700..

I quote....

Exmor™ + BIONZ™ multi-stage
noise reduction

"The A700 uses advanced technologies at every stage
of image acquisition and processing to reduce
electrical signal noise that can degrade image quality,
especially with long exposures or at high ISO
sensitivity settings. During shooting, the Exmor™ CMOS
sensor’s on-chip circuitry suppresses noise both
before and after analogue-to-digital signal conversion"

"Within the BIONZ™ image processing engine, noise
reduction is also applied during both RAW data
conversion and image processing to ensure an
absolute minimum of noise in the fi nal image data.
In addition, high ISO noise reduction is automatically
applied when shooting at ISO 1600 or more"

Thats good enough for me!
This is what has been being said. The key wording is "conversion"
when RAW is converter to JPG it has more NR applied , this is why
often the JPG is better than the RAW because some RAW converters suck
comparied to Sony's processor. Then the JPG gets a final pass of
image processing as a jpg.

This is what Mark told me and what Walt has been telling you for a
long time. I am glad you finally found a source you trust.

-
------------
Ken - Happy A700 Owner
http://www.cascadephotoworks.com
--
Direct your eye right inward, and you'll find a thousand
regions in your mind Yet undiscovered. Travel them, and
be Expert in home-cosmography.
-H.D. Thoreau
 
If you look closely at the images you posted you will see that each
one of the files contains the EXACT same noise splotches in the EXACT
same places. The only difference is that it looks like Bible is
sharpening the images more than Adobe and also there is an obvious
difference in color correction between the two.

If you look at the numbers 1 and 7 in the number MCA217 on the bottle
and if you examine the tips of both of these numbers both of the
numbers have a big red splotch of noise. It is there in every image
from both adobe and bible, I have circled it in red in all three
images so you can see the area I am talking about.

You have to realize that no matter what RAW converter you are using
it is still processing the exact same RAW file pixel for pixel. If
the RAW file has a splotch of red noise that is 4 pixels big at
location 120,232 that same splotch is going to be there no matter
what RAW converter you use on that file. It may no longer be red but
the splotch is going to be there no matter what. So if and I say if
the RAW file is cooked it is cooked no matter what you process it
with, the detail is still not there. I am not saying that I don't
think the final Bible image is more visually appealing what I am
saying is that file contains the exact same amount of detail "maybe
less because the color noise is gone" as the non color reduced bible
image and the adobe image.

I have said this numerous times in the past, I hope that you can see
and uderstand what I and others have been talking about from looking
at these images and my explanation. Everytime in the past that I try
to explain this I get bashed for being a Sony hater bla bla bla. I am
not a Sony hater but I am tired of everyone being told they are wrong
when they try to explain this.
There are two types noise color which is usually seen as red and blue noise and

and luminance which is were the grain effect is. Color comes out of the deomosaicing process where color is created from near by pixels and luminance comes from the lluminance level data is is unique to each pixel location.

No RAW converter is going to change the location of the noise. But they can either pass it through or supress it.

Adobe is supressing the luminance noise before the user gets a choice in the matter. Look at the watch subdials.. try as I might..with LR .. bibble always gave me more of the concentric circle pattern because it doesn't try to guess at what fine non color detail is noise.. it just takes the data. The Adobe takes the finer luminance noise out of the image and also some of the detail.

So what you see with adobe is the color noise is soft blobs and with Bibble you see the same color areas but with a finer luminance noise pattern too.

Adobe is the worst at doing this. Sony's IDC is a bit better.. everyone else I have tried with NR zeroed shows the finer Noise pattern not just the blobs.

This is where water color happens.. when Adobe smoothes out the smaller noise pattern and when you start to sharpen it sees the color blobs as detail and puts and edge around the blob.
First the adobe image



Then the Bible image



Now the NR bible image. The splotch is still there it just contains
no color now.


Below is the proof. Both are from the SAME A700 RAW file ISO 3200.
With defau;t settings, and sharpening, and NR settings set to ZERO.
Adobe applies way more NR without user option to change than Sony
ever has. Still I see little outrage from some posters nor from Phil
on this.

Before you look at these. These shots are made to SHOW nose
patterns. These are not how good the camera takes images. This is to
show how much detail and noise at high ISO Sony still leaves to user
choice if the user is smart enough to choose a good RAW tool.

This is Adobe.. and what Phil and others think is OK to use for
testing cameras.

This is bibble which leaves the choices on noise to the user.

And for those who for artistic reasons like grain this is an image
processed to remove color noise leave the REAL Sony Luminance grain
pattern in.. It has been sharpened and had only color NR applied..
all grain you see is from the A700 none added by software. This is
Bibble...

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

------------
Ken - Happy A700 Owner
http://www.cascadephotoworks.com
--
Thanks,
Chadd
http://www.pbase.com/chadd
--
------------
Ken - Happy A700 Owner
http://www.cascadephotoworks.com
 
Spent sometime last night going over them at all ISO's - then it occurred to me that I may have been reviewing the original firmware version in relation to the D300.

Based on forum input, the V2 (new) firmware makes a "considerable" (my definition of what I think I'm hearing :-) difference in A700 Noise.

Thanks again, Bob

--
Bob
 
B. Bionz does something that we would like to have a on/off switch for.
I got the impression from your talk with Mark that the Bionz was NOT affecting the RAW data stored as RAW, but the RAW data stored as JPEG. Did I miss something??
 
Barry (That's good enough for me...),

You still haven't yet answered my direct question to you, regarding
Bibble and the A700.

Do you still have a problem with the images, raw files, etc, now with
the results that we're seeing with Bibble? I really want to
understand if this is a stand on principal with you, despite the
nice/decent results (no smearing/splotches) we're getting with Bibble?
Sure ACR doesnt do a great job on A700 files..no worries there. Buble does a better job, buts its more down to sharpening artifacts, than a finer grain pattern.

Its still not what I am looking for..I want that "Off" and pretty bad..no problems with an A700 with NR off.

--



Clint is on holiday! Soon to return! ;-)
 
Taken from their A700 marketing pdf..







--



Clint is on holiday! Soon to return! ;-)
 
B. Bionz does something that we would like to have a on/off switch for.
I got the impression from your talk with Mark that the Bionz was NOT
affecting the RAW data stored as RAW, but the RAW data stored as
JPEG. Did I miss something??
As I understand it and as the literature says in obscure markting talk.

The Exmore processor has
Noise Avoidance (short circuote paths etc)
CMOS Noise canceling and NR before and after the on Sensor AD.
That Data is the raw data.
If you shoot RAW.. that gets written to a file.

If you shoot jpg the RAW data gets another pass of NR on the full 12 bits of RAW data before it is converted to jpg and then gets one file pass. Then buffer raw is discarded.

The goal of this appears to be to do the NR at each appropriate stage which may contribute to the color accuracy etc. VS just doing it after it making a JPG.

Mark specifically mentioned on Sensor NR needs to be firmware addressable and that the Bionz doesn't change the RAW file.

In the beginning there was CCD and CMOS. CMOS was more noisy so most cameras used CCD. Then Canon added on sensor noise handeling and that removed the noise and grain of higher ISO and it was good. Then on the next Day.. Sony added more Noise control and it was good except for those crying in the dessert for more grain...

Then we found the grain was still there.. different but there .. but the evil Adobe had been hiding it from the eyes of the viewers.. And many said it was good, but still some cried in the desert because they only accept one way of dealing with noise... and they would not see a new way.

That is the story so far.
------------
Ken - Happy A700 Owner
http://www.cascadephotoworks.com
 
Chadd,

But to the naked eye, it looks DAMN good...??!? So I don't understand what the problem is? :(

You basically had to circle it for ME to see it and honestly even then I still had to look very carefully and double check just to see it. Regardless of what process Bibble is user (Sharpening Noise, NR Artifacting, etc...)

Bibble is doing an amazing job, touch up, repair, post processing you name it, it's the BEST image out of the three and is very acceptable.

In a nutshell, no matter what you call it, Bibble is making the HIGH ISO A700 RAW images look good.

Chadd, Barry, IMHO Bibble is giving the best possible results, whether or NOT the NR can be totally turned off at this time is irrelevant as the end image converted from SONY A700 is very usable. Sure the D300 and E3 and other cameras may have slightly better results but USING Bibble the IQ difference in HIGH ISO RAW conversions between Olympus, NIKON, CANON and SONY has been greatly reduced and it SHOWS that the A700 can certainly hold it's own against the competition without compromise...

--
-Alex

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

http://www.pbase.com/lonewolf69
 
Taken from their A700 marketing pdf..





Keep posting this its good stuff.

This is key.. Mark wanted to reference this very diagram in talking about why the Bionz is not working on the raw file.

This diagram is of the process from sensor to jpg. As it says an I know you can read.. NR is applied to raw data as part of "conversion." It never says it is applied as the file is written. There is no conversion in writing the RAW file it doesn't need conversion its RAW already.

Is the language obscure enough that I could be wrong.. could be.. but as of right now this doesn't say anything about RAW data written to the file. And Mark said that was not affected by the Bionz...

he was also pretty annoyed at Phil making guesses about a system he doesn't understand. Which again had to so with Phil ASSuming he knows the RAW file flow.
--



Clint is on holiday! Soon to return! ;-)
--
------------
Ken - Happy A700 Owner
http://www.cascadephotoworks.com
 
Ken with all due respect what you are saying is just not true.

Here is a copy of your Adobe LR image that I edited in CS2 all I did was reduce color noise "slightly" it still has some color noise left, adjusted contrast, sharpen the image and then correct the color balance.

As you can see there is no notable difference between this image and your bible image that you say is processing the luminance noise differently. You also have to realize that I edited this from your jpeg so there is clipping happening on my edited image.

Here is your Adobe LR image that I edited in CS2



and here is your bible image.



They are virtually identical. Even though I edited a jpeg that is being clipped. Explain to me how this is possble if Adobe is creating the water color issues?
There are two types noise color which is usually seen as red and blue
noise and
and luminance which is were the grain effect is. Color comes out of
the deomosaicing process where color is created from near by pixels
and luminance comes from the lluminance level data is is unique to
each pixel location.
No RAW converter is going to change the location of the noise. But
they can either pass it through or supress it.

Adobe is supressing the luminance noise before the user gets a choice
in the matter. Look at the watch subdials.. try as I might..with LR
.. bibble always gave me more of the concentric circle pattern
because it doesn't try to guess at what fine non color detail is
noise.. it just takes the data. The Adobe takes the finer luminance
noise out of the image and also some of the detail.

So what you see with adobe is the color noise is soft blobs and with
Bibble you see the same color areas but with a finer luminance noise
pattern too.

Adobe is the worst at doing this. Sony's IDC is a bit better..
everyone else I have tried with NR zeroed shows the finer Noise
pattern not just the blobs.

This is where water color happens.. when Adobe smoothes out the
smaller noise pattern and when you start to sharpen it sees the color
blobs as detail and puts and edge around the blob.

------------
Ken - Happy A700 Owner
http://www.cascadephotoworks.com
--
Thanks,
Chadd
http://www.pbase.com/chadd
 
My argument is not that the bible doesn't produce a better looking image than Adobe straight from the RAW conversion, it is that Adobe is not mucking up the RAW conversion as Ken is implying, look at the two images I just posted in response to Ken. Adobe is not throwing away noise before the user has a change to adjust it as they want. The detail and noise in both the bible images and LR images are exactly the same.

My point still stands that "if the RAW files are cooked then it is the A700 that is doing it not your raw converter".

Now do you guys see why I wish this whole thing would just be dropped? We are never going to come to a conclusion on it.
Chadd,

But to the naked eye, it looks DAMN good...??!? So I don't
understand what the problem is? :(

You basically had to circle it for ME to see it and honestly even
then I still had to look very carefully and double check just to see
it. Regardless of what process Bibble is user (Sharpening Noise, NR
Artifacting, etc...)

Bibble is doing an amazing job, touch up, repair, post processing you
name it, it's the BEST image out of the three and is very acceptable.

In a nutshell, no matter what you call it, Bibble is making the HIGH
ISO A700 RAW images look good.

Chadd, Barry, IMHO Bibble is giving the best possible results,
whether or NOT the NR can be totally turned off at this time is
irrelevant as the end image converted from SONY A700 is very usable.
Sure the D300 and E3 and other cameras may have slightly better
results but USING Bibble the IQ difference in HIGH ISO RAW
conversions between Olympus, NIKON, CANON and SONY has been greatly
reduced and it SHOWS that the A700 can certainly hold it's own
against the competition without compromise...

--
-Alex

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

http://www.pbase.com/lonewolf69
--
Thanks,
Chadd
http://www.pbase.com/chadd
 
Ken with all due respect what you are saying is just not true.

Here is a copy of your Adobe LR image that I edited in CS2 all I did
was reduce color noise "slightly" it still has some color noise left,
adjusted contrast, sharpen the image and then correct the color
balance.

As you can see there is no notable difference between this image and
your bible image that you say is processing the luminance noise
differently. You also have to realize that I edited this from your
jpeg so there is clipping happening on my edited image.
First of all they are not virtually the same..

If you look at the one you processes and and sharpened in CS2.. Adobe is placing sharpening artifacts along the edges of the color blobs with almost no "grain" artifacts in the blobs.. this irregular pattern is already starting a watercolor in the gray table top area at the bottom of an image. as the sharpening artifacts wind between areas of smoothed color blobs.

The bible image has noise that I did sharpen for this "show Barry the grain" version. and it is more uniform because Bibble didn't start with smoothing out the luminance noise..

You did get PS to to pull up the watch fact detail well with the heavy sharpening.. I will note that

bible has that in the RAW file before sharpening unlike LR so I don't need to apply as much sharpening to recover the smoothed out luminance data.

IE in PS you can recreate a lot of things with work including making something look like Grainy 400 ISO ektachrome.

The difference being that Bibble has more detail from the base line so needed less sharpening to do what you did in PS to create sharpening errors/artifacts along the color blob edges.
Here is your Adobe LR image that I edited in CS2



and here is your bible image.
I linked back to the orginal so no Pbase processing if any was included because for this test the small text patches you circled are not where the key data is.


They are virtually identical. Even though I edited a jpeg that is
being clipped. Explain to me how this is possble if Adobe is creating
the water color issues?
There are two types noise color which is usually seen as red and blue
noise and
and luminance which is were the grain effect is. Color comes out of
the deomosaicing process where color is created from near by pixels
and luminance comes from the lluminance level data is is unique to
each pixel location.
No RAW converter is going to change the location of the noise. But
they can either pass it through or supress it.

Adobe is supressing the luminance noise before the user gets a choice
in the matter. Look at the watch subdials.. try as I might..with LR
.. bibble always gave me more of the concentric circle pattern
because it doesn't try to guess at what fine non color detail is
noise.. it just takes the data. The Adobe takes the finer luminance
noise out of the image and also some of the detail.

So what you see with adobe is the color noise is soft blobs and with
Bibble you see the same color areas but with a finer luminance noise
pattern too.

Adobe is the worst at doing this. Sony's IDC is a bit better..
everyone else I have tried with NR zeroed shows the finer Noise
pattern not just the blobs.

This is where water color happens.. when Adobe smoothes out the
smaller noise pattern and when you start to sharpen it sees the color
blobs as detail and puts and edge around the blob.

------------
Ken - Happy A700 Owner
http://www.cascadephotoworks.com
--
Thanks,
Chadd
http://www.pbase.com/chadd
--
------------
Ken - Happy A700 Owner
http://www.cascadephotoworks.com
 

Keyboard shortcuts

Back
Top