Digital BW, The Print

> (There are about 50% more green sensors than those filtered for red
> > 
> > and blue, once again to provide more luma information.)
> 
> THERE ARE TWICE AS MANY GREEN SENSOR ON A NON-FOVEON BAYER ARRAY SENSOR, 
NOT 
> HALF AS MANY. 

Oh my goodness. Please read more carefully. 50% (ahem) MORE I said, not less. I'm 
-talking-about the bayer type.

> > Luma is just brightness. It has nothing to do with sharpness. ALL the
> > sensors are responding to luma, but each in only one portion of the
> > spectrum.
> 
> 
> HAS EVERYTHING TO SO WITH MODULATION THEREOF BY THE LENS USING DIFFERENT 
> COLOR ILLUMINANCE AND REFLECTANCE, SO IT HAS A GREAT DEAL TO DO WITH IT.


I'll confess I have not got a clue what you're saying here. I'd be pleased to learn how 
-sharpness- at any single given photo-site even exists, much less what a colored filter 
above the site has to do with it. All that is happening is that some portion of the spectrum 
is filtered out as the light-level is converted to electricity by the sensor. True, a full 
spectrum appears "brighter" to the human eye (and the sensor) than a single color will, but 
it's still just a luma value that is being recorded... And I'll continue to maintain that 
brightness is not sharpness.
> 
> > 
> > AntiAliasing is a "cure" to color issues, and has nothing (much) to do
> > with luma. The anti-aliasing filter may be physical (which can lend to
> > softness issues that can be severe, and easily discernable, or it
> > calculated, which produces information depending on how well done the
> > algorythms are, and the layout of the sensors.)
> > 
> > The need for anti-aliasing comes from trying to deal with a
> > sensor-group which has two different colors meeting in a sharp line
> > across it. 
> 
> 
> ANTI ALIASING FILTERS HAVE NO COLOR BIAS, THEY ARE SIMPLY A LOWPASS FILTER 
> AND ARE ABSOUTELY CALCULATED IN CONCERT WITH COLOR SMEARING ALGORTITHMS 
IN THE 
> DIGITAL SIGNAL PROCESSORS OF THE SYSTEM. THE SYMPTOMS OF THEIR ABSENCE ON 
BAYER 
> FILTER ARRAYS ARE VISUALLY MANIFESTED IN COLOR, BUT THEY ARE NOT A COLOR 
> ISSUE PER SE.

Exactly what I said: sensors are strictly luma information. They do not respond to color, 
except as determined by the RGB filter over them.

> 

> THE FOVEON ONLY SHOOTS RAW 

all cameras "shoot raw"....whether record the file RAW or allow further in-camera 
processing is an option with other cameras, but not the Foveon.

AND HAS EVERYTHING TO DO WITH SINGLE PIXEL 
> SHARPNESS. WHERE DID YOU GET THIS BS WHEN TALKING ABOUT BAYER ARRAYS? WHAT 
IS 
> PERFECTLY SHARP? 1/4 OF THE ARRAY PIXEL COUNT? WHAT PICTCH? WHAT FILTER?

Please read carefully.... my entire -point- is that without the use of antil-aliasing, 
sharpness remains. "if you shoot RAW, you'll get perfectly sharp information..." Please see 
comments below.

And, as I noted, I was trying to figure that out that 'sharpness' phrase having been used in 
a previous post. Sharpness, I suggest, has to do with resolution. To take a "reducto 
absurdum" example, imaging a photo of a human face made with 10 pixels vs one made 
with 6 million pixels.
> 
> > 
> > However, all that said, simply take a RAW camera image (say 3024 x
> > 2024, as it comes from the S2 pro) and select split channels in Photoshop.
> > 
> > You'll get three images with different ranges of gray tones, depending
> > on the range implicit in the color filter. Each image will be some
> > range of "black and white" and each image will be exactly 3024 x 2024,
> > and contain just over 6 million pixels... not 1.5 million. 
> 
> 
> THIS IS AFTER FUJI DOES THEIR SQUARE ROOT OF TWO, OCTAGONAL TO ORTHOGONAL 
> VOODOO MATH, SO IT'S PROCESSED INFORMATION WHICH IS 5/6 INTERPOLATED AND 
NOT 
> NATIVE DATA.

As far as I know, (and at least on the Fuji), this is not true. The raw camera luma values 
from the sensor sites is saved. I believe that color interpolation only occurs if one chooses 
any of the other non-raw formats for saving (such as TIFF, or JPEG).

> 
> YOU ARE STILL TALKING A PROCESSED IMAGE NOT RAW, UNLESS IT''S FOVEON WHICH IS 
> RAW. BTW, I HAVE OWNED 4 DIFFERENT FOVEONS CAMERAS FOR 5 YEARS AND PRINTED 
> MANY IMAGES FROM THEM, B&W AND COLOR.
> 
> > 
> > Take the color information out of a 3.5 meg photo, blend the results,
> > and you'll still have 3.5 megs of resolution.
> 
> 
> COMPARED TO WHAT? NOT ALL PIXELS ARE CREATED EQUAL AND FOVEON PIXELS ARE 
> ROUGHLY THE SAME AS TWICE AS MANY BAYER PIXELS IN THE FILE.

I'm afraid you're confusing resolution with color information. A Pixel is a single picture 
element, comprised of r,g,& b information in three channels. The Forveon advantage is 
that it directly reads all that information from a single photo sensor site, while a Bayer 
array interpolates the site information across approximately 2/3 of the sensor locations 
when the information is processed. It needs to be processed, of course, because the color 
filters are side by side instead of stacked. The Foveon method is vastly superior to Bayer 
on a photosite by photosite basis, since there is no need for interpolation. But a PIXEL is a 
pixel is a pixel.  It's only when you start putting anti-aliased pixels side by side, in groups 
larger than one, that the problems with false colors begins to show up.

> 
> > 
> > Remember, we're talking B&W here, not color, which is what
> > anti-aliasing is used for...
> 
> 
> THE SYMPTOMS OF ALIASING SHOW UP IN COLOR THE MOST, BUT AA FILTERS ARE NOT 
AS 
> COLOR DEPENDENT AS YOU SUGGEST. THE COLOR DEPENDENCY OF THE 
> LUMINANCE/ILLUMINANCE IS SMALL FRACTION OF THEIR TOTAL EFFECT.

You may be correct, but that is not how I understand it. The luma information in a file is 
approximated 1000 times greater than the color information. However, we may be talking 
with different words here again.

> 
> > 
> > And it is impossible for there to be such a thing as an "interpolated'
> > RAW file: either it is RAW info, or it is interpolated info, but it
> > cannot be both, by definition. RAW is what the CCD photo sensors put
> > out: period. (I understand the confusion, the tyranny of words, that
> > can arise, since the Foveon only puts out a "raw" file, but to keep
> > the terminology correct, once that file is interpolated, it is no
> > longer "raw" in the sense normally used when speaking of digital
> > camera RAW files.
> 
> 
> ARE YOU TALKING ABOUT COLOR INTERPOLATION? FOVEON DOES NOT COLOR 
INTERPOLATE, 
> EVER. TYRANNY INDEED.

"interpolated raw" was not my phrase: I was responding to what was said in another post 
and I was correcting it to reflect what you're saying.

So, please, please -read what I said-. That is -exactly- what I said. The Foveon does not 
need interpolation. 

> 
> > 
> > Yes, the Foveon is great, particularly for color. Excellent stuff. But
> > there is a trade-off on sheer resolution vs the lack of need for
> > anti-aliasing. Bottom line, for color, is how you like the result.
> > It's the photo, stupid! :-)
> 
> 
> THIS IS THE ONLY TRUE STATEMENT YOU HAVE MADE SO FAR. JUST LUCKY I GUESS.

An unnecessary dig when all that is apparently happening is a failure to understand the 
terminology used (file size versus sensor sites).

> 
> > 
> > But for B&W, if using RAW files, then there is no anti-aliasing, and
> > no matter how you slice the pie, 3.5 megs is less information than 6.1. 
> 
> 
> FOVEON RAW FILES HAVE HISTORICALLY BEEN 12 MEGAPIXELS, 16 MEGAPIXELS (ON 
> THEIR MONOCHROME SENSOR, NEVER PRODUCED) OR 10.5 MEGAPIXELS ON THEIR 
CURRENT X3 
> CHIPS. THEY HAVE NEVER BEEN 3.5 IN RAW. THEY ARE 4.5 ON THEIR SMALLEST CHIP, 
NOT 
> YET IN PRODUCTION CAMERAS, BUT FORTHCOMING IN POLAROID CAMERAS, IT WOULD 
> SEEM.

Interesting. In their ad for the SD10, appearing in Shutterbug, page 21, Jan 05 , they say 
"3.4 red, 3.4 blue, 3.4 green". That would be, as I understand their technology, a CCD 
array of 3.4 million sensors sites, with pure red, green and blue data from each one.

The point you are missing is that I was describing sensor sites, receptors, since we were 
discussing how the Bayer interpolation works, which is only meaningful if you're talking 
about sensor sites. The Foveon has 3.4 million. Other, non- stacked arrays have 6, 
8,12 16, or (as with Leafs) 22 million or more sensor sites.)

My only point is that the Foveon 3.4 (I misspoke when I said 3.5) sensor sites to work with. 
That is less than 6 million sensor sites. 

Sounds to me like you're confusing file size with pixel resolution. A Pixel, "picture 
element" is the smallest 'dot' you'll see on the screen. It is composted of R, B and B 
information.

Your camera has sensor sites. The Fuji, for example has 6.1 million sensor sites. The 
Foveon has 3.4 million sensor sites. That is the native resolution of the camera.

 I was pointing out that 6 million pixels (even if you're not a big fan of how the color in 
each site was arrived at) is a higher resolution than 3.4, and that there is trade-off 
between what the human eye will see in a photo made of 3.4 million pixels, and one made 
with 6, 8, or 12 million pixels.

I apologize if I was not clear enough in separating out the discussion as to sensor sites/
pixels and file size. I assumed that since we were discussing anti-aliasing, and "sharpness" 
that we all understood we were talking about pixels and not file size.

You are perfectly correct that the Foveon puts out 3.4 million bits of raw, uninterpolated 
information for red, green and blue. A 6 megapixel camera directly reads 1.5 megs of 
"pure" red data, and interpolates the remaining 4.5 million megapixels; ditto for blue, and 
3 megapixels of green with the other 3 interpolated.

All added up, that is 6 megabytes of information in 6 megaPIXELS .

Add up all the Foveon colors and you'll get 10.2 megabytes of information, in 3.4 
megaPIXELS.

I should also note that I've got back to the technical documents, and must correct myself: 
the CCD in the Fuji -does- have a low pass filter on it (along with an infra-red).

I am sorry that this exchange deteriorated to such an extent.

May I only suggest that one -carefully- read posts, as I'm sure that doing so, and using 
words with mutually agreed-upon meanings will improve the quality of communications.

Cordially

Tracy