Foveon talk at Stanford.

• He mentioned 1.7x as a good ballpark conversion factor when comparing X3 resolution to Bayer filter based sensors. (The F7 chip used in the Sigma camera has 3.5 million X3 pixels. Thus it weighs in right around 6 million Bayer filtered pixels by this metric. 1.7 is the square root of 3 BTW.)

• The F7 sensor produces images that are sharper than those from the Foveon Studio Camera products. This has to do with optical issues in using existing lenses with the beam splitter prism mechanism. (Or being less charitable, the prism introduces optical artifacts) This certainly explains why they aren't selling that product anymore...

• Sensitivty advantages for the X3 sensor were mentioned. (The point about how Bayer filters throw away 1/3 of the light.) I asked about why then the F7 has lower ISO ratings than Bayer filtered chips of similar pixel pitch. Richard's answer is basicaly that ISO ratings for digital cameras are very arbitrary and he's not sure the F7 will actually be any worse in real world use than existing sensors. The existing chips produce really lousy quality at high sensitivity and Foveon can do that too.

• He said dynamic range will definitely be better than slide film and should approach negative film. (But did not want to say how many bits that is )

• Example shots were shown to demonstrate how the X3 sensor avoids color artifacts and gives much more consistent rendering of detail.

• He talked a bit about workflow and in-camera image processing vs. on a PC vs. in a printer vs. at a photo service. It seems Foveon is working to establish their RAW file format as a standard that is built-in to Windows and MacOS. He denigrated the quality of JPEG images quite a bit. Overall this rang hollow with me. Sure JPEGs aren't the best, but they're quite good and for a lot of photography, workflow ease is more important than absolute quality. A camera that can't produce JPEGs is less useful. He said they understand this and that Foveon equipped cameras will eventually do the whole RAW+JPEG recording thing. (It doesn't have anything to do with the sensor technology really other than that the RAW data is much larger.)

Z (is real) said:

I just got back from a talk by Richard Lyon at Stanford about the
history of photography and Foveon's X3 technology. The history part
was quite interesting though overall I'd hoped to learn more
technical detail about X3.

Fundamentally, all methods of establishing color from an inherently
monochromatic sensing function were used with silver based
photography. Three shot techniques are fairly well known, but there
were also a variety of mosaic filter techniques. He showed a
picture of an "autochrome" color print. These used a fine coating
of three different colors of dyed potato starch on a piece of
glass. One would a expose normal black and white print through the
glass and then develop it and look at the print through the same
"starched" glass.

Ultimately, the development of decent quality three layer film
eliminated all the other techniques. Of course he predicted the
same thing will happen in digital...

Some interesting details:

• He mentioned 1.7x as a good ballpark conversion factor when

comparing X3 resolution to Bayer filter based sensors. (The F7 chip
used in the Sigma camera has 3.5 million X3 pixels. Thus it weighs
in right around 6 million Bayer filtered pixels by this metric. 1.7
is the square root of 3 BTW.)

• The F7 sensor produces images that are sharper than those from

the Foveon Studio Camera products. This has to do with optical
issues in using existing lenses with the beam splitter prism
mechanism. (Or being less charitable, the prism introduces optical
artifacts) This certainly explains why they aren't selling that
product anymore...

• Sensitivty advantages for the X3 sensor were mentioned. (The

point about how Bayer filters throw away 1/3 of the light.) I asked
about why then the F7 has lower ISO ratings than Bayer filtered
chips of similar pixel pitch. Richard's answer is basicaly that ISO
ratings for digital cameras are very arbitrary and he's not sure
the F7 will actually be any worse in real world use than existing
sensors. The existing chips produce really lousy quality at high
sensitivity and Foveon can do that too.

There is a forthcoming standard for measuring sensitivity in
digital cameras which will take into account noise. He didn't want
to comment on where they will end up but expected the F7 to come
out at around 200 in full res mode. (The standard will apparently
quote three ISO ratings, probably at different quality levels. I
expect he was talking about the highest quality.)

(At some point higher ISO is like push processing. A sensor with
much better useable (low noise) dynamic range can be pushed to
deliver quality equivalent to a sensor with lower dynamic range.)

• He said dynamic range will definitely be better than slide film

and should approach negative film. (But did not want to say how
many bits that is )

• Example shots were shown to demonstrate how the X3 sensor avoids

color artifacts and gives much more consistent rendering of detail.

• He talked a bit about workflow and in-camera image processing vs.

on a PC vs. in a printer vs. at a photo service. It seems Foveon is
working to establish their RAW file format as a standard that is
built-in to Windows and MacOS. He denigrated the quality of JPEG
images quite a bit. Overall this rang hollow with me. Sure JPEGs
aren't the best, but they're quite good and for a lot of
photography, workflow ease is more important than absolute quality.
A camera that can't produce JPEGs is less useful. He said they
understand this and that Foveon equipped cameras will eventually do
the whole RAW+JPEG recording thing. (It doesn't have anything to do
with the sensor technology really other than that the RAW data is
much larger.)

Overall, I'd say they have a promising technology and one hopes it
will meet its promise and become widely deployed in a variety of
photographic instruments. However it is obvious that its still
early yet and they are competing against much more mature product
lines. Its going to take a year or two before this is viable in
anything really mass market. Meanwhile, the Sigma SD9 may gain a
following as an excellent studio camera for some...

-Z-

Click to expand...

Terry Sessford said:

Just one small question, your final sentence includes, "the Sigma
SD9 may gain a following as an excellent studio camera for some".
Is there any particular reason why you specified "studio camera"?
I am seriously considering buying an SD9 (but will compare it with
Canon D60 and Nikon D100 before making a final decision), however I
would not be using it in a studio, more as a general purpose camera
but with an emphasis on wildlife photography (from insect and
flower macros through to mammals and birds).

Also, was there any mention of frame-rate in the talk -- the Foveon
specs on their website suggest a maximum 2 fps for the F7 sensor,
however I have been told by Sigma that the SD9 will shoot at 3.5
fps (at maximum resolution).

Click to expand...

Z (is real) said:

I just got back from a talk by Richard Lyon at Stanford about the
history of photography and Foveon's X3 technology. The history part
was quite interesting though overall I'd hoped to learn more
technical detail about X3.

Fundamentally, all methods of establishing color from an inherently
monochromatic sensing function were used with silver based
photography. Three shot techniques are fairly well known, but there
were also a variety of mosaic filter techniques. He showed a
picture of an "autochrome" color print. These used a fine coating
of three different colors of dyed potato starch on a piece of
glass. One would a expose normal black and white print through the
glass and then develop it and look at the print through the same
"starched" glass.

Ultimately, the development of decent quality three layer film
eliminated all the other techniques. Of course he predicted the
same thing will happen in digital...

Some interesting details:

• He mentioned 1.7x as a good ballpark conversion factor when

comparing X3 resolution to Bayer filter based sensors. (The F7 chip
used in the Sigma camera has 3.5 million X3 pixels. Thus it weighs
in right around 6 million Bayer filtered pixels by this metric. 1.7
is the square root of 3 BTW.)

• The F7 sensor produces images that are sharper than those from

the Foveon Studio Camera products. This has to do with optical
issues in using existing lenses with the beam splitter prism
mechanism. (Or being less charitable, the prism introduces optical
artifacts) This certainly explains why they aren't selling that
product anymore...

• Sensitivty advantages for the X3 sensor were mentioned. (The

point about how Bayer filters throw away 1/3 of the light.) I asked
about why then the F7 has lower ISO ratings than Bayer filtered
chips of similar pixel pitch. Richard's answer is basicaly that ISO
ratings for digital cameras are very arbitrary and he's not sure
the F7 will actually be any worse in real world use than existing
sensors. The existing chips produce really lousy quality at high
sensitivity and Foveon can do that too.

There is a forthcoming standard for measuring sensitivity in
digital cameras which will take into account noise. He didn't want
to comment on where they will end up but expected the F7 to come
out at around 200 in full res mode. (The standard will apparently
quote three ISO ratings, probably at different quality levels. I
expect he was talking about the highest quality.)

(At some point higher ISO is like push processing. A sensor with
much better useable (low noise) dynamic range can be pushed to
deliver quality equivalent to a sensor with lower dynamic range.)

• He said dynamic range will definitely be better than slide film

and should approach negative film. (But did not want to say how
many bits that is )

• Example shots were shown to demonstrate how the X3 sensor avoids

color artifacts and gives much more consistent rendering of detail.

• He talked a bit about workflow and in-camera image processing vs.

on a PC vs. in a printer vs. at a photo service. It seems Foveon is
working to establish their RAW file format as a standard that is
built-in to Windows and MacOS. He denigrated the quality of JPEG
images quite a bit. Overall this rang hollow with me. Sure JPEGs
aren't the best, but they're quite good and for a lot of
photography, workflow ease is more important than absolute quality.
A camera that can't produce JPEGs is less useful. He said they
understand this and that Foveon equipped cameras will eventually do
the whole RAW+JPEG recording thing. (It doesn't have anything to do
with the sensor technology really other than that the RAW data is
much larger.)

Overall, I'd say they have a promising technology and one hopes it
will meet its promise and become widely deployed in a variety of
photographic instruments. However it is obvious that its still
early yet and they are competing against much more mature product
lines. Its going to take a year or two before this is viable in
anything really mass market. Meanwhile, the Sigma SD9 may gain a
following as an excellent studio camera for some...

-Z-

Click to expand...

Z (is real) said:

• He mentioned 1.7x as a good ballpark conversion factor when

comparing X3 resolution to Bayer filter based sensors. (The F7 chip
used in the Sigma camera has 3.5 million X3 pixels. Thus it weighs
in right around 6 million Bayer filtered pixels by this metric. 1.7
is the square root of 3 BTW.)

Click to expand...

Adam Tow said:

Thanks for the great recap. How did you hear about the talk at
Stanford? I would have made it had I known about it.

-adam

Z (is real) said:

I just got back from a talk by Richard Lyon at Stanford about the
history of photography and Foveon's X3 technology. The history part
was quite interesting though overall I'd hoped to learn more
technical detail about X3.

Fundamentally, all methods of establishing color from an inherently
monochromatic sensing function were used with silver based
photography. Three shot techniques are fairly well known, but there
were also a variety of mosaic filter techniques. He showed a
picture of an "autochrome" color print. These used a fine coating
of three different colors of dyed potato starch on a piece of
glass. One would a expose normal black and white print through the
glass and then develop it and look at the print through the same
"starched" glass.

Ultimately, the development of decent quality three layer film
eliminated all the other techniques. Of course he predicted the
same thing will happen in digital...

Some interesting details:

• He mentioned 1.7x as a good ballpark conversion factor when

comparing X3 resolution to Bayer filter based sensors. (The F7 chip
used in the Sigma camera has 3.5 million X3 pixels. Thus it weighs
in right around 6 million Bayer filtered pixels by this metric. 1.7
is the square root of 3 BTW.)

• The F7 sensor produces images that are sharper than those from

the Foveon Studio Camera products. This has to do with optical
issues in using existing lenses with the beam splitter prism
mechanism. (Or being less charitable, the prism introduces optical
artifacts) This certainly explains why they aren't selling that
product anymore...

• Sensitivty advantages for the X3 sensor were mentioned. (The

point about how Bayer filters throw away 1/3 of the light.) I asked
about why then the F7 has lower ISO ratings than Bayer filtered
chips of similar pixel pitch. Richard's answer is basicaly that ISO
ratings for digital cameras are very arbitrary and he's not sure
the F7 will actually be any worse in real world use than existing
sensors. The existing chips produce really lousy quality at high
sensitivity and Foveon can do that too.

There is a forthcoming standard for measuring sensitivity in
digital cameras which will take into account noise. He didn't want
to comment on where they will end up but expected the F7 to come
out at around 200 in full res mode. (The standard will apparently
quote three ISO ratings, probably at different quality levels. I
expect he was talking about the highest quality.)

(At some point higher ISO is like push processing. A sensor with
much better useable (low noise) dynamic range can be pushed to
deliver quality equivalent to a sensor with lower dynamic range.)

• He said dynamic range will definitely be better than slide film

and should approach negative film. (But did not want to say how
many bits that is )

• Example shots were shown to demonstrate how the X3 sensor avoids

color artifacts and gives much more consistent rendering of detail.

• He talked a bit about workflow and in-camera image processing vs.

on a PC vs. in a printer vs. at a photo service. It seems Foveon is
working to establish their RAW file format as a standard that is
built-in to Windows and MacOS. He denigrated the quality of JPEG
images quite a bit. Overall this rang hollow with me. Sure JPEGs
aren't the best, but they're quite good and for a lot of
photography, workflow ease is more important than absolute quality.
A camera that can't produce JPEGs is less useful. He said they
understand this and that Foveon equipped cameras will eventually do
the whole RAW+JPEG recording thing. (It doesn't have anything to do
with the sensor technology really other than that the RAW data is
much larger.)

Overall, I'd say they have a promising technology and one hopes it
will meet its promise and become widely deployed in a variety of
photographic instruments. However it is obvious that its still
early yet and they are competing against much more mature product
lines. Its going to take a year or two before this is viable in
anything really mass market. Meanwhile, the Sigma SD9 may gain a
following as an excellent studio camera for some...

-Z-

Click to expand...

--
Adam Tow
tow.com
Tomorrow begins today.

Click to expand...