SD9 + Foveon = Problems

harry wrote:
The

Member said:

Foveon technology IS here, not a pipe dream.

Click to expand...

Harry,

Have YOU actually touched this camera? Do you know ANY store who has taken money from a customer and given him this camera?

Maybe you should learn the difference between Vaporware and the real McCoy: Apple makes a computer with 2 G4 1.25mhz processors. That is a real McCoy. Apple users want a G5 processor. That is vaporware! Until you, as a consumer, can put your hands on the item, it is a pipe dream.

Have you ever heard of Irvine Sensors? They were trying to make a product called Digital Film. It was to fit into lmost any 35mm film camera and give you 1280x960 res for around $1300. When they first annouced the idea it would have been a good deal. It was high res for the times and it would work just like your own SLR. Well, they kept pushing the ship date off and before you knew it, 1.3 megapixel was low and $1300 was too high. The Irvine sensors went bankrupt before it shipped.

Will the same happen to Foveon?

hoang nguyen said:

I only have one question. Did it take a company like Foveon to
make a cool sensor X3 for Canon and Nikon and the likes to Suddenly
introduce "superior technologies" why haven't they done so until
now?

Click to expand...

Foveon isnt actually new. It is based off a 20 yrs old Kodak patent that couldnt be done back then. Foveon is using NSC's 0.18 um mfg'er process for their chip, and even your Canon CMOS is still being produced with the 0.5 um process.

--
jc
Sony F707
http://www.reefkeepers.org/gallery/f707
http://www.reeftec.com/gallery

G. Lentz said:

Most probably think in terms of what resolution means relative to
what they see when they look at their computer monitor. If we want
to play that game, then I could say that my computer monitor
resolution is really 3072x2304, if you count each Red, Green, and
Blue pixel component seperately.

Click to expand...

No, because if they used your "logic" they'd claim that a 6MP CFA sensor using a typical Bayer-algorithm is actually a 54MP plus sensor. They don't.

Every pixel in a Foveon sensor contributes it's luminance and color data one-for-one to the output. No sampling. So 3.54M x 3 (stacked RGB) = 10.62 MPS (photo-sites) in, 3.54MP out.

Every pixel in a CFA sensor contributes it's luminance data one-for-one to the output. As the eye is more sensitive to luminance, this contributes more to perceived resolution. Every pixel in a typical bayer algorithm determines it's color value by sampling itself and it's adjacent neighbors.

So in effect, 6M x (1 (source) + 8 (adjacents)) = 54MPS in, 6MP out. Advanced algorithms will sample additional h, v, and diagonal pixels for better edge detection.

If you're interesting in learning the facts, I can can point you to some sites that describe the algorithms and processes I've described.

Summary:

Is the Bayer system a hack? Yes.
Is it a very elegant hack that works exceedingly well? Yes.
Are companies misrepresenting pixel counts? No. (Maybe Fuji)
Will the Foveon deliver more accurate color information? Yes.
Do I wish the 1DS was an 11MP Foveon? Yes.
Will the Foveon have it's own problems? Yes.
Are existing Bayer sensors as brain dead as some seem to think? No.

I've seen this comment other places and I just wanted to say something. The 4/3 system hasn't even broken the surface. Specifically regarding lenses, though, the promise is to make them lighter, smaller, more efficient, and better suited for sensor size..all for cheaper. What if you could sell 1 of your Nikon lenses and buy 2 or 3 4/3 lenses for starters. Then, what if those lenses deliver what they promise?

psd250 said:

I still woudl only want my Nikon lenses...so whats the point.

Click to expand...

Michael Long wrote:
[snip]

Michael Long said:

There have been many, many companies that have introduced
"superior" technologies -- and not managed to properly market and
sell them.

I do hope -- and I mean it sincerely -- that Foveon isn't one of them.

Click to expand...

Hear hear. I would love to see this chip in a 4/3 system. That's a camera that would really tempt me.

As for now, even if I was in the market for a D-SLR now (and I'm not), I wouldn't buy the SD-9: the 1.7x focal length multiplier and lack of JPG capability are showstoppers for me.

The other known limitations (Sigma lens mount, max ISO400) I probably could live with: there are Sigma lenses good enough for my purposes, and with 12 bits of data in RAW it shouldn't be a problem to push the CCD to 1600 and have usable quality; with a bright lens the low-light performance ought to be acceptable..

Petteri
--
http://www.seittipaja.fi/index/

What are you talking about "mfg'er" sounds more liek you are swearing and that "um" sounds more like you are pausing in midsentance...um well we were walking along and um...you get teh point. Anyways if you do have any type of info please support with links or a refernce to a source so we can follow up on it. Other than that I would say that you have spent a "little" too much time with your Sony f707...AHHAHAHA...I knew that Sony product would get to you.

Jimmy Chen said:

Jimmy Chen said:

I only have one question. Did it take a company like Foveon to
make a cool sensor X3 for Canon and Nikon and the likes to Suddenly
introduce "superior technologies" why haven't they done so until
now?

Click to expand...

Foveon isnt actually new. It is based off a 20 yrs old Kodak patent
that couldnt be done back then. Foveon is using NSC's 0.18 um
mfg'er process for their chip, and even your Canon CMOS is still
being produced with the 0.5 um process.

--
jc
Sony F707
http://www.reefkeepers.org/gallery/f707
http://www.reeftec.com/gallery

Click to expand...

All parallels aside, I agree that Nikon is already a "defacto" standard for the 35mm dSLR but a standard lens mount is only one of the advantages of 4/3. An equally, if not more important advantage of 4/3 is sensor size. Being approximately half the size of a full frame 35mm sensor, it should single handedly knock $500-$1000 off the production costs of the sensor. Not a small difference. The problem with 35mm dSLR's is that they will always be too expensive for most and a gaping hole has opened in the middle of the market. It will take something like 4/3 to fill this hole. Another possibility is that the mid-range market will stick to the model pursued to date and abandoned interchangable lenses altogether. Who needs them with the long range zooms available now? Well...maybe, but of course long zooms can't do everything and they do have their disadvangtages. But that is another topic

Getting more on track with the original topic, in summary I feel that if the SD9 was built to 4/3 specs (with an X3) and the price was right, the demand would be so high that Sigma wouldn't be able to make them fast enough. Speculation of course, but there sure seem to be a lot of people on these forums begging for a product like this.

Fred

work_permit said:

I think the 4/3 concept is more like Linux then the IBM PC.

As has been stated elsewhere, IBM gave the PC Industry needed
credentials. People wrote software for its platform (specifcally,
Lotus). The clones came in and in the end the IBM/Windows/Intel
architecture dominated the market.

Today, It feels like Nikon is playing the role of IBM. Other DSLR
makers like Kodak and Fuji are essentially "cloning" the DSLR by
creating bodies that are compatible with the Nikon lens mount.

4/3 is an attempt to create a standard when a "defacto" Nikon
standard exists. Feels more like Linux. And while I really do
hope Linux succeeds, the jury is still out on whether it can
dethrone Wintel.

I DO feel there are significant advantages to the 4/3 project. For
example, lens design is simplified if image post-processing is
assumed. Digitally-correctable defects include distortion (barrel,
pincushion, gullwing), lateral chromatic aberration, and
vignetting. Fewer glass elements yield lower weight, less flare,
lower price.

I do hope 4/3 succeeds. But somehow, I doubt it will.

Click to expand...

Well, for that matter, I'm sure there's a significant amount of processing that could be done with the Foveon to invent data that isn't there.

I guess I was referring to the sensor specifically, not the DSP algorithms employed in 'synthesizing' a final output (good Lord, then we'd be comparing Braeburns to Galas to Golden Delicious to GrannySmiths to.... ;-) )

Michael Long said:

Michael Long said:

Most probably think in terms of what resolution means relative to
what they see when they look at their computer monitor. If we want
to play that game, then I could say that my computer monitor
resolution is really 3072x2304, if you count each Red, Green, and
Blue pixel component seperately.

Click to expand...

No, because if they used your "logic" they'd claim that a 6MP CFA
sensor using a typical Bayer-algorithm is actually a 54MP plus
sensor. They don't.

Every pixel in a Foveon sensor contributes it's luminance and color
data one-for-one to the output. No sampling. So 3.54M x 3 (stacked
RGB) = 10.62 MPS (photo-sites) in, 3.54MP out.

Every pixel in a CFA sensor contributes it's luminance data
one-for-one to the output. As the eye is more sensitive to
luminance, this contributes more to perceived resolution. Every
pixel in a typical bayer algorithm determines it's color value by
sampling itself and it's adjacent neighbors.

So in effect, 6M x (1 (source) + 8 (adjacents)) = 54MPS in, 6MP
out. Advanced algorithms will sample additional h, v, and diagonal
pixels for better edge detection.

If you're interesting in learning the facts, I can can point you to
some sites that describe the algorithms and processes I've
described.

Summary:

Is the Bayer system a hack? Yes.
Is it a very elegant hack that works exceedingly well? Yes.
Are companies misrepresenting pixel counts? No. (Maybe Fuji)
Will the Foveon deliver more accurate color information? Yes.
Do I wish the 1DS was an 11MP Foveon? Yes.
Will the Foveon have it's own problems? Yes.
Are existing Bayer sensors as brain dead as some seem to think? No.

Click to expand...

I've jumped into the middle here without reading the rest of this thread but surely apart from the 1Ds and the new Kodak all the other mainstream DSLRs are based on approx 1/2 frame sized sensors? I don't see how the marginally smaller 4/3 size can lower production costs much.

lockwoodfe said:

Getting more on track with the original topic, in summary I feel
that if the SD9 was built to 4/3 specs (with an X3) and the price
was right, the demand would be so high that Sigma wouldn't be able
to make them fast enough. Speculation of course, but there sure
seem to be a lot of people on these forums begging for a product
like this.

Fred

work_permit said:

I think the 4/3 concept is more like Linux then the IBM PC.

As has been stated elsewhere, IBM gave the PC Industry needed
credentials. People wrote software for its platform (specifcally,
Lotus). The clones came in and in the end the IBM/Windows/Intel
architecture dominated the market.

Today, It feels like Nikon is playing the role of IBM. Other DSLR
makers like Kodak and Fuji are essentially "cloning" the DSLR by
creating bodies that are compatible with the Nikon lens mount.

4/3 is an attempt to create a standard when a "defacto" Nikon
standard exists. Feels more like Linux. And while I really do
hope Linux succeeds, the jury is still out on whether it can
dethrone Wintel.

I DO feel there are significant advantages to the 4/3 project. For
example, lens design is simplified if image post-processing is
assumed. Digitally-correctable defects include distortion (barrel,
pincushion, gullwing), lateral chromatic aberration, and
vignetting. Fewer glass elements yield lower weight, less flare,
lower price.

I do hope 4/3 succeeds. But somehow, I doubt it will.

Click to expand...

Click to expand...

The problem I see with Sigma SD9 (and all other 35mm based DSLRs)
is that basically they are 35mm based.

The lense is designed to evenly cover a 35 mm diagonal rectangle
with an image. The camera is designed with the lens mounted far
enough away from the sensor (film or digital) that a > 35 mm diagonal
mirror has space to sit at 45 degrees and flip up out of the way when
the photo is being taken.

In the SD9 the sensor is only a fraction of that size however, making
the size of the mirror and the image circle of the lens redundant. The
multiplication factor is 1.7. This can be described as a FOV crop or
other ways. The implications for the SD9 are that a lens designed to
have a certain focal length, actually has a focal length 1.7 x longer. Also
the depth of field will be significantly greater. Problems arise when wide
angle shots are needed. The widest zoom lens available I think is 15-35mm
which translates into a focal length equivalence of 25.5-59.5. My current
widest 35mm lens (pentax KAF mount) is 18mm which is conciderably wider
than what is available on the Sigma SD9.

Prosumer cameras do have multiplication factors for their lenses when
comparing focal length with 35mm, and these are still calculated in the
same way as for DSLRs. However, the implications of the multiplication

factor is much less because the lens and the sensor were actually designed to work with one another in the first place (there is no 35mm mirror in the
Sony S707 for example). Certainly there is no field of view crop. However
there is a significant increase in depth of field (which can be a good or a
bad thing depending on the type of shot required).

Frank B said:

Michael Long said:

Frank B said:

Sorry, I don’t agree. The original poster used FOV (field of
view) not FLM (focal length multiplier). The FOV of the Sony is
the same as the FOV of a 35 mm camera at a focal length ratio of
approximately 3.91. I don’t see what was wrong with his post.

Click to expand...

The Sigma SD9 / Foveon has a FOV factor / FLM of 1.7 due to the
smaller sensor size under a 35mm lens. The lens / sensor in the
Sony F707 has a FOV factor / FLM of 1 (not 3.91) due to the matched
lens / sensor size.

Click to expand...

My statement above is accurate. I believe that is all the poster
was saying.

Frank B said:

Read the original reply again:

Frank B said:

Frank B said:

6) They need wide angle capabilities (1.7x FOV factor, worse than
D60).

Click to expand...

And compared to my Sony F707's 3.91x FOV factor, the 1.7x FOV factor is a bueaty.

Click to expand...

The Sony does not have a "3.91x FOV factor" (his words) in the same
way that the SD9 has a 1.7x factor, as the Sony does not have a
"cropped" sensor area.

Click to expand...

Neither does the Sigma have a cropped sensor area. The sensor is
fully covered by the lens used. The so called "crop" is of the 35
mm film size just as theSony when compared to a 35 mm camera could
be said to represent a "cropped" portion. The only difference is
that the lenses used on the Sigma cover the 35 mm area. So what,
the ratio is still based on the sensor size of the digital camera
and the sensor size of a 35 mm camera. If the sensor size of the
Sigma were larger the ratio would be smaller. It really has
nothing to do with the lens. If Sigma made a lens that covered
only the sensor the ratio of the sensor to 35 mm film would remain
the same.

Frank B said:

His direct comparison of the Sony's FOV factor (which, when used
that way, it doesn't have) to the Sigma's FOV factor (which does)
was the same as comparing apples to oranges.

The post starting this subthread and the one following this one
shows that others are reading it the same way.

http://www.dpreview.com/forums/read.asp?forum=1000&message=3488205

As I said earlier, I've spent enough time illustrating this point.
I'll let you have the last word.

Click to expand...

Closer to apples to apples.

Click to expand...

--
J

Very good explanation. I agree, except that I'm not so sure the implications are so different. For example, if Sigma made dedicated lenses for their CCD the lenses could be lighter and might be a little sharper than their 35 mm lenses, but I believe their characteristics (e.g. DOF, multiplier when compared to their 35 mm counterparts) would otherwise be the same. You don't find any really wide angle lenses on prosumer camera's either. Although, maybe with a dedicated lens Sigma could make a wider angle lens than is currently available when you consider the multiplier.

Frank B

jonathan68 said:

The lense is designed to evenly cover a 35 mm diagonal rectangle
with an image. The camera is designed with the lens mounted far
enough away from the sensor (film or digital) that a > 35 mm diagonal
mirror has space to sit at 45 degrees and flip up out of the way when
the photo is being taken.

In the SD9 the sensor is only a fraction of that size however, making
the size of the mirror and the image circle of the lens redundant. The
multiplication factor is 1.7. This can be described as a FOV crop or
other ways. The implications for the SD9 are that a lens designed to
have a certain focal length, actually has a focal length 1.7 x
longer. Also
the depth of field will be significantly greater. Problems arise
when wide
angle shots are needed. The widest zoom lens available I think is
15-35mm
which translates into a focal length equivalence of 25.5-59.5. My
current
widest 35mm lens (pentax KAF mount) is 18mm which is conciderably
wider
than what is available on the Sigma SD9.

Prosumer cameras do have multiplication factors for their lenses when
comparing focal length with 35mm, and these are still calculated in
the
same way as for DSLRs. However, the implications of the multiplication
factor is much less because the lens and the sensor were actually
designed to work with one another in the first place (there is no
35mm mirror in the
Sony S707 for example). Certainly there is no field of view crop.
However
there is a significant increase in depth of field (which can be a
good or a
bad thing depending on the type of shot required).

Frank B said:

Michael Long said:

Frank B said:

Sorry, I don’t agree. The original poster used FOV (field of
view) not FLM (focal length multiplier). The FOV of the Sony is
the same as the FOV of a 35 mm camera at a focal length ratio of
approximately 3.91. I don’t see what was wrong with his post.

Click to expand...

The Sigma SD9 / Foveon has a FOV factor / FLM of 1.7 due to the
smaller sensor size under a 35mm lens. The lens / sensor in the
Sony F707 has a FOV factor / FLM of 1 (not 3.91) due to the matched
lens / sensor size.

Click to expand...

My statement above is accurate. I believe that is all the poster
was saying.

jonathan68 said:

Read the original reply again:

jonathan68 said:

jonathan68 said:

6) They need wide angle capabilities (1.7x FOV factor, worse than
D60).

Click to expand...

And compared to my Sony F707's 3.91x FOV factor, the 1.7x FOV factor is a bueaty.

Click to expand...

The Sony does not have a "3.91x FOV factor" (his words) in the same
way that the SD9 has a 1.7x factor, as the Sony does not have a
"cropped" sensor area.

Click to expand...

Neither does the Sigma have a cropped sensor area. The sensor is
fully covered by the lens used. The so called "crop" is of the 35
mm film size just as theSony when compared to a 35 mm camera could
be said to represent a "cropped" portion. The only difference is
that the lenses used on the Sigma cover the 35 mm area. So what,
the ratio is still based on the sensor size of the digital camera
and the sensor size of a 35 mm camera. If the sensor size of the
Sigma were larger the ratio would be smaller. It really has
nothing to do with the lens. If Sigma made a lens that covered
only the sensor the ratio of the sensor to 35 mm film would remain
the same.

jonathan68 said:

His direct comparison of the Sony's FOV factor (which, when used
that way, it doesn't have) to the Sigma's FOV factor (which does)
was the same as comparing apples to oranges.

The post starting this subthread and the one following this one
shows that others are reading it the same way.

http://www.dpreview.com/forums/read.asp?forum=1000&message=3488205

As I said earlier, I've spent enough time illustrating this point.
I'll let you have the last word.

Click to expand...

Closer to apples to apples.

Click to expand...

--
J

Click to expand...

my humble opinion
unless sigma knocks the socks off with their "quality" it will be a big bust
somewhat like the contax n digital

limlh said:

The SD9 is probably going to be priced a few hundred dollars lower
than the D60. And with cheaper Sigma lenses, you are talking about
SD9 being the first "prosumer" DSLR that has an image quality that
beats the D60. So the real buyers are out there just waiting to
pounce on it once the SD9 goes on sale.

--
http://www.pbase.com/limlh

Click to expand...

--
beam me up scotty

im giving it all shes got captain

psd250 said:

What are you talking about "mfg'er" sounds more liek you are
swearing and that "um" sounds more like you are pausing in
midsentance...um well we were walking along and um...you get teh
point.

Click to expand...

I am sorry if you do not understand the standard technical terms when you try to get yourself involve in a technical discussion. May I suggest trying to learn the technical terms instead of throwing out flame baits.

--
jc
Sony F707
http://www.reefkeepers.org/gallery/f707
http://www.reeftec.com/gallery

You are correct when referring to 1/2 frame sensors. If you read my argument again you will notice that I qualified my statement by saying full frame 35mm sensors. I am assuming that long term, the 35mm dSLR will migrate to full frame sensors to eliminate the lens multiplier.

DMillier said:

I've jumped into the middle here without reading the rest of this
thread but surely apart from the 1Ds and the new Kodak all the
other mainstream DSLRs are based on approx 1/2 frame sized sensors?
I don't see how the marginally smaller 4/3 size can lower
production costs much.

Click to expand...