FOVEON - Dead in the Water?

kecohen · Jan 10, 2004

particularly knowlegeable on Foveon. They have a stupendous digital back that delivers amazing stuff.

If Nikon rejected Foveon, you might just consider that Nikon was already neck deep in a contract with Sony, licensing and/or purchasing another ccd might not have been a good way for them to go financially and still be "competitive." It may well be that the Foveon chip may not be able to deliver "fast" images suitable to a pj camera. Unlike Canon who has opted for two different imagers, CMOS and CCD, Nikon for financial considerations may have decided that one imager that could serve both of its pro markets would be best.

Imagining that just because a company doesn't buy Foveon chips doesn't mean there is anything wrong with them; nor does it mean the chips aren't the best imagers money can buy. It only means that company's priorities may prevent it from doing so. After all, would you buy a D2H today if it were $1000 more? Companies have to make decisions (and compromises) based on their assessment of their target market. I assure you that Nikon spends big money to accurately evaluate their markets. Their decisions aren't based on any single criteria.
--
Karen

...but if you try sometimes, you just might find, you get what you need.

http://www.e-designarts.com
http://www.edaphoto.com
http://www.pbase.com/kecohen/sports_action

Erik Magnuson · Jan 10, 2004

Laurence Matson said:
China has huge market potential which seems much more oriented
toward what we would call smaller brands, such as Samsung.

It's cetrainly stretching the point to call a multi-billion dollar conglomerate like Samsung a "smaller brand".

Laurence Matson said:
Therefore, were Sigma to
come out with something, which could well be a collaborative effort
with an outsourcer,

OK, what does Sigma bring to this party? In other words, let's assume that someone (or a partnership of multiple someones) has designed and funded'an X3-based P&S to be manufactured by a Chinese OEM. What part of that camera would be from Sigma? Would it be the lens? (In some ways that would be the most logical part because Sigma is known lens maker, but the've not competed in this space before.) The electronics? Or just the "brand" name? (And will NatSemi make the sensors in Maine and ship them to China or will the sensors also be OEMed out to a local fab?)

Sigma certainly has some name recognition among SLR owners/buyers. But does it have any "presence" among the punters in the more general target market? (Whenever I've mentioned it, most people who did not already have a recent SLR said "Who?") Certainly here in the states, such a camera would possibly do better if it were distributed under a name like "Polaroid".

Which brings up the other issues of brand names: import, distribution and service. Did Sigma retain the rights to the Sigma name in all of the interesting markets? For example, we've already heard that the Swedish distributor for Sigma may not be interested in handling the SD10. Could Sigma bypass reluctant distributors or do those distributors have exclusive rights to the Sigma trademark for their countries? Are the Sigma distributors capable of handling the potentially larger volume of sales/returns/repairs that such a product would bring? Do they have the connections to get the product into the right retail shops? (Face it: if it's only available by special order/mail order or by ones and twos in some shops, it won't sell in large numbers.)

Laurence Matson said:
and bring it to market for a reasonable price
with say 3 million pixel locations or 9 million pixel sensors and 4
mu pixel pitch,

The devil is in the details as to price/features/performance. This one better have JPEG. If it's limited to selling to only to those who are willing to process RAW, it's not likely to be any more successful than the Sigma SD series. One area where Foveon does not seem to have made much progress is in infrastructure support for x3f files. You can't just bring your memory card into the local 1-hour place and get prints made. Or hook the camera directly up to an inkjet or dye-sub printer. For the non-computer savvy and snapshot market, this is a critical feature.

Laurence Matson said:
it would probably do a lot better than you think in
the global market.

Of course this depends on defining appropriate values for "a lot better" and "you think". It's not even clear yet whether the competition will be the 828/A1/S7000 class or the more crowded 5MP 3x zoom class.

--
Erik

Erik Magnuson · Jan 10, 2004

kecohen said:
particularly knowlegeable on Foveon. They have a stupendous digital
back that delivers amazing stuff.

You're not too knowledgeable yourself. It's not "have", it's "had" (past tense: no longer being made/sold because it's not competitive) And like the Sigma SD series, it may have had some technical merits, but had very little sales/impact on the digital back market (which Foveon has abandoned for the time being.)

kecohen said:
Imagining that just because a company doesn't buy Foveon chips
doesn't mean there is anything wrong with them; nor does it mean
the chips aren't the best imagers money can buy.

It strongly suggests they are not obviously superior for some reasonable set of criteria (or that Nikon is not rational.) But remember, we're not just talking about Nikon. When every major SLR vendor (save one minor vendor) also makes another decision, then the assumption of "best that money can buy" is a little suspect.

--
Erik

Joseph S Wisniewski · Jan 10, 2004

Laurence Matson said:
Sean Keegan said:

The RAW processing of a foveon image is VERY
processor intensive, which is probably why there is no JPEG in
camera. The in camera processor probably cannot handle it.

Click to expand...

Like those CAPS. Got any evidence of this? Is the raw processing in
the SD10 really actually more processor-intensive than say the
Bayer process?

He might not, but I do.

A Bayer camera can create an in-camera JPEG with a very simple algorithm, some form of interpolation that normally doesn't involve more than 13 cells in a fix weighted pattern (13 macs, multiply and add operations) then a simple color adjustment using a 3x3 matrix (9 macs), for a grand total of 22 macs per cell.

I've been unable to get what I consider acceptable color (under 5E, even in scenes without metamerism problems) via a 3x3. Acceptable color takes a 2D LUT. That's 7 macs to normalize, 4 to index, 48 (16x3) to interpolate, and 3 to restore the normalized luminance. A grand total of 62. About 3x what it takes to do a cell in a Bayer camera. Then consider that there's 1.7x more cells that 6mp Bayer cameras with similar sized processors and you're up to 4.8x (more or less) what you need to do for a Bayer camera.

Both Foveon and Bayer need some form of noise processing and sharpening before the final JPEG output. The Bayer will typically need more sophisticated (and mor computationally intensive) sharpening, the Foveon makes up for this in it's need for more agressive noise work.

And one other thing you have to watch out for is the Bayer "blows out" gracefully, the Foveon doesn't. That's why SPP 2 handles blowouts better than SPP 1, because they added an improved "blowout fixer", probably an inpainting routine.

Laurence Matson said:
Are you telling me that the camera does not make a
jpeg in-camera right now? Have you checked that? Are you using one?

Just the thumbnail, and no one really cares if a thumbnail uses a 3x3 matrix color algorithm and ends up over 5E off. The little LCD on the camera is already worse than 5E.

Laurence Matson said:
Be careful here, because I am right and you are wrong on all the
issues above. Do your homework before JL shows up. You don't want
to deal with her again, do you?

I kind of miss her.

--
Ciao!

Joe

http://www.swissarmyfork.com

Joseph S Wisniewski · Jan 10, 2004

Jason Busch said:
There is no single best camera out there, but there sure are some
lesser-than DSLRs, no?

No.

There are no "lesser-than" DSLRs. There are some that are less suited to particular purposes than others. That's being "different", not "lesser", not "greater".

Jason Busch said:
What kind of lens selection do you have? Do
you have a 5-year warranty on your lenses? Does your camera accept
other brand lenses?

Good points, which do affect the Sigma's suitability to some purposes. As I've mentioend before, if my business hinged around portrait work, I'd say a Nikon trounced the Sigma thoroughly in lenses. In landscape it's a wash. Same with wildlife.

Jason Busch said:
I have a D100 and I can also buy Sigma lenses.
Almost any lens you put on your SD10, I can get that same lens, and
it will probably work better on a Canon or Nikon DSLR. Can you put
Canon or Nikon lenses on your SD10?

Yes to the Nikon, anyway, and to M42 and M39. As soon as Satoru-san comes through with a delivery date for my Nikon adapter, one of my incom thax check goodies is going to be an SD10 to go with my Nikon system.

Jason Busch said:
What about your focusing
speeds? What about your dynamic range? What about 3rd party
accessories?

Sucks, about as bad as the competition, and sucks, respectively.

Jason Busch said:
I think your purchase sounds like a rationalization. By just
comparing features of Nikon and Sigma, Nikon easily comes out as
having more features.

True. But I try to avoid comparing features. Comparing features leads manufacturers to a concept called "shovelware" or "creeping featuritis", where you see how many features you can cram in, to get the count up higher than yor competator's feature count, regaudless of how crappy those featurea are implemented.

Jason Busch said:
Hypothetically, even if the SD10 had better image quality, I can
make up for that in photo editing.

Possibly, but what does that do to your workflow. Remember, time spent on each printable image is part of your TCO (total cost of ownership).

Jason Busch said:
What you and I cannot make up for, is in features and performance.
You can't improve a SD10's focusing speeds. You can't tell Sigma
that their lenses will now have a 5-year warranty.

You can't make a Nikon D100 deliver a 10.2 megasomething image.

Jason Busch said:
Remember, I'm formerly an Olympus owner, so I had no interest in
either a Canon or a Nikon, or Fuji, or whatever. I simply compared
features, performance, and price, and voila, my D100 has not let me
down for over a year and a half.

Neither have either of mine. But that's not a statistically significant sample. You'll find Nikon D100 users who have had problems (needing to go in for focus recalibration or meter recalibration or firmware upgrades, or broken aperture coupling levers, or stuck secondary mirrors). You'll find plenty of Sigma users who have also had problem free experiences, and others who have had their share of problems.

The real problem is getting meaningful statistics.

Jason Busch said:
I wish people wouldn't make it so hard for themselves, but, that's
why Sigma and lesser known brands exist, because some people just
like being outcast DSLR owner and don't want to buy what everyone
else is buying.

True of cameras, cars, houses, spouses, life, the universe, and everything.

Jason Busch said:
Well, there are very good reasons why we buy the
cameras from the big boys, and it's not just because of the name.

I guess what it comes down to is that you don't use a camera like I
do.

Perfect, exactly what it comes down to. You bought the camera that best fits their needs. Others buy the ones that best suit their needs.

Like you, for my bread and butter, a few Nikon bodies and a big bag of Nikon lenses best suits my needs. But I'm not everyone. I try not to speak "for everyone". Sometimes I screw up, and in those cases, I try to accept correction gracefully.

And Jason, you're not everyone, either.

--
Ciao!

Joe

http://www.swissarmyfork.com

kecohen · Jan 10, 2004

Erik Magnuson said:
kecohen said:

particularly knowlegeable on Foveon. They have a stupendous digital
back that delivers amazing stuff.

Click to expand...

You're not too knowledgeable yourself. It's not "have", it's "had"
(past tense: no longer being made/sold because it's not
competitive) And like the Sigma SD series, it may have had some
technical merits, but had very little sales/impact on the digital
back market (which Foveon has abandoned for the time being.)

No, they still have and still support the digital back. They also have a product for video cameras, in addition to the X3 for digital still 35mm imaging.

Were you also aware that as of August of 2003 they have a new CEO who's job is to get the products in mass production and marketing? Do you know who they are talking with and what deals are possibly in the works? Do you also know that the company is still keeping their technology "close to the vest?"

Member said:
Member said:

Imagining that just because a company doesn't buy Foveon chips
doesn't mean there is anything wrong with them; nor does it mean
the chips aren't the best imagers money can buy.

Click to expand...

It strongly suggests they are not obviously superior for some
reasonable set of criteria (or that Nikon is not rational.) But
remember, we're not just talking about Nikon. When every major SLR
vendor (save one minor vendor) also makes another decision, then
the assumption of "best that money can buy" is a little suspect.

No, it doesn't matter whether or not it's the best that money can buy if it's too expensive, or otherwise doesn't meet a company's requirements.

We would like to think that Nikon's or Canon's highest priority is image quality, but I assure you it isn't. "Sufficiently good" image quality for the needs of the targeted market segment is satisfactory. Let me emphasize, "best image quality" and "obviously superior" are still relative to price and utility.

Additionally, there are many considerations that go into the decision to choose one component of a complex product over another, not the least of which may be the soundness of the company who manufactures it, its track record at producing mass quantities, their ability to maintain quality control and meet deadlines. If I were Nikon or Canon, would I necessarily jump at the chance to buy a key component from an R&D company with no experience in production? And if that R&D company insisted on manufacturing it themselves in order to safeguard their secrets, well, that would sure be a deal breaker for me.

I'm not saying Foveon's chip is the best that money can buy, only that even if it were, there are plenty of possible reasons why other companies could consider it "undesirable."

It's like a mentor once told me: A good idea can fail without the proper support, but a bad idea will fail all by itself and needs no help from you.

I could be wrong, but I believe that the Foveon chip is a good idea that's failing because of very poor management. If the new CEO is successful, we might actually find out.

--
Karen

...but if you try sometimes, you just might find, you get what you need.

http://www.e-designarts.com
http://www.edaphoto.com
http://www.pbase.com/kecohen/sports_action

Laurence Matson · Jan 10, 2004

Joe,

Well you've got me wondering.

Joseph S. Wisniewski said:
A Bayer camera can create an in-camera JPEG with a very simple
algorithm, some form of interpolation that normally doesn't involve
more than 13 cells in a fix weighted pattern (13 macs, multiply and
add operations) then a simple color adjustment using a 3x3 matrix
(9 macs), for a grand total of 22 macs per cell.

Just because you cannot figure it out does not mean that it has to that much more complicated. Indications are that the X3 calculation might actually be easier and less processor-intensive than the CFA. Give it some thought.

Joseph S Wisniewski said:
I've been unable to get what I consider acceptable color (under 5E,
even in scenes without metamerism problems) via a 3x3. Acceptable
color takes a 2D LUT. That's 7 macs to normalize, 4 to index, 48
(16x3) to interpolate, and 3 to restore the normalized luminance.
A grand total of 62. About 3x what it takes to do a cell in a Bayer
camera. Then consider that there's 1.7x more cells that 6mp Bayer
cameras with similar sized processors and you're up to 4.8x (more
or less) what you need to do for a Bayer camera.

That's your approach, but it does not mean it is Foveon's approach. Consider that there might be other, more efficient ways of dealing with this. Consider that you are dealing with an analog chip. Think outside the box for a moment.

Joseph S Wisniewski said:
Both Foveon and Bayer need some form of noise processing and
sharpening before the final JPEG output. The Bayer will typically
need more sophisticated (and mor computationally intensive)
sharpening, the Foveon makes up for this in it's need for more
agressive noise work.

Maybe, but it could be handled by purely tuning the sensor electronics before the image capture. That would not add a lot of overhead. Think outside the box.

Joseph S Wisniewski said:
And one other thing you have to watch out for is the Bayer "blows
out" gracefully, the Foveon doesn't. That's why SPP 2 handles
blowouts better than SPP 1, because they added an improved "blowout
fixer", probably an inpainting routine.

Are you sure the improved "blow out" handling is purely a function of SPP? Have you looked at the publically available spec sheet on SD9 firmware improvements. I gotta get you outside the box for a bit, Joe.

Joseph S Wisniewski said:
Just the thumbnail, and no one really cares if a thumbnail uses a
3x3 matrix color algorithm and ends up over 5E off. The little LCD
on the camera is already worse than 5E.

Is it really just that? Why can I look at it at 200% then?

Joseph S Wisniewski said:
I kind of miss her.

She misses you too, Joe. Sean too.

Joseph S Wisniewski said:
Ciao!

dog food

--
There is a tide in the affairs of men,
Which, taken at the flood, leads on to fortune;
Omitted, all the voyage of their life
Is bound in shallows and in miseries.

Laurence

http://www.pbase.com/lmatson/root
http://www.pbase.com/sigmasd9/root
http://www.pbase.com/cameras/sigma/sd10
http://www.pbase.com/cameras/sigma/sd9
http://www.beachbriss.com (eternal test site)

Erik Magnuson · Jan 10, 2004

kecohen said:
No, they still have and still support the digital back.

" Foveon has phased out production of Foveon II Studio cameras to concentrate on image capture systems."

Most people do not use "have" for products that are no longer made. You can still get support for a 50-year old Leica. Does that mean Leica "has" the IIIe?

kecohen said:
They also
have a product for video cameras, in addition to the X3 for digital
still 35mm imaging.

They do? There are only 2 products on their web site: the X3 10M and 5M. AFAIK, the 5M is not shipping in any current product, still or video. You might say the HANVISION camera does video with the 10M, but not in any useful consumer sense.

kecohen said:
who's job is to get the products in mass production and marketing?

Yes, and that's the antithesis of the large format digtal back market. (And maybe even the DSLR market.)

kecohen said:
Do you know who they are talking with and what deals are possibly
in the works?

No, I'm only commenting on what is and not what might be.

kecohen said:
Do you also know that the company is still keeping
their technology "close to the vest?"

Perhaps a little too close. They need to get more product out.

kecohen said:
No, it doesn't matter whether or not it's the best that money can
buy if it's too expensive, or otherwise doesn't meet a company's
requirements.

According to Foveon's marketing, "too expensive" should not have been the case. But in any case, not meeting the requirements of the customer base is not a convincing argument for superiority.

kecohen said:
"Sufficiently good" image
quality for the needs of the targeted market segment is
satisfactory. Let me emphasize, "best image quality" and "obviously
superior" are still relative to price and utility.

I've shot the 300D and the SD9 side by side for several hundred images. I think I have a pretty good grasp of the relative merits of at least these two implementations. Each has their strengths and weaknesses, but annointing one "obviously superior" in terms of image quality is rather dubious or at least highly subjective.

kecohen said:
And if that R&D company insisted on manufacturing it
themselves in order to safeguard their secrets,

Two minor flaws in this argument:

1. Foveon does not manufacture the sensor, NatSemi does.

2. Sony probably does not share their sensor manufacturing or design secrets with Nikon either and that did not stop the D100.

kecohen said:
I could be wrong, but I believe that the Foveon chip is a good idea
that's failing because of very poor management. If the new CEO is
successful, we might actually find out.

I don't know if it's failing or not because there is not enough information on what might constitute success. It certainly has not lived up to the hype at annoucement.

--
Erik

Laurence Matson · Jan 10, 2004

Karen,

Interesting comments. But I wonder about this:

kecohen said:
I could be wrong, but I believe that the Foveon chip is a good idea
that's failing because of very poor management. If the new CEO is
successful, we might actually find out.

Is it really failing? Does it take an overnight success these days to prove that something is good. Some good ideas also take a long time to get to the really successful level. Consider the PDA (which I don't; don't own one), which did not do well at Apple but has become a famous success otherwise. The X3 concept could go that route if Foveon did not learn from the Apple experience. But I believe they did.

One could see the current phase as failure or as ramping up. Given the fact that Foveon have had only about 2 1/2 years to really develop this thing, I see it more as ramping up.

Interesting site of yours, btw.

--
There is a tide in the affairs of men,
Which, taken at the flood, leads on to fortune;
Omitted, all the voyage of their life
Is bound in shallows and in miseries.

Laurence

http://www.pbase.com/lmatson/root
http://www.pbase.com/sigmasd9/root
http://www.pbase.com/cameras/sigma/sd10
http://www.pbase.com/cameras/sigma/sd9
http://www.beachbriss.com (eternal test site)

kecohen · Jan 10, 2004

Laurence Matson said:
Interesting comments. But I wonder about this:

kecohen said:

I could be wrong, but I believe that the Foveon chip is a good idea
that's failing because of very poor management. If the new CEO is
successful, we might actually find out.

Click to expand...

Is it really failing? Does it take an overnight success these days
to prove that something is good. Some good ideas also take a long
time to get to the really successful level. Consider the PDA (which
I don't; don't own one), which did not do well at Apple but has
become a famous success otherwise. The X3 concept could go that
route if Foveon did not learn from the Apple experience. But I
believe they did.

Well, if a tree falls in the forest and no one hears it, does it really make a sound? Failing is the term I'm using to refer to its acceptance and use. If the company is bought out, as Thom thinks possible, it may take its rightful place in the industry (whatever place that may be.) If the company continues as it has been going, it is likely that it will fade into oblivion and whatever its capabilities or potential capabilities will never be explored.

Laurence Matson said:
One could see the current phase as failure or as ramping up. Given
the fact that Foveon have had only about 2 1/2 years to really
develop this thing, I see it more as ramping up.

I would agree, and hope that they are successful.

Laurence Matson said:
Interesting site of yours, btw.

When my parents used to say something was "interesting," it was their attempt to be kind and not say anything negative. How do you mean "interesting?" Are you being kind or haven't you made up your mind yet?

--
Karen

...but if you try sometimes, you just might find, you get what you need.

http://www.e-designarts.com
http://www.edaphoto.com
http://www.pbase.com/kecohen/sports_action

Search

FOVEON - Dead in the Water?

kecohen

Senior Member

Erik Magnuson

Guest

Erik Magnuson

Guest

Joseph S Wisniewski

Forum Pro

Joseph S Wisniewski

Forum Pro

kecohen

Senior Member

Laurence Matson

Forum Pro

Erik Magnuson

Guest

Laurence Matson

Forum Pro

kecohen

Senior Member

Keyboard shortcuts