If this is not false advertisement, what is?

[Joseph S Wisniewski]

that on the SD10/SD9 only 3 mega pixels are active out of 9
megapixels at any given time while the 6 mega pixels on the 10D are
fully activated ? We need to have some sort of standard in counting
active pixels here.

We do. The image processing industry has had such a standard for years. It's quite simple. By any accepted definition, the Foveon sensor is 3.5 megapixels. Any departure from this is marketing, not science.

Granted, they're 3.5 million really, really nice pixels...

I think Sigma and Fuji are "deceiving" the
general public don't you think ? I do.

Yes. And it's a pity, because the Sigma, Fuji, and Oly products are excellent in their own right. They're just painted weird colors by their respective marketing folks.

(had to throw Oly in. The "4/3 system" is hype above and beyond whatever Sigma and Fuji are doing, even though the actual camera and lenses really are quite nice).

--
Ciao!

Joe

http://www.swissarmyfork.com

 

[Just Looking]

We need to have some sort of standard in counting
active pixels here.

We do. The image processing industry has had such a standard for
years. It's quite simple. By any accepted definition, the Foveon
sensor is 3.5 megapixels. Any departure from this is marketing, not
science.

And the Fuji S2 is 12 MP by the same definition.

j

 

[Joseph S Wisniewski]

We do. The image processing industry has had such a standard for
years. It's quite simple. By any accepted definition, the Foveon
sensor is 3.5 megapixels. Any departure from this is marketing, not
science.

And the Fuji S2 is 12 MP by the same definition.

Nope. Just 6mp, in a grid rotated 45 degrees.

--
Ciao!

Joe

http://www.swissarmyfork.com

 

[Just Looking]

We do. The image processing industry has had such a standard for
years. It's quite simple. By any accepted definition, the Foveon
sensor is 3.5 megapixels. Any departure from this is marketing, not
science.

And the Fuji S2 is 12 MP by the same definition.

Nope. Just 6mp, in a grid rotated 45 degrees.

Oh, I see, you want to use the definition of only counting the actual pixel sensors, not the pixels in the output file. In that case, the 10.2 million for the SD9/10 was correct.

j

 

[Halldor Eiriksson]

I think Sigma is really cool. This is not deceiving, what is? They
make 10 MP

[the rest snipped]

People!!! Buy af Sigma!!! It's a gem!!!

Man, you folks are easy. A fairly new Canon user trolls the Sigma
forum just once, and hooks so many wriggly fish. He even got Phil.

Amazing.

--
Ciao!

Joe

http://www.swissarmyfork.com

true, the Sigma is clearly not for the imbecile

 

[Roland Karlsson]

While risky it's not UNTRUE and I certainly wouldn't call it
pathetic. Your reasoning is incorrect, the 10D has 6 million
photodetectors, it can only detect green, red or blue at anyone
position and then interpolates the color neighbouring pixels to
produce a 'best guess' of the actual color.

The X3 sensor has 10 million photodetectors, stacked in 3 layers.
It reads red, green and blue for each pixel location and hence
doesn't interpolate at all.

Hi Phil,

The SD10 might be a fine camera that can take very
good pictures, I have no doubt. But .. it is 3.4 Mpixels,
just as SD9 is 3.4 Mpixels. The spatial resolution is
2268x1512, and that is 3.4 Mpixels.

The camera then has the possibility to detect color
information at each pixel. This increases the color depth,
but not the resolution.

The Sigma press release says explicitely 10.2 Million Pixels.

It is the Bayer pixel counting that is dubious, but the Foveon
pixel counting is straight forward.

Roland

 

[Roland Karlsson]

...

Personally I think we shall take a step back and look
at Bayer pixel counting with some skepticism, and the
Fuiji 45 degress (twice as many pixels) with even more so.
The normal (non tilted) Bayer can (in the best cases)
produce that resolution, but the Fuiji - no way.

The Foveon on the other hand has exactly the resolution
that the pixel count tells - in this case 3.4 Mpixels.

Now - to be fair - both Foveon and Bayer needs anti alias
filters. If you don't have it, your pictures will look sharp
but weird at pixel level. There has been a lengthy discussion
about this in news:rec.photo.digital. A discussion where
some Foveon fan has shown weird photos of the moon and
other stuff, full of aliasing artefacts.

Roland

 

[William Wilgus]

. . . because they term their 3mp and 6mp sensors that are in-camera interpolated to 6mp and 12mp respectively as 6mp and 12mp cameras. Their now out of production 1.8mp 4900Z pro-sumer camera was touted as a 3mp camera by the same reasoning.

Until the SD10, Sigma advertised the sensor as 3.4mp. No doubt the practice of counting every sensor as a pixel element was forced upon them by habit of ttheir competitors doing so.

It's rather unfortunate for consumers that the digital camera industry has zeroed in on mega-pixels as being the only measure of a camera's capability. Just as mega-Herz doesn't tell the whole story with computers, mega-pixels don't tell the whole story with digital cameras.

William

 

[Roland Karlsson]

. . . because they term their 3mp and 6mp sensors that are
in-camera interpolated to 6mp and 12mp respectively as 6mp and
12mp cameras. Their now out of production 1.8mp 4900Z pro-sumer
camera was touted as a 3mp camera by the same reasoning.

Yes, Fuiji is out on very deep water.

Until the SD10, Sigma advertised the sensor as 3.4mp. No doubt the
practice of counting every sensor as a pixel element was forced
upon them by habit of ttheir competitors doing so.

They might feel the pressure to lie to the customers. But,
in my world a lie is a lie no matter how you motivate it.
So - Foveon lies. Just so you know. How do you like liers?

Roland

 

[Bruce Oudekerk]

Just when I thought I knew it all This issue needs clarification for MILLIONs of digital photographers.

If I understand what you are saying; when Canon (for instance) says the 300D has 6 MP they really mean there are 6 million discrete photo sensors (made up of an unequal numbered assortment of RGB sensors...Also that each one of these one-hue sensors is reinterpolated to provide "true” color depth information? ...Am I understanding you correctly??? I had assumed there was some pattern of R&G&B sensors that was concatenated to provide a discrete “True” color Pixel (24 bit + info.) In its most simplistic form that would be an RGB triad of pixels. Life is never simple and it looks as though this issue is even more complex than I had envisioned. It appears I can “sort of” understand a Bayer pattern without understanding how the pixels are couted.

To a certain extent the "proof is in the pudding" and the Foveon sensor produces a nice pudding. But I'm a big believer of looking over the ingredients on the box to see what I’m ingesting. We all need to know what we are dealing with and I suspect there are a lot of fairly knowledgeable people like myself that are being misled by the marketing hype and by misinformed assumptions. I hope you can shed some light on this for others and me. It might even make a really good article.

Bruce

While risky it's not UNTRUE and I certainly wouldn't call it
pathetic. Your reasoning is incorrect, the 10D has 6 million
photodetectors, it can only detect green, red or blue at anyone
position and then interpolates the color neighbouring pixels to
produce a 'best guess' of the actual color.

The X3 sensor has 10 million photodetectors, stacked in 3 layers.
It reads red, green and blue for each pixel location and hence
doesn't interpolate at all.

Phil Askey
Editor / Owner, dpreview.com

 

[Zone8]

Hi Andy,

Sounds as though you cannot accept the fact that the SD9 produces images superior to other DSLR systems, let alone what the SD10 offers. It really does not matter what the "pixel count" is - just look at the images. That says it all - provided they are by a competent photographer, of course.

Why don't you just go and take pictures. It really does not matter which camera system you select as your first choice, as long as YOU are happy with it. Personally, I was not happy with any DSLR UNTIL the SD9 came along. That was when I felt digital had come of its own and I have seen nothing to change my mind. But then, I wanted the highest quality available and for my moiney, the SD9 produced that and most importantly, still does excel.

The SD10 may be better in some ways but I am already quite content with the superb quality from my SD9 so whatever the "detractors" are shouting - and we have heard LOTS AND LOTS of them over the past year - users, who know how to take good photos, are still lauding the high quality of their SD9 images.

As SPP 2.0 should be available soon also to SD9 users, no doubt this will help produce even finer quality from past RAW files. Now - that says a lot for wanting RAW files over JPEGs from the camera.

May I humbly suggest that the best way to rant your feelings is to stand in front of a mirror, then you will be ranting to someone who agrees with you

Me? I'm off to take even more high quality images. Well, isn't that what photography is supposed to be about?

Zone8

 

[Mr. Pickles]

This is what Phil has posted that I guess if a Press Release from Sigma UK. It clearly states what the heck the sensor does and contains, and also what the others sensors are. And, I didn't see anywhere where they claim it is 3.4mp's or 10.2mp's or 10.2mpd's....

"1. FOVEON X3® Direct Image Sensor

Foveon® X3™ direct image sensor in the Sigma SD10 digital SLR can capture all RGB colors at each and every pixel. The revolutionary design of Foveon® X3™ direct image sensor features three layers of pixel sensors

Using three silicon-embedded layers of pixel sensors, stacked to take advantage of silicon's ability to absorb red, green, and blue light at different respective depths, the Foveon X3® direct image sensor can thereby directly capture full color and detail at each and every pixel location, without interpolation.

Other conventional image sensors feature just single layer of pixel sensors in a tiled mosaic pattern. Each pixel detects only one color of light, which has been filtered through a single red or green or blue color filter. As a result, two thirds of the color information at each pixel is blocked out. To compensate, the missing color is estimated through a complicated process of color interpolation, leading to color errors, color artifacts and loss of image detail.

Sigma SD10, powered by Foveon X3® direct image sensor, however, captures all three colors at every pixel location, ensuring the capture of full, complete color. It efficiently reproduces color more accurately, and offers sharper resolution, pixel for pixel, than any conventional image sensor."

--
Mr. Pickles

 

[Just Looking]

Until the SD10, Sigma advertised the sensor as 3.4mp.

Really? All the SD9 ads recently have said 10.2 Million Pixels, 10.2 Million Pixel Sensors, 10.2 Million photodetectors, things like that.

I've never seen Sigma or Foveon call it 3.4 MP. They started out with "3.4 MP x 3" as a way to make the total more implicit and the organization more explicit, but the times 3 kept getting lost, or confused with the trademark X3.

No doubt the
practice of counting every sensor as a pixel element was forced
upon them by habit of ttheir competitors doing so.

I think that's what they're saying.

It's rather unfortunate for consumers that the digital camera
industry has zeroed in on mega-pixels as being the only measure of
a camera's capability. Just as mega-Herz doesn't tell the whole
story with computers, mega-pixels don't tell the whole story with
digital cameras.

Agreed.

 

[Alatar]

If I understand what you are saying; when Canon (for instance) says
the 300D has 6 MP they really mean there are 6 million discrete
photo sensors (made up of an unequal numbered assortment of RGB
sensors...Also that each one of these one-hue sensors is
reinterpolated to provide "true” color depth information? ...Am I
understanding you correctly???

Yes, this is correct. A normal Bayer 6MP sensor has 1.5M red sensors, 3.0M green and 1.5M blue. They are arranged in an R-G-B-G pattern (see Phil's review). All the sensors are in different locations so they call it a 6MP sensor.

Since a 6MP colour bitmap contains 18M values, getting such an image from a 6MP Bayer sensor, 67% of the values in the image have to be calculated, and only 33% are actual measurements.

 

[Bruce Oudekerk]

This statement is true but CAN BE VERY misleading. I don’t think it is succinct at all. It misled me for a LONG time in its various incarnations. I’m still not completely convinced I understand it yet. Language is such an imprecise form of communication, even under ideal circumstances. I had always believed that the “tiled mosaic pattern” they are talking about here was the Bayer “Pixel” and I assumed that was how Foveon’s competitors counted it, that is, as one pixel. However it appears they use the individual RG or B sensor on the chip in determining the pixel count. Fuji and Sony’s new chip just obfuscate the issue more. If a distinct triad of sensors determined the information of an output pixel at the picture level it would be more obvious to everyone what was happening but that is not the case. To extract the most accurate information from modern CCDs and CMOS photo sensors requires some serious number crunching. The nature of the Foveon arrangement seems as though it produces, or has the capability to produce, a more accurate information packet in a much more straightforward computational manor.

Unfortunately the term “pixel” is bandied about and everyone calls it anything they want. Apparently “they” could be talking about an input pixel, a virtual pixel, a sensor pixel, and/or an output pixel that may or may not be the same thing.

Bruce

Quote from Foveon

"Other conventional image sensors feature just single layer of pixel
sensors in a tiled mosaic pattern. Each pixel detects only one
color of light, which has been filtered through a single red or
green or blue color filter. As a result, two thirds of the color
information at each pixel is blocked out. To compensate, the
missing color is estimated through a complicated process of color
interpolation, leading to color errors, color artifacts and loss of
image detail."

Mr. Pickles

 

[Jason Busch]

Andy,

I understand what you're saying. I'd like to see a Sigma SD9/10 photo next to a Canon 1Ds photo. Since there is only 1MP difference, I'm sure the Sigma will do amazing well, and everyone can save $6000. Right? The SD10 has more ISO range than a 1Ds, even.

Sigma reminds me of a local car dealership. By the time you leave their shop, you'll swear a Ford Pinto can outrace a Mustang GT. But then reality sets in, the false advertising fades away, you car is puffing oil smoke about a mile down the road, and you're stuck with a liability.

-------

"The name X3 comes from a unique capability that the Foveon X3 technology brings – the ability to capture three colors at each single pixel location..."

"The Foveon X3 breakthrough is accomplished by stacking three photodetectors within each pixel..."

Taking from those two statements, pixels and photodetectors are two different things. This is Foveon's own definition here. The sensor is NOT a 10MP sensor--it is a 10 Million photodetecting sensor, which has 3.34 million pixels. This is why Foveon is calling their sensor a 10M and not a 10MP.

Pixels are the picture elements of an image. A photodetector, is the TRANSLATOR, not the actual pixel itself (you can have as many "translators" as you want, but it still has to pass to a singular pixel location). The photodector translates light energy to electrical energy, to the pixel (picture element). The pixel understands electrical energy, and gives us what we call, an image.

Since Sony has made their 4-color sensor, they could make a 4-layer sensor and call their 8MP sensor, a 32 million photodetecting sensor, or a 32M 838. Just by that statement, Sony would sell a lot of cameras, and most people probably wouldn't even notice. That's the sad thing.

As Phil stated, Foveon is daring, and their statements are indeed, risky. Olympus is demonstrating some riske verbage themselves. But that's another story.

--
http://www.digitaldingus.com
http://pub103.ezboard.com/bthedigitaldinguscommunity

 

[Thomas Ko]

Here are some choices:
11M/4 RGBG, 6M/4 RGBG and 10M/3 RGB
vs
2.75MX4 RGBG, 1.5MX4 RGBG and 3.4MX3 RGB
vs
11M, 6M and 10M

See, lie can provide more "Simple", "acurate" and "Useful" measurement to general audiences. Probably that's why when a patient is in critical situation and the doctor will say "you are ok, you will be fine" in order to tell him "u are very likely dying, tell me the last word in case i die".

--
Thomas the C.Wolf 8^)
Gallery:
http://www.pbase.com/sigmasd9/c_wolf
http://www.pbase.com/c_wolf

 

[Jason Busch]

Here are some choices:
11M/4 RGBG, 6M/4 RGBG and 10M/3 RGB
vs
2.75MX4 RGBG, 1.5MX4 RGBG and 3.4MX3 RGB
vs
11M, 6M and 10M

I don't think you understand what's going on. We're talking megapixels and photodetectors. These are two entirely different things. Foveon is betting you don't know the difference either.

Unless I'm missing something. Your examples are to make an implication that an 11MP and a 6MP camera have somewhat less resolution than the Foveon 3.43MP sensor. Pixels. Photodetectors. Remember, they are not the same.

All you have to do is compare a Sigma SD9/10 photo with a Canon 1Ds photo. You tell me which has more resolution and clarity. And then even go a step further, and take crop from the Sigma photo, and a crop from a 1Ds photo. The results should be obvious.

--
http://www.digitaldingus.com
http://pub103.ezboard.com/bthedigitaldinguscommunity

 

[Joseph S Wisniewski]

Now - to be fair - both Foveon and Bayer needs anti alias
filters. If you don't have it, your pictures will look sharp
but weird at pixel level. There has been a lengthy discussion
about this in news:rec.photo.digital. A discussion where
some Foveon fan has shown weird photos of the moon and
other stuff, full of aliasing artefacts.

And it's nice to see Foveon finally admitting this. Apparently, the new SD-10 has an AA filter.

--
Ciao!

Joe

http://www.swissarmyfork.com

 

[Just Looking]

All you have to do is compare a Sigma SD9/10 photo with a Canon 1Ds
photo. You tell me which has more resolution and clarity. And then
even go a step further, and take crop from the Sigma photo, and a
crop from a 1Ds photo. The results should be obvious.

Have you ever done that test?

While you're at it, compare the SD9 picture to a 3MP and a 4 MP picture.

Stop speculating and just do it.

j