Re: Sigma SD9 - from the show floor

You keep on saying X x Y x 3 X3 sensor.
Are you not repeating yourself.

The X3 is supposed to be the symbol that represents the new sensor, not the number of pixels.
This has appeared in the news headlines and the news article.
--...Lyall

Yes, the phrase "3.54 x 3 million pixel X3 sensor" is totally bogus. (For a start, if anything it should be "3.54 million x 3 pixel X3 sensor" but even that's not justified.)

Some might consider it approrpriate to call the sensor in, say, the Canon D-30 a "2160 x 1440 1/3rd-pixel sensor", but just because no-one does, you can't magically triple the number of pixels on an X3. A pixel is a pixel, no matter many (or few) primaries it captures.

(BTW, the I-R folks are just as guilty of this kind of nonsense. Their first story on the X3 had exactly the same logic-defying aithmetic.)

Cheers,
Pete

Lyall Pearce said:

You keep on saying X x Y x 3 X3 sensor.
Are you not repeating yourself.
The X3 is supposed to be the symbol that represents the new sensor,
not the number of pixels.
This has appeared in the news headlines and the news article.

--
...Lyall

Click to expand...

-- http://www.peter-cockerell.net:8080/

Pete,

In actual fact, the X3 is not totally bogus, it is the Symbol that Foveon use to signify their sensor. (A bit like Microsoft using XP to identify their Operating System)

What is bogus, however, is the 2160x1440x3 X3 - people are repeating themselves. (Do people write Microsoft Windows XP XP?)

I think Foveon really made a mistake with their symbol, it causes lots of confusion. Maybe they should have used 3 horizontal bars one for each primary colour instead - that would have prevented this x3 misnomer from surfacing)

Pete Cockerell said:

Some might consider it approrpriate to call the sensor in, say, the
Canon D-30 a "2160 x 1440 1/3rd-pixel sensor", but just because
no-one does, you can't magically triple the number of pixels on an
X3. A pixel is a pixel, no matter many (or few) primaries it
captures.

(BTW, the I-R folks are just as guilty of this kind of nonsense.
Their first story on the X3 had exactly the same logic-defying
aithmetic.)

Cheers,
Pete

Lyall Pearce said:

You keep on saying X x Y x 3 X3 sensor.
Are you not repeating yourself.
The X3 is supposed to be the symbol that represents the new sensor,
not the number of pixels.
This has appeared in the news headlines and the news article.

--
...Lyall

Click to expand...

--
http://www.peter-cockerell.net:8080/

Click to expand...

--...Lyall

Oh, don't get me wrong. Foveon can call their technology what they like. X3 is as good a name as any. What I don't like is "x 3" where "x" is being used as a multiplication operator. This has been more of a problem with the news web sites than Foveon itself, e.g Imaging Resource's:

"Foveon has initially announced two products based on its X3 design; the F7 and F10 image sensors - part numbers F7-35X3-A25B and F10-14X3-D08A respectively. The F7 is a 2304 x 1536 pixel (2268 x 1512 effective) chip that would be a 3.43 megapixel sensor in terms of dimensions of the image alone, but actually records 10.29 megapixels of color data."

Total nonsense. A pixel is a spatial location on the sensor (it stands for "picture element" after all), not some fuzzy concept whose count can be manipulated arbitrarily.

Anyway, that was my point...

Cheers,
Pete

Lyall Pearce said:

Pete Cockerell said:

Some might consider it approrpriate to call the sensor in, say, the
Canon D-30 a "2160 x 1440 1/3rd-pixel sensor", but just because
no-one does, you can't magically triple the number of pixels on an
X3. A pixel is a pixel, no matter many (or few) primaries it
captures.

(BTW, the I-R folks are just as guilty of this kind of nonsense.
Their first story on the X3 had exactly the same logic-defying
aithmetic.)

Cheers,
Pete

Lyall Pearce said:

You keep on saying X x Y x 3 X3 sensor.
Are you not repeating yourself.
The X3 is supposed to be the symbol that represents the new sensor,
not the number of pixels.
This has appeared in the news headlines and the news article.

--
...Lyall

Click to expand...

--
http://www.peter-cockerell.net:8080/

Click to expand...

--
...Lyall

Click to expand...

-- http://www.peter-cockerell.net:8080/

I agree it may sound a bit confusing but only because it is the name of the sensor, X3. Maybe it should say '3.43 million x 3 pixel Foveon X3 Sensor' And no, there is nothing deceitful about it. Foveon's 3.34 million pixel chip is made up of three very thin blue, green and red layers each with their own photodetectors, were as a traditional chip has a single layer made up of a mosiac design of red blue and green pixels with only 1 photodetector. Which means only 1/3 of the color is actually captured, the other colors must be interpolated. This usually means loss of detail and color artifacts. Some better than others, however.

If anything, the Foveon X3 chip is the only one not lying about the true amount of pixels being used. All other sensors have to interpolate. In my opinion, Foveon got it right. However, I'm not thrilled about the Sigma camera either. The results are stunning though.

Mike

Michael

You have fallen into the same trap that I am talking about.

'3.43 million x 3 pixel Foveon X3 Sensor'
should read
'3.43 million pixel Foveon X3 Sensor'

It is not 3.43 times 3, it is 3.43 'the symbol X3' Sensor

Anyway, enough about that. I was only commenting because it appears Phil has fallen into the same trap as yourself, to quote his news headline :- (see the main dp review page)

'The SD9 digital SLR is fitted with a 3.54 x 3 million pixel X3 sensor which produces a 2268 x 1512 x 3 image (captures RAW in-camera).'

I believe it should be written as

'The SD9 digital SLR is fitted with a 3.54 million pixel X3 sensor which produces a 2268 x 1512 image (captures RAW in-camera).'

Phil is actually propogating the confusion, rather than alleviating it. I suppose it is the excitement of the PMA...

michael collins said:

I agree it may sound a bit confusing but only because it is the
name of the sensor, X3. Maybe it should say '3.43 million x 3
pixel Foveon X3 Sensor' And no, there is nothing deceitful about
it.

Mike

Click to expand...

--...Lyall

Call it whatever you or they want. Have any of you guys in this thread seen a large print from the x3 chip? The prints in the booth that are as large as 30 x 30 or larger far surpass anything that could have been produced from a regular 3.54 mp camera file. That said, we can't just compare it on pixel resolution alone because it is like comparing apples and oranges. If you don't believe me, just run on down here to Orlando and check it out.
Wes

Maybe we'll call it y3 so everyone isn't confused. Y Not?

Lyall Pearce said:

You have fallen into the same trap that I am talking about.

'3.43 million x 3 pixel Foveon X3 Sensor'
should read
'3.43 million pixel Foveon X3 Sensor'

It is not 3.43 times 3, it is 3.43 'the symbol X3' Sensor

Anyway, enough about that. I was only commenting because it appears
Phil has fallen into the same trap as yourself, to quote his news
headline :- (see the main dp review page)
'The SD9 digital SLR is fitted with a 3.54 x 3 million pixel X3
sensor which produces a 2268 x 1512 x 3 image (captures RAW
in-camera).'

I believe it should be written as

'The SD9 digital SLR is fitted with a 3.54 million pixel X3 sensor
which produces a 2268 x 1512 image (captures RAW in-camera).'

Phil is actually propogating the confusion, rather than alleviating
it. I suppose it is the excitement of the PMA...

michael collins said:

I agree it may sound a bit confusing but only because it is the
name of the sensor, X3. Maybe it should say '3.43 million x 3
pixel Foveon X3 Sensor' And no, there is nothing deceitful about
it.

Mike

Click to expand...

--
...Lyall

Click to expand...

Wes,

I was not comparing image quality or anything else for that matter, I was just commenting on the way people write the sensor spec. People are writing "123x456x3 X3 sensor", not "123x456 X3 sensor".

Yes, I agree, call it Y3 and the confusion goes away! It gets written as 123x456 Y3 sensor. No confusion there!

Wes Siebe said:

Call it whatever you or they want. Have any of you guys in this
thread seen a large print from the x3 chip? The prints in the booth
that are as large as 30 x 30 or larger far surpass anything that
could have been produced from a regular 3.54 mp camera file. That
said, we can't just compare it on pixel resolution alone because it
is like comparing apples and oranges. If you don't believe me, just
run on down here to Orlando and check it out.
Wes

Maybe we'll call it y3 so everyone isn't confused. Y Not?

Click to expand...

--...Lyall

I'm with you Wes. Yes I was there to see the prints. They were totally awsome. I guess the point Lyall is trying to make is that it isn't a 3.54 x 3 number of pixels. No, in the end only 3.43 (effective) million pixels wiil be used. But because no interpolation of color is used, you do get the results equal to a 9+ million pixel chip. Believe me, the results look that good.

Traditional CCD =i nterpololation (color & sharpness loss). Foveon X3 + no interpolation. Who's really falling into a trap here? But having said that, I still love my D1X and don't plan on switching for awhile. I'm sure in a year or two they'll have some new technology that blows all this new stuff away.

Mike

More can be learned here: http://www.foveon.com

Well, what Lyall said. I can never understand why people have to interpret a cirticism of one aspect of (say) the way a chip is being presented as dissing the chip itself. No-one is doubting the quality of the images produced by the Foveon sensors (which doesn't mean there aren't some reservations), but the point is, to call it a 10M-pixel chip is just BS, and no-one is going to redefine the meaning of the word 'pixel' just by pointing out that Bayer sensors only captures one primary per pixel. This isn't news. The F7 has 3,429,216 (visible) pixels, they're just much nicer pixels than the average Bayer-mask sensor. End of story (he says, hopefully).

Cheers,
Pete

Wes Siebe said:

Maybe we'll call it y3 so everyone isn't confused. Y Not?

Lyall Pearce said:

You have fallen into the same trap that I am talking about.

'3.43 million x 3 pixel Foveon X3 Sensor'
should read
'3.43 million pixel Foveon X3 Sensor'

It is not 3.43 times 3, it is 3.43 'the symbol X3' Sensor

Anyway, enough about that. I was only commenting because it appears
Phil has fallen into the same trap as yourself, to quote his news
headline :- (see the main dp review page)
'The SD9 digital SLR is fitted with a 3.54 x 3 million pixel X3
sensor which produces a 2268 x 1512 x 3 image (captures RAW
in-camera).'

I believe it should be written as

'The SD9 digital SLR is fitted with a 3.54 million pixel X3 sensor
which produces a 2268 x 1512 image (captures RAW in-camera).'

Phil is actually propogating the confusion, rather than alleviating
it. I suppose it is the excitement of the PMA...

michael collins said:

I agree it may sound a bit confusing but only because it is the
name of the sensor, X3. Maybe it should say '3.43 million x 3
pixel Foveon X3 Sensor' And no, there is nothing deceitful about
it.

Mike

Click to expand...

--
...Lyall

Click to expand...

Click to expand...

-- http://www.peter-cockerell.net:8080/

michael collins said:

If anything, the Foveon X3 chip is the only one not lying about the
true amount of pixels being used. All other sensors have to
interpolate. In my opinion, Foveon got it right.

Click to expand...

I agree. The other manufacturers claiming to have 6 million pixels or so... it's simply not true. Maybe better call them "3000 x 2000 BS pixels"?

I think that stiil right! There are 3 times X x Y pixels read from this X3 sensor: for R, G, and B, totalling 3x 3.5 Mp = 10.5M 12bit-words of information. That's why frame rate may be a bit limited at this time...
Jake.

Lyall Pearce said:

You keep on saying X x Y x 3 X3 sensor.
Are you not repeating yourself.
The X3 is supposed to be the symbol that represents the new sensor,
not the number of pixels.
This has appeared in the news headlines and the news article.

--
...Lyall

Click to expand...

--Jake.

Jake said:

Lyall Pearce said:

You keep on saying X x Y x 3 X3 sensor.
Are you not repeating yourself.
The X3 is supposed to be the symbol that represents the new sensor,
not the number of pixels.
This has appeared in the news headlines and the news article.

Click to expand...

Click to expand...

I interpret X3 as being the name of the product range, therefore when specifying its image sampling capability I consider it perfectly appropriate to state: X x Y x 3, in order to distinguish it from a conventional mosaic sensor with a specification of X x Y. The extra 'x 3' seems to me to be an accurate and honest way of making the distinction, better than actually stating the product of X x Y x 3 (10.5 Mpixel in the case of the X3) which is highly misleading. The alternative would be to change the manner in which conventional mosaic sensors are specified, as in X x Y x 1/3 (as has already been suggested in this thread), which is not going to happen.

Jake, you raise the issue of frame rate -- I have been told directly by Sigma that the frame rate of the SD-9 will be 3.5 fps, ie. as fast or even faster than its competitors, the Canon D60, Nikon D100 and the Fuji S2 Pro. This figure has yet to be confirmed by Phil.

Terry.

Jake said:

I think that stiil right!

Click to expand...

Yes, but that's because you don't know what a pixel is.

Jack40001 said:

There are 3 times X x Y pixels read from
this X3 sensor: for R, G, and B, totalling 3x 3.5 Mp = 10.5M

Click to expand...

That would be true if the separate colors of a pixel were also called pixels, but they're not. They're called "components" or "samples" or "bands", depending on whose terminology you're using. But not pixels. Really, trust me on this. It doesn't matter how many times you say it, it won't make it true. Even the fact that Bayer sensors only have one component per pixel doesn't make a difference. As I've already said, choose to call them "partial pixels" or somesuch if it makes you feel better, but don't start saying the X3 sites are anything greater than a pixel, because they're not. Can we move on now?

Jack40001 said:

12bit-words of information. That's why frame rate may be a bit
limited at this time...

Click to expand...

No argument there.

Jack40001 said:

Jake.

Lyall Pearce said:

You keep on saying X x Y x 3 X3 sensor.
Are you not repeating yourself.
The X3 is supposed to be the symbol that represents the new sensor,
not the number of pixels.
This has appeared in the news headlines and the news article.

--
...Lyall

Click to expand...

--
Jake.

Click to expand...

-- http://www.peter-cockerell.net:8080/

Lyall Pearce said:

Michael

You have fallen into the same trap that I am talking about.

'3.43 million x 3 pixel Foveon X3 Sensor'
should read
'3.43 million pixel Foveon X3 Sensor'

Click to expand...

I think '3.43 million x 3 pixel Foveon X3 Sensor' is right.
There are read: '3.43 million x 3 pixel values, totalling 10.29M words of data

representing 10.29M values for the whole image, where in a Bayer pattern only 1/3 would be read. The the technology is named X3, see it as a name.

You don't need to multiply again by 3 because of the name of the technology, but you should do it once as the technolgy name suggests.

I considder it as very right to speak of: '3.43 million x 3 pixel Foveon X3 Sensor' although '3.43 million x 3 pixel Foveon Sensor' would also be right

Jake.

Lyall Pearce said:

It is not 3.43 times 3, it is 3.43 'the symbol X3' Sensor

Anyway, enough about that. I was only commenting because it appears
Phil has fallen into the same trap as yourself, to quote his news
headline :- (see the main dp review page)
'The SD9 digital SLR is fitted with a 3.54 x 3 million pixel X3
sensor which produces a 2268 x 1512 x 3 image (captures RAW
in-camera).'

I believe it should be written as

'The SD9 digital SLR is fitted with a 3.54 million pixel X3 sensor
which produces a 2268 x 1512 image (captures RAW in-camera).'

Phil is actually propogating the confusion, rather than alleviating
it. I suppose it is the excitement of the PMA...

michael collins said:

I agree it may sound a bit confusing but only because it is the
name of the sensor, X3. Maybe it should say '3.43 million x 3
pixel Foveon X3 Sensor' And no, there is nothing deceitful about
it.

Mike

Click to expand...

--
...Lyall

Click to expand...

--Jake.

Terry Sessford said:

I interpret X3 as being the name of the product range, therefore
when specifying its image sampling capability I consider it
perfectly appropriate to state: X x Y x 3, in order to distinguish
it from a conventional mosaic sensor with a specification of X x Y.

Click to expand...

Why? If X3 is sufficient to distinguish the kind of sensor, why use the misleading "x 3" as well (or instead, it's not clear)?

Member said:

The extra 'x 3' seems to me to be an accurate and honest way of
making the distinction, better than actually stating the product of
X x Y x 3 (10.5 Mpixel in the case of the X3) which is highly
misleading.

Click to expand...

No, because people will do the math anyway and come up with 10.5M pixels. It also implies that Bayer sensors can produce no usable information for the two uncaptured components, which is clearly not true by the time the image is ever displayed. I think it's next to impossible to put a numerical figure on what equivalence a give sized X3 is to a larger-sized Bayer sensor, because, as we know, even Bayer sensors of the same pixel count can differ dramatically in apparent resolution.

This "x 3" approach also ignores the (admittedly fairly rare) existing sensors that don't use a Bayer mask, e.g. multi-sensor devices, multi-exposure devices, and monochrome devices. I don't remember anyone claiming those have three times the number of pixels as a Bayer sensor.

Member said:

The alternative would be to change the manner in which
conventional mosaic sensors are specified, as in X x Y x 1/3 (as
has already been suggested in this thread), which is not going to
happen.

Click to expand...

Right.

Cheers,
Pete-- http://www.peter-cockerell.net:8080/

Pixel

any of the small discrete elements that together constitute an image (as on a television screen)

any of the detecting elements of a charge-coupled device used as an optical sensor
(from Mirriam Webster Dictionary)

A combination of PIX [picture] and Element, meaning the smallest addressable unit on a display screen. The higher the pixel resolution (the more rows and columns of pixels), the more information can be displayed.

In storage, pixels are made up of one or more bits. The greater this "bit depth," the more shades or colors can be represented. The most economical system is monochrome, which uses one bit per pixel (on/off). Gray scale and color displays typically use from 4 to 24 bits per pixel, providing from 16 to 16 million colors.

On the screen of a CRT or LCD, pixels are made up of three dots of color RGB. Monochrome and gray scale systems use one dot per pixel. Pixels are energized with different intensities, creating a range from dark to light. creating a range of colors (perceived as the mixture of thee dots). Black is all three dots off, white is all dots on at max.

IBM called them PELs, Picture Elements, but their terminology never caught on, by then IBM had lost its clout in the PC world, its last big name claim being VGA and to a lesser extent in the display market XGA.

Jake.

Pete Cockerell said:

Jake said:

I think that stiil right!

Click to expand...

Yes, but that's because you don't know what a pixel is.

pcockerell said:

There are 3 times X x Y pixels read from
this X3 sensor: for R, G, and B, totalling 3x 3.5 Mp = 10.5M

Click to expand...

That would be true if the separate colors of a pixel were also
called pixels, but they're not. They're called "components" or
"samples" or "bands", depending on whose terminology you're using.
But not pixels. Really, trust me on this. It doesn't matter how
many times you say it, it won't make it true. Even the fact that
Bayer sensors only have one component per pixel doesn't make a
difference. As I've already said, choose to call them "partial
pixels" or somesuch if it makes you feel better, but don't start
saying the X3 sites are anything greater than a pixel, because
they're not. Can we move on now?

pcockerell said:

12bit-words of information. That's why frame rate may be a bit
limited at this time...

Click to expand...

No argument there.

pcockerell said:

Jake.

Lyall Pearce said:

You keep on saying X x Y x 3 X3 sensor.
Are you not repeating yourself.
The X3 is supposed to be the symbol that represents the new sensor,
not the number of pixels.
This has appeared in the news headlines and the news article.

--
...Lyall

Click to expand...

--
Jake.

Click to expand...

--
http://www.peter-cockerell.net:8080/

Click to expand...

--Jake.

Pete Cockerell said:

Terry Sessford said:

I interpret X3 as being the name of the product range, therefore
when specifying its image sampling capability I consider it
perfectly appropriate to state: X x Y x 3, in order to distinguish
it from a conventional mosaic sensor with a specification of X x Y.

Click to expand...

Why? If X3 is sufficient to distinguish the kind of sensor, why use
the misleading "x 3" as well (or instead, it's not clear)?

Click to expand...

Because your average punter going into a shop will probably not know the significance of the name X3, and I really do not see what is at all misleading about including "x 3" after the pixel count to indicate that these pixels are in some way different ('better'), which is perfectly true in terms of their information content.

pcockerell said:

pcockerell said:

The extra 'x 3' seems to me to be an accurate and honest way of
making the distinction, better than actually stating the product of
X x Y x 3 (10.5 Mpixel in the case of the X3) which is highly
misleading.

Click to expand...

No, because people will do the math anyway and come up with 10.5M
pixels. It also implies that Bayer sensors can produce no usable
information for the two uncaptured components, which is clearly not
true by the time the image is ever displayed.

Click to expand...

No, because we are talking purely about the specification of the sensor, not about any post processing of the sensor's output.

pcockerell said:

I think it's next to
impossible to put a numerical figure on what equivalence a give
sized X3 is to a larger-sized Bayer sensor, because, as we know,
even Bayer sensors of the same pixel count can differ dramatically
in apparent resolution.

Click to expand...

I agree with you here -- it is extremely difficult to quantitatively compare an X3 sensor with a Bayer sensor. In this regard I am very eager to see objective reviews by Phil and others of the Sigma SD-9 vis-a-vis the Canon D60, Nikon D100, etc. Until these reviews appear all discussion is mere speculation.

Terry.

pcockerell said:

This "x 3" approach also ignores the (admittedly fairly rare)
existing sensors that don't use a Bayer mask, e.g. multi-sensor
devices, multi-exposure devices, and monochrome devices. I don't
remember anyone claiming those have three times the number of
pixels as a Bayer sensor.

pcockerell said:

The alternative would be to change the manner in which
conventional mosaic sensors are specified, as in X x Y x 1/3 (as
has already been suggested in this thread), which is not going to
happen.

Click to expand...

Right.

Cheers,
Pete
--
http://www.peter-cockerell.net:8080/

Click to expand...

The link to dust protection filter picture is broken

Wes Siebe said:

The prints in the booth
that are as large as 30 x 30 or larger far surpass anything that
could have been produced from a regular 3.54 mp camera file.

Click to expand...

Yes this disturbs me...does anyone really believe those 30 x 30 prints where made from a rectangular sensor and then cropped?

Not likely so the prints we see from the booth are not what we can expect from the Sigma camera.