Digital BW, The Print

Having just come from a lecture by Carver Mead, CEO of Foveon, I think that its going to be the camera of the future.
I'm not tch inclined, but the gist of what he spoke about and showed slides is..
With traditonal CCD's each pixel records R G or B, then math algorythms, figure out 2/3rds of the missing information to construct the entire photo.  With the CCD's that we are used to, only 1/3 of picture info is real, the rest is mathematically figured out or guessed at.  With the Foveon silicon chip, it acts like a piece of Kodachrome film, where there are 3 color reading levels in layers, just like traditional film.  therefore no information has to be guessed at to fill in the entire photo, its all there.  The reason they say 3 times the quality of the conventional chip is that all of the RGB info is recorded, not 1.3 of the info, and then mathematiccaly guessed at to fill in the photo.
Hope some of this makes sense to you.
The new Sigma camera coming out with the X3 chip will be under $3000.00 and be the equivalent in megapixels , compared to actual megapixels in Nikon or Canon cameras, it will be 10.2 megapixels.  He said they were printing 48 inch prints from the Sigma, and compared it to the quality of single capture backs on a Hasselblad.
Camers should be available May 1st in stores.
Lets wait and see!
David

Hi David,

> Having just come from a lecture by Carver Mead, CEO of Foveon, I
> think that its going to be the camera of the future.
> I'm not tch inclined, but the gist of what he spoke about and
> showed slides is..

I'm sure it was a very entertaining lecture!  I would have loved to have
been there ;-)

> With traditonal CCD's each pixel records R G or B, then math
> algorythms, figure out 2/3rds of the missing information to
> construct the entire photo.

That is not how it works, and if that was what was presented, it is a
misrepresentation of how the Bayer reassimilation algorithm works.  It is
only the COLOR information that is interpolated, NOT the edge information.
In fact, television works in a very similar way...on a simple fact that our
eyes are far far less sensitive to color information then they are to edge
information.

> With the CCD's that we are used to,
> only 1/3 of picture info is real, the rest is mathematically
> figured out or guessed at.

Again, that is completely wrong.

> Hope some of this makes sense to you.

It makes complete sense to me.  What you believe you heard is not what
happens with Bayer pattern sensors.

You know they DO have a good idea, but they are creating absolutely bogus
straw men to fight against, and that, in my opinion, is not really "proper".

>  He said they were printing 48 inch prints from the Sigma, and
> compared it to the quality of single capture backs on a Hasselblad.

Really?  Did he SHOW you any of the supposed 48 inch prints?  The sensor
that was used in the DPReview camera was a 2k x 2k sensor, and the F7 sensor
is 2268 x 1512, and the F10 sensor is 1280 x 960.  Without interpolation,
assuming the 2k x 2k sensor, that would make a 48 inch print have only 42.6
pixels per inch.  I do not believe that.  I can't imagine how he can make
the claim he has, unless the image was run through something like Genuine
Fractals...

Austin

Austin Franklin wrote:

>
> Really?  Did he SHOW you any of the supposed 48 inch prints?  The sensor
> that was used in the DPReview camera was a 2k x 2k sensor, and the F7 
> sensor
> is 2268 x 1512, and the F10 sensor is 1280 x 960.  Without interpolation,
> assuming the 2k x 2k sensor, that would make a 48 inch print have only 
> 42.6
> pixels per inch.  I do not believe that.  I can't imagine how he can make
> the claim he has, unless the image was run through something like Genuine
> Fractals...
>

Well, this should lead to some interesting and esoteric marketing battles...

When you are not interpolating color OR edge data, what ratio of  
similarly sized pixels would be needed to achieve the same results with 
a chip that interpolates color on a 1/3 basis (Bayer Pattern), but does 
not interpolate edge values..?

The question comes down to an interesting issue of human visual 
perception.. generally color is less acutely perceived by the eye than 
is edge definition (it's part of our hard-wiring to pick up movement). 
 The question becomes one of  how many mPix on the new three layer chip 
would be equivalent to a 4 mPix Bayer pattern chip... Given our 
pre-wiring for edges I would say it's likely to be need to be well more 
than 1/2 the resolution of the Bayer CCD.

BUT, the debate does not end there.  Because what might not be visibly 
perceptible to us in output of the capture may well create artifacts 
that upon PhotoShop or other processing become intolerable.  Like that 
one orange pixel on a field of denim..  Therefore, it is quite likely 
that FOVEON chip captures may size up even say 3x as well as a Bayer CCD 
capture would (absent extensive filtering).  When you size up 
incorrectly interpolated points they increase in significance within the 
visual field by the simple reality of their increase in finite size.. 
 What might have been an issue of ratio to total image size is now one 
of fixed anomaly size..

[Keith]
 
 

****************************************************************
CONFIDENTIALITY & COPYRIGHT NOTICE:
This e-mail message, including attachments and contents, is © Copyright, 
Keith Krebs, 2001, All Rights Reserved. It is expressly for the sole use 
of the intended recipient and may contain confidential and privileged 
information. Absent the express written authorization of the author, any 
unauthorized review, use, disclosure, transfer, or distribution is 
explicitly prohibited and taken at your own risk. If you are not, or are 
unsure whether you are, the intended recipient, please contact the 
sender immediately and destroy all copies of the original message. 
Violations will be prosecuted to the FULL extent allowed under 
applicable civil and criminal law. Imagery published or distributed in 
violation of these conditions shall be subject to a $1500/image 
liquidated damages charge, in addition to any applicable Copyright 
violation penalties.

[POV IMage Service Banner]
****************************************************************
{ The P.O.V. Image Service Website is still at http://www.p-o-v-image.com/ }




[Non-text portions of this message have been removed]

Colour interpolation ISN'T the same as picture interpolation, and claiming 
that a 3.3 megapixel RGB sensor is equivalent to a 10 megapixel Bayer sensor 
is just plain wrong.

It's a characteristic of human vision that you can slash and burn the colour 
information without causing much damage to the overall picture.

TV depends upon it, JPEG depends upon it and four colour screen printed 
magazines depend upon it. 

Don't get me wrong, I think the X3 sensor looks like a significant advance, 
but I hate unnecessary hype.

On Tuesday 19 Mar 2002 11:39 pm, daschkenas@... wrote:
> Having just come from a lecture by Carver Mead, CEO of Foveon, I think that
> its going to be the camera of the future. I'm not tch inclined, but the
> gist of what he spoke about and showed slides is.. With traditonal CCD's
> each pixel records R G or B, then math algorythms, figure out 2/3rds of the
> missing information to construct the entire photo.  With the CCD's that we
> are used to, only 1/3 of picture info is real, the rest is mathematically
> figured out or guessed at.  With the Foveon silicon chip, it acts like a
> piece of Kodachrome film, where there are 3 color reading levels in layers,
> just like traditional film.  therefore no information has to be guessed at
> to fill in the entire photo, its all there.  The reason they say 3 times
> the quality of the conventional chip is that all of the RGB info is
> recorded, not 1.3 of the info, and then mathematiccaly guessed at to fill
> in the photo. Hope some of this makes sense to you.
> The new Sigma camera coming out with the X3 chip will be under $3000.00 and
> be the equivalent in megapixels , compared to actual megapixels in Nikon or
> Canon cameras, it will be 10.2 megapixels.  He said they were printing 48
> inch prints from the Sigma, and compared it to the quality of single
> capture backs on a Hasselblad. Camers should be available May 1st in
> stores.
> Lets wait and see!
> David
>
>
> Please visit the Group Homepage to check the Files, Bookmarks, Polls and
> other resources as they are often being updated. The page is at:
>
> http://groups.yahoo.com/group/DigitalBlackandWhiteThePrint
>
> Please follow these basic guidelines:
> - Include your full name with your message.
> - Include the address of your website, if you have one.
> - As threads develop, trim off excess portions of earlier messages to keep
> them short. - As the topic of a thread changes remember to change the
> subject header. - Good manners are required at all time. No personal
> attacks or "flames." - Complete your Yahoo profile.
> - Before posting a question, search the message archives and the various
> resources on the homepage.
>
>
>
>
> Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/

Er so what?

Medium format digital backs usually only have 6 megapixel sensors. 

On Tuesday 19 Mar 2002 11:39 pm, daschkenas@... wrote:
> \ufffdHe said they were printing 48 inch prints from the Sigma, and compared it
> to the quality of single capture backs on a Hasselblad.

This just came in from Digital Photo News, thought it might be of 
interest to the group
Gates backs Foveon
Microsoft billionaire Bill Gates declared his support for Foveon's
revolutionary new X3 sensor (BJP, 02 February) at this year's Photo
Marketing Association Annual Trade Show in Florida. In a pre-recorded
video clip played at the event, Gates publicly endorsed the new
technology, saying Microsoft is working with Foveon to provide native
support for X3 raw images from within the Windows XP operating system.
"Foveon's X3 sensor technology promises to take digital imaging to the
next level, with even higher quality images and new capabilities - well
beyond what is capable with today's film and digital camera technology -
available at a lower cost to users," he said. Currently, the technology
is available in two configurations - a small 6.4x4.8mm chip and the
20.7x13.8mm (APS size) sensor found in the Sigma SD-9. Rated at 3.43
million pixels, multiplied by three for each colour channel, the new
SD-9 provides colour resolution equivalent to a 10 million pixel
single-shot camera, says Foveon. The SD-9 is estimated to arrive in the
UK in June, with a price around £2000.
-- 
  ******* ********
Ostrom Photography
  ostrom@...
    ******  *******

> Gates publicly endorsed the new
> technology, saying Microsoft is working with Foveon to provide native
> support for X3 raw images from within the Windows XP operating system.

What the hell does that mean?  How are X3 "RAW IMAGES" any different than
any other images?  I thought that WAS the point with this sensor, that the
data was in standard 3 color pixel format, and required no "support"?

That is an unbelievably foolish statement, and is entirely laughable.  I can
only guess he bought stock in the company, and had to come up with something
to say...and that's the best he could do.

Austin

Carver Mead stated that the equivalent of a 4megapixel bayer type chip would 
be equal to a 12 MP Foveon chip, without any artifacts.  He also stated that 
they are currently working with the Adobe folks, so that in the near future 
Photoshop would recognize the Foveon system, and should work seamlessly.
David

> Carver Mead stated that the equivalent of a 4megapixel bayer type
> chip would
> be equal to a 12 MP Foveon chip, without any artifacts.  He also
> stated that
> they are currently working with the Adobe folks, so that in the
> near future
> Photoshop would recognize the Foveon system, and should work seamlessly.
> David

Ah, more tidbits of...well...misinformation.  Does anyone have ANY idea what
so ever why Adobe would need to "recognize the Foveon system"?  Why wouldn't
it just work seamlessly already?  It's just data in some image format...
There have been different sensor types out for 10+ years, both interline,
one shot and scanning...and there hasn't been ANY need to handle the data
any differently for these in the past 10 years, now, why, all of a sudden,
does someone make some claim that there is a "problem" that never existed
before?  And, IMO, doesn't exist today.

Austin

Spoken by someone that obviously doesn't have a clue what current digital camera 
'RAW'  format files mean or what can be done with them  when using the proper 
software.

I agree with others about this thread being clearly off topic for this group. Besides, 
it's getting boring. Like trying to pound nails into a steel plate.

There's clearly an agenda at work that doesn't belong here.

> Ah, more tidbits of...well...misinformation.  Does anyone have ANY
> idea what so ever why Adobe would need to "recognize the Foveon
> system"?  Why wouldn't it just work seamlessly already?  It's just
> data in some image format... There have been different sensor types
> out for 10+ years, both interline, one shot and scanning...and there
> hasn't been ANY need to handle the data any differently for these in
> the past 10 years, now, why, all of a sudden, does someone make some
> claim that there is a "problem" that never existed before?  And, IMO,
> doesn't exist today.

You can't be talking about me?  I have FAR more than a clue what the raw
data out of digital cameras is, and if you think the "Foveon system" needs
some special recognition, you are mistaken.   The ONLY thing RAW means is
the bit depth.

Austin

> Spoken by someone that obviously doesn't have a clue what current
> digital camera
> 'RAW'  format files mean or what can be done with them  when
> using the proper
> software.
>
> I agree with others about this thread being clearly off topic for
> this group. Besides,
> it's getting boring. Like trying to pound nails into a steel plate.
>
> There's clearly an agenda at work that doesn't belong here.
>
> > Ah, more tidbits of...well...misinformation.  Does anyone have ANY
> > idea what so ever why Adobe would need to "recognize the Foveon
> > system"?  Why wouldn't it just work seamlessly already?  It's just
> > data in some image format... There have been different sensor types
> > out for 10+ years, both interline, one shot and scanning...and there
> > hasn't been ANY need to handle the data any differently for these in
> > the past 10 years, now, why, all of a sudden, does someone make some
> > claim that there is a "problem" that never existed before?  And, IMO,
> > doesn't exist today.

> Spoken by someone that obviously doesn't have a clue what current
> digital camera
> 'RAW'  format files mean or what can be done with them  when
> using the proper
> software.

Let me clarify this so there is no misunderstanding.  There is absolutely no
relation between what you are calling the "CURRENT digital camera 'RAW'
format files" (which all are not the same), and what you will get as raw
data out of the Foveon.  This is assuming you are not talking about scanning
backs/cameras, but cameras with Bayer pattern CCDs.  If you understand the
structure of the current CCDs (the Bayer pattern) as well as how the Foveon
sensor is architected, it is obvious why.

The BIG "feature" with the Foveon chip is it uses three sensors per pixel,
and as such, eliminates the need to process the Bayer pattern RGBG sensor
data.  Given that, the Foveon SHOULD return full color values, as in each
pixel contains N bits of red, green and blue per pixel.  Given that, the raw
data files from the Foveon are just like those from a scanning digital
camera back (or for that matter, a film scanner), NOT like the raw data
files from Bayer pattern cameras...so my point is, the only thing that is
raw about the Foveon sensors returned data in raw mode is purely the pixel
values.  The need for the Bayer pattern processing is unnecessary and is
therefore eliminated, and therefore the raw data returned is entirely
different, and has no bearing on the "current" digital camera raw
format...as you apparently believe.

That is why I don't understand what special support would be required by
anyone to use these files, and why I believe the Microsoft claim of
"support" is bogus.

Austin

Er it isn't foolish if the sensor delivers more than 8 bits per colour as 
then a non-24 bit colour format will be needed like 36-bit TIFF, and it would 
be perfectly logical to call it a raw format even if it isn't Beyer data.

On Wednesday 20 Mar 2002 1:55 pm, Austin Franklin wrote:
> > Gates publicly endorsed the new
> > technology, saying Microsoft is working with Foveon to provide native
> > support for X3 raw images from within the Windows XP operating system.
>
> What the hell does that mean?  How are X3 "RAW IMAGES" any different than
> any other images?  I thought that WAS the point with this sensor, that the
> data was in standard 3 color pixel format, and required no "support"?
>
> That is an unbelievably foolish statement, and is entirely laughable.  I
> can only guess he bought stock in the company, and had to come up with
> something to say...and that's the best he could do.
>
> Austin

Hi Derek,

It clearly IS a raw format, no doubt...I have no problem with that, but for
Microsoft to claim they are "providing native support" for something that is
merely the same unprocessed pixel depth as, say, our raw scanner files,
seems a bit silly to me.  Seems more like hype than substance anyway.

What can he possibly mean by "support"?  The raw image data still needs the
setpoints set, and the tonal curves applied (PRNU)...and if wanted,
sharpening.  That seems more like an image processing program issue than an
OS issue, doesn't it?  What could one use the raw data for, without applying
the setpoints and curves etc.?

Regards,

Austin

> Er it isn't foolish if the sensor delivers more than 8 bits per colour as
> then a non-24 bit colour format will be needed like 36-bit TIFF,
> and it would
> be perfectly logical to call it a raw format even if it isn't Beyer data.
>
>
> On Wednesday 20 Mar 2002 1:55 pm, Austin Franklin wrote:
> > > Gates publicly endorsed the new
> > > technology, saying Microsoft is working with Foveon to provide native
> > > support for X3 raw images from within the Windows XP operating system.
> >
> > What the hell does that mean?  How are X3 "RAW IMAGES" any
> different than
> > any other images?  I thought that WAS the point with this
> sensor, that the
> > data was in standard 3 color pixel format, and required no "support"?
> >
> > That is an unbelievably foolish statement, and is entirely laughable.  I
> > can only guess he bought stock in the company, and had to come up with
> > something to say...and that's the best he could do.
> >
> > Austin

So as just one examle, you are saying that the Foveon technology does not need to be 
white balanced? Setpoints and curves adjusted?  

I'll bet that just like every other manufacturer, they use a proprietary file format for their 
'raw' files. Hence, the need for native support for that format. Just like jpg, tiff, eps, etc. 

Nearly every pro I know using pro quality digital today is shooting in 'raw' file mode. The 
advantages are just too great to ignore.

Greg

> > Spoken by someone that obviously doesn't have a clue what current
> > digital camera 'RAW'  format files mean or what can be done with
> > them  when using the proper software.
> 
> Let me clarify this so there is no misunderstanding.  There is
> absolutely no relation between what you are calling the "CURRENT
> digital camera 'RAW' format files" (which all are not the same), and
> what you will get as raw data out of the Foveon.  This is assuming you
> are not talking about scanning backs/cameras, but cameras with Bayer
> pattern CCDs.  If you understand the structure of the current CCDs
> (the Bayer pattern) as well as how the Foveon sensor is architected,
> it is obvious why.
> 
> The BIG "feature" with the Foveon chip is it uses three sensors per
> pixel, and as such, eliminates the need to process the Bayer pattern
> RGBG sensor data.  Given that, the Foveon SHOULD return full color
> values, as in each pixel contains N bits of red, green and blue per
> pixel.  Given that, the raw data files from the Foveon are just like
> those from a scanning digital camera back (or for that matter, a film
> scanner), NOT like the raw data files from Bayer pattern cameras...so
> my point is, the only thing that is raw about the Foveon sensors
> returned data in raw mode is purely the pixel values.  The need for
> the Bayer pattern processing is unnecessary and is therefore
> eliminated, and therefore the raw data returned is entirely different,
> and has no bearing on the "current" digital camera raw format...as you
> apparently believe.
> 
> That is why I don't understand what special support would be required
> by anyone to use these files, and why I believe the Microsoft claim of
> "support" is bogus.
> 
> Austin
> 
> 
> ------------------------ Yahoo! Groups Sponsor
> ---------------------~--> Access Your PC from Anywhere It's Easy. It's
> Fast. - Free Download
> http://us.click.yahoo.com/AxtVhB/6XkDAA/_ZuFAA/ucIolB/TM
> ---------------------------------------------------------------------~
> ->
> 
> Please visit the Group Homepage to check the Files, Bookmarks, Polls
> and other resources as they are often being updated. The page is at:
> 
> http://groups.yahoo.com/group/DigitalBlackandWhiteThePrint
> 
> Please follow these basic guidelines:
> - Include your full name with your message.
> - Include the address of your website, if you have one.
> - As threads develop, trim off excess portions of earlier messages to
> keep them short. - As the topic of a thread changes remember to change
> the subject header. - Good manners are required at all time. No
> personal attacks or "flames." - Complete your Yahoo profile. - Before
> posting a question, search the message archives and the various
> resources on the homepage. 
> 
> 
> 
> 
> Your use of Yahoo! Groups is subject to
> http://docs.yahoo.com/info/terms/ 
> 
> 
>

Hi Greg,

> So as just one examle, you are saying that the Foveon technology
> does not need to be
> white balanced? Setpoints and curves adjusted?

No, not at all.  I am saying the raw data from the Foveon camera is not
[your list above], but that is true for even the Bayer pattern sensors...but
with the Bayer pattern sensors, the data still has to have the color
interpolation done, but with the Foveon it does not.  It is the need for the
Bayer interpolation that causes the raw image data from these cameras to
have to be treated "special", and that's eliminated with the Foveon.

> I'll bet that just like every other manufacturer, they use a
> proprietary file format for their
> 'raw' files.

Well, they kind of do.  It's Run Length Encoded (RLE).  These formats aren't
really proprietary, in the sense that they're "secret", they just have to be
decoded.  RLE is very very simple to decode.  It's simply a list of two
numbers, first number being the count and the second number being the value.
I'm somewhat surprised they even do it at all.

I really wish they would just use TIFF with the Foveon.  With the Foveon,
using TIFF makes sense, since it's exactly like raw data from a scanner,
unlike the Bayer pattern sensors, which is simply single color information
per what ever you call a unit of a single color, as it isn't really a
pixel...and there's no official name for it...

> Nearly every pro I know using pro quality digital today is
> shooting in 'raw' file mode. The
> advantages are just too great to ignore.

What software do "they" use to do the Bayer pattern interpolation?  But to
go back to my original point on this particular discussion, the raw data
from the Foveon is entirely different than the raw data from a Bayer pattern
camera, since you don't have to do any interpolation, and every pixel does
contain full color information.

Austin

Austin Franklin wrote:

>
> What can he possibly mean by "support"?  The raw image data still 
> needs the
> setpoints set, and the tonal curves applied (PRNU)...and if wanted,
> sharpening.  That seems more like an image processing program issue 
> than an
> OS issue, doesn't it?  What could one use the raw data for, without 
> applying
> the setpoints and curves etc.?
>
>
Two possibilities...


We know Gates wants to push embedded Windows OS's, the new tablet OS 
version of Windows, and whatever they call the PDA OS version these days..

1)   If he provides built in support (within the OS) to do the RAW data 
acquisition, set points, curve adjustment, etc... what does this do?  It 
means vendors can build the device into a product and let the OS deal 
with most of the interface issues..  That would make Windoze products 
the OS of choice for products incorporating the chip.. Cell Phones, 
PDAs, tablets, laptops, etc..

2)    Slightly different, but along the same lines...  One could 
potentially use USB or firewire FOVEON webcams etc with very little 
hardware.  Just Plug it into a port of a Win machine and voila!

The interesting thing is that a move like this would allow what is  now 
firmware to be patched (and hacked) much more simply..  With all the 
processing power in most cpus these days, it may make sense to take the 
minor hit that conversion of the data represents, instead of doing ANY 
on device processing..  Heck, you could use new sharpening algorithms as 
part of the acquisition workflow, simply by updating the acquisition 
software on your machine...  Much more elegant than doing it in firmware 
and on-board the device..

Obviously, this is less an advantage in devices like cameras, not 
tethered to a Windoze box or not using a Windows embedded OS..  

[Keith]
 
 

****************************************************************
CONFIDENTIALITY & COPYRIGHT NOTICE:
This e-mail message, including attachments and contents, is © Copyright, 
Keith Krebs, 2001, All Rights Reserved. It is expressly for the sole use 
of the intended recipient and may contain confidential and privileged 
information. Absent the express written authorization of the author, any 
unauthorized review, use, disclosure, transfer, or distribution is 
explicitly prohibited and taken at your own risk. If you are not, or are 
unsure whether you are, the intended recipient, please contact the 
sender immediately and destroy all copies of the original message. 
Violations will be prosecuted to the FULL extent allowed under 
applicable civil and criminal law. Imagery published or distributed in 
violation of these conditions shall be subject to a $1500/image 
liquidated damages charge, in addition to any applicable Copyright 
violation penalties.

[POV IMage Service Banner]
****************************************************************
{ The P.O.V. Image Service Website is still at http://www.p-o-v-image.com/ }




[Non-text portions of this message have been removed]

Of course the numbers returned from the sensor will probably need some kind 
of spectral tuning before they would make sense so a raw format would still 
apply.

On Wednesday 20 Mar 2002 4:53 pm, Austin Franklin wrote:

> The BIG "feature" with the Foveon chip is it uses three sensors per pixel,
> and as such, eliminates the need to process the Bayer pattern RGBG sensor
> data.  Given that, the Foveon SHOULD return full color values, as in each
> pixel contains N bits of red, green and blue per pixel.  Given that, the
> raw data files from the Foveon are just like those from a scanning digital
> camera back (or for that matter, a film scanner), NOT like the raw data
> files from Bayer pattern cameras...so my point is, the only thing that is
> raw about the Foveon sensors returned data in raw mode is purely the pixel
> values.  The need for the Bayer pattern processing is unnecessary and is
> therefore eliminated, and therefore the raw data returned is entirely
> different, and has no bearing on the "current" digital camera raw
> format...as you apparently believe.