Olympus-OM
[Top] [All Lists]

[OM] E1, 4/3 format and Foveon

Subject: [OM] E1, 4/3 format and Foveon
From: Hughes <hi100@xxxxxxxxxxxxx>
Date: Wed, 25 Jun 2003 02:16:10 -0700 (PDT)
Dear Zuiks,
           earlier this year I attended an Institute of Electrical Engineers 
presentation here in
Silicon Valley, given by Foveon. There were some very interesting discussions 
relating to both
Foveon as well as digi imaging and the history of imaging in general.  It was 
particulalry
interesting hearing their perspective on aliasing and related issues, which I 
asked a couple of
questions about. For example the cost of adding an anti-aliasing optical filter 
is prohibitive,
almost as much as the sensor chip itself. The Sigma camera using the Foveon 
chip does not have use
such a filter, although of course, the Foveon chip does not need one nearly as 
much as similar
sized conventional chips.

       Their comments on the 4/3 system were, that they thought the idea of 
redefining camera/size
format was basically right and the sensor was a reasonable size but that the 
consortium had made a
major mistake in making the flange distance too large. My understanding of what 
they were
suggesting was that they should have got rid of the the pentaprism and mirror. 
If this is what
they meant, you then land up with a very nice small camera (Leica like) and 
smaller lenses since
they don't need to be retro-focus, but an electronic (read: difficult to 
manually focus)
viewfinder.

On sensor size, they felt that the current 1.5x-1.7x (equiv 35mm) size was the 
best compromise
between yield,cost,noise performance. They did not seem to have plans to make 
much larger chips or
much smaller chips. They said Sony is the leader in really small pixel sensors 
developed for video
cams with sizes down in the 1-2uM range. 

An interesting wrinkle mentioned, was that with lensed sensor chips you can 
angle the on pixel
lenses as you move away from the centre of the chip, to make up for non 
paraxial rays where light
comes from a bigger angle. Of course Foveon's sensor does not suffer nearly so 
much from the well
effects of metalization off axis, with their vertically stacked RGB sensors. 
They don't currently
use on chip lenses. This is also probably a big cost/yield issue as the then 
required extra
layers/steps increase opportunity for defects.  

Somebody on this list mentioned that the Foveon sensor has a low pixel count. 
This is not really
true, since Foveon counts a vertically stacked R+G+B "pixel" as one pixel while 
everybody else
counts their R,G,B pixels individually. What is stored to disk is usually 
interpolated data still
further muddying the "resolution".  For illustration a 3Meg Foveon chip would 
have 9Meg sensor
pixels in everybody elses counting system, so they are at the higher end of 
resolution/pixel
count. I would hate to be in Foveon's marketing dept and have to explain this 
everyday!

An interesting feature of the Oly E1 is the inclusion of digital lens 
correction data stored in
the lens itself. What it appears, is that they correct for linear distortion 
and probably
vignetting/light fallof, for each lens design. This is a great idea as now you 
have an extra
degree of freedom in the lens design. You could optimize the lens for 
resolution and then correct
for say horrible barrel distortion and still produce normal images. In fact, 
apart from resolution
a fisheye lens could then have dual function linear or fisheye at flick of 
switch.

Regards,
Tim hughes

< This message was delivered via the Olympus Mailing List >
< For questions, mailto:owner-olympus@xxxxxxxxxxxxxxx >
< Web Page: http://Zuiko.sls.bc.ca/swright/olympuslist.html >


<Prev in Thread] Current Thread [Next in Thread>
Sponsored by Tako
Impressum | Datenschutz