Comments at bottom.
At 3:56 AM +0000 12/6/03, olympus-digest wrote:
>Date: Fri, 5 Dec 2003 20:12:50 -0500 (EST)
>From: "Alienspecimen" <alienspecimen@xxxxxxxxxx>
>Subject: Re: [OM] E-1 4/3 vs. full sensor, read it, this is an order
>
>Sorry Skip,
>I was waaay off on this one. Dugged out the mag and here is what I found:
>The photographer's name is Stephen Johnson (www.sjphoto.com)
>The article is called New Eye on National Parks and here is what it says:
>As a pioneer in digital imaging, Johnson has been using various methods of
>digital capture for several years. The outfit he employs now consists of a
>specially developed digital back fitted to a Sinar Bron view camera, with
>possible resolution of 48 MP, generating nearly 150 MB files. That setup
>gives him a dynamic range that dramatically outstrips film's usual five stops.
>as well as extremely accurate color rendition. His digital camera represents
>the latest tool in Johnson's long term struggle to be freed from the confines
>of film and the stylistic choices film forces upon us. No longer limited to
>shooting at the edges of the day in warm light, Johnson can make beautiful
>images at times and in conditions that would send many film photographers
>looking for a place to wait for the lad-day light.
>
>From the previous page:
>
>The digital camera has a hight dynamic rang-nine to 11 stops versus film's
>five to seven and accurate color rendition, which allows him to continue down
>a path he had been on since the late 70"s with film, "of tring to be perfectly
>straight with the color and finding the romance and beauty in what was there,
>rather than in Warm light". In addition, it captures an amazing 6000X8000
>pixels, equivalent to a 48MP camera capable of seeing outstandig detail in
>very large prints.
>
>I am yet to visit his website...been busy talking about SUV's and politics...I
>kid, I kid...I spend most of my time looking for a job
>Boris
It's a very interesting website, and the pictures are spectacular. Thanks.
The camera specs are at <http://www.betterlight.com/emodels_info.asp>.
The Model 6000E seems to be the one in question. It's a "filmpack" that fits
into a Sinar Bron 4x5 view camera and contains a mechanically scanned Kodak
trilinear CCD array. A "trilinear" CCD contains three parallel linear (1-D)
CCD imagers, one per color. This array is physically scanned perpendicular the
parallel linear arrays, and electronically scanned along the arrays. It takes
66 seconds to take a full-resolution picture, although the exposure time per
pixel is far shorter. The scan will be along the long direction (to allow a
shorter CCD to be used), so the effective exposure time cannot exceed 66/8000=
0.0083 second, or about 1/121 sec, so it's probably 1/125 sec nominal. The
"film speed" is 100 to 1600 ISO.
The output is 6000x8000 pixels, each having all three colors, so in marketing
megapixels, this is a 144 Mpix camera. And here we are, arguing about 3 to 12
Mpix in the E1 versus the world.
Each pixel is full RGB, using a 12-bit analog-digital converter (ADC) per
color, but the internal datapaths and image software computations are 14-bit,
and one can store images in either 8-bit or 16-bit format. The dynamic range
is 11 bits. The bit about using 14-bit paths to handle 12-bit data is
standard, the intent being to cause roundoff errors in the image computations
to contaminate only noise bits, thus preserving the image data to the best
practical degree. Losing only one bit (12 bits becoming 11 bits) is a very
good result. This 11-bit dynamic range is what I have called "instantaneous
dynamic range" in earlier postings.
The file size is qouted at 137 MB, but this cannot be for the full-resolution
images: (6000)(8000)(3)(12)/(8*1024^2)= 206 MB, if the bits are packed as
tightly as possible. If one instead chooses 16-bit format, each image file
will be (6000)(8000)(3)(16)/(8*1024^2)= 275 MB= 0.268 GB. No wonder it comes
with a 9-GB disk drive, which will hold 9/0.268= 33.6 such images, at least in
theory. So, in round numbers, a 9GB drive is more or less equal to a
36-exposure 35mm roll. A very large and fast computer will be required to
photoshop such images.
The scan area is 72 by 96 mm, so each pixel is 72/6000= 12 microns by 96/8000=
12 microns. Because the trilinear CCD is moving, the effective light
collection may be somewhat less that one would expect from the 12-micron square
pixel size, but the camera datasheet doesn't address the issue.
The above specs are a perfect match for the Kodak KLI-6013 Image Sensor, whose
pixels saturate at 275,000 electrons, with a claimed 72-db dynamic range, or
11.96 bits, almost the full 12 bits.
For comparison, the Oly E-1 uses the KAF-5101CE, whose pixels saturate at
41,000 electrons, about 1/7th as many, so the E-1 will be far more vulnerable
to image noise. The claimed dynamic range is 67 db, or 11.13 bits. The
11.96-11.13= 0.83 bit difference is due to greater internal and shot noise in
the smaller sensor; both would use a 12-bit ADC.
This is quite the camera.
Joe Gwinn
< 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 >
|