Introduction
The claim I am offering herein, that Roger Patterson had a 15mm lens on his camera that day at Bluff Creek, will
be subjected to rigorous examination and challenge by others in the coming months. I anticipate this and given the potential of how
it reframes the whole discussion of the subject in the film, it should be challenged and examined by as many people as possible, people
who have technical expertise to determine the fact or error in the claim.
My goal here is to offer up as much supporting material
as I now have, and continue to add to it in the coming months, so that others have as much data to test and evaluate as possible.
This section presents both the general concept foundation ideas anyone evaluating this issue should be aware of, first, and then explains
specifics that I have done or considered in my analysis.
The K-100 Camera - This camera is a spring-driven standard 16mm camera which takes a 100' daylight load of film stock. In its standard
configuration, it holds one lens for the camera proper, and a companion lens for the side viewfinder. It is not equipped for viewing
through the camera lens. In general principle, this restricts the camera to using the lenses supplied by Kodak, with their companion
viewfinder lenses, and precludes the prospect of using a zoom lens. This is not a mechanical restriction, so much as question of reason
and intent to use the camera successfully.
Mechanically, the camera accepts lenses with a "C" mount, and also an "S" mount with
an adapter (Kodak's 15mm lens for this camera is an "S" mount, with a Kodak-made adapter for "S" to "C" mount attached, as one example).
So in theory, any "C" mount lens can be put on the camera, and should render an acceptable picture. But if you do not have the companion
viewfinder lens, then what you see in the viewfinder is not the composition you get on the film. So we must ask, what logical purpose
would a camera user anticipate to deliberately configure the camera to insure what he/she sees in the viewfinder is guaranteed NOT
TO BE what will result on film?
And when you consider the camera was a rental, what would motivate the rental person to even
allow the renter to take out the camera under these circumstances, when the resulting footage, not to the camera user's liking, could
put the blame on the rental company for failing to equip the camera properly with the correctly matching camera and viewfinder lens
pair?
So on this issue, one can argue for the mechanical possibility of using any "C" mount lens on the camera and successfully
filming, but in the practical sense, the sheer irrationality of using a camera and viewfinder lens pair that are deliberately mismatched
in focal length, suggests to me the prospect is unlikely and testing for this is simply not a high priority in my research effort.
Far more likely and probable things have priority for my research time. So with this in mind, I have given priority to the Kodak 25mm
lens and the Kodak 15mm lens that do have companion viewfinder lens available for that camera. Other researchers may, of course, test
for lenses of other focal lengths, if they feel that the effort is a worthwhile pursuit.
There will continue to be arguments
that the camera used by Roger for the PG Film had the 25mm lens on it, because rental documents and some type of police report (when
he apparently failed to return it in a timely manner) specify a 25mm lens on those documents. It is reasonable to say, when he rented
it, months before, it did have a 25mm lens on it. But changing a lens requires about one minute of time, and so the prospect that
he changed lenses on the camera somewhere between the time of the rental and the filming at Bluff Creek is a perfectly reasonable
expectation. And from an argumentative standpoint, the 15mm lens is a better lens for scenic outdoor photography, and that is substantially
what Roger had been filming in the months he had the camera.
Ultimately, however, the lens determination must be made by optical
science and mathematics, not arguments of speculated human behavior. So while all the above notes are relevant to the discussion,
they are not being used in any proof of the lens focal length.
Verifying the Lens
There are several approaches to verifying the lens used on Roger's camera. The methods I am intending to use
in this analysis are Replication, Mathematical Model, and Photogrammetry Software Processing.
Replication - This is the process
of duplicating the process of filming, with a model or mockup of the original scene of Bluff Creek, using specified camera positions
and lens specifications to correspond to specific frames of the real film. The greater the number of camera positions used, and the
greater the number of objects which match in position and size, the greater the likelihood that the replication is a truthful determination.
An example, in ballistics, would be to look at a bullet slug recovered from a crime scene, and studying the bullet deformation from
impact, and then replicate the event by firing the same type bullet into various physical substances to study the deformations on
impact of the study bullets, as compared to the deformations of the crime scene bullet. A replication of deformation argues for the
study criteria as being consistent with the crime scene criteria.
In this case, a digital model has been constructed of the Bluff
Creek site, and matched to seven film frames representing seven different camera positions. And 15 objects, most representing multiple
points (which allows for comparing both their position and size relative to the objects in the film images), have been used. A score
sheet charting the points or objects which match in varied film frames has been developed, and a full data sheet of object coordinates
(dimension, position and rotation) has been provided for other researchers to test the model independently. Camera coordinates (position
and rotation) have also been provided.
This replication method first allows me to compare my site model to the film, and determine
that an excellent match of objects and camera positions occurs for the lens specification I am testing (the 15mm lens), and also allows
for other researchers to replicate my efforts independently and verify the results. It further provides both myself and other researchers
with the materials to test other lens specifications with similar model replication, by adjusting the lens angle of view, and them
setting different camera and object positions to attempt an alternate replication where we can grade the number of matches. If they
do so, their results can be easily documented in a consistent manner for other researchers and myself to verify.
Mathematical Model - A mathematical model should allow for calculations of object positions and camera positions based entirely on
the object position and dimension data that can be extracted from the film. This mathematical model should allow for exact computations
of relative position of objects and cameras, independent of a viewing angle, so it can be done without a pre-determined lens angle
of view put into the process (as the replication method does require). So this method, once accomplished, would be an independent
verification of the lens, determined after the mathematical model is complete. Work is currently underway to define the various tree
objects in the highest quality image form, as they appear in multiple frames, so the data has a high degree of accuracy and a low
margin of error.
Once the data is prepared for this method, it will be released for independent researchers to evaluate and
test.
Photogrammetry Software Analysis - Software which performs photogrammetry analysis uses an automated version of the mathematical
model described above, with specific algorithms developed first to analyze film footage and identify points, objects or patterns
(like pattern recognition or edge detection methods), and then additional algorithms to process the data and determine point positions,
camera positions and lens focal length. I have been using one product, boujou 4.1 made by 2D3 Ltd of the UK, for this method. The
boujou software is a CGI industry "Matchmove" software which specifically tracks points, objects, cameras and lens data so a digital
model or effect can be composited into real film footage effectively. I chose it specifically for its capability of locating cameras
and determining lens focal lengths.
A preliminary analysis I have done confirmed the 15mm lens on the PG Film camera, and a separate
test using footage I generated with a known lens horizontal field of view was tested by the software and determined to be accurate
in its lens determination. I am currently running additional tests with more footage, as well as setting up the PG Film frames for
a more lengthy and detailed analysis, which I will document thoroughly. The staff of the software developer have also graciously offered
to run their own analysis using the footage, and preparations for that effort are being made.
Links To Lens Data
The following pages will go to the various discussions and image documents which show the Replication with
the Digital Site Model, and provide more information about the process. Additionally, some links will go to the material other researchers
will need to test the Digital Model.
Verification and Falsification - A correct proof needs both verification and falsification to be complete. Verification is the positive
or proactive demonstrating or determination that the fact claimed is correct. Falsification is the demonstration that alternatives
cannot be correct. Both are considered strongest in factual certainty when results are replicated by persons other than the claimant
of the proof. So I am making the data available for other researchers with that intent, that they may test the models and data, and
either concur or contest the results I have obtained.
My primary verification of the 15mm lens claim at this point is the replication,
and the high number of matches in multiple objects and multiple camera positions. The compatability of the camera position determinations
as compared to site map measurement data and other filming at the site is supportive of this claim. What is currently lacking is the
margin of error calculation, still under study.
My primary falsification of the 25mm lens (thus far) is that after two months
of attempting to build a digital model of Bluff Creek using the 25mm specifications, the model failed to solve by every effort or
method. But at the time I was doing so, I wasn't thinking in terms of trying to falsify the 25mm lens. I was actually trying to prove
it valid. I did not maintain the appropriate documentation of the experiments, and so now, these must be redone. The work will be
forthcoming, but isn't done yet.