Comparison of
35M67 wide-conversion lenses
UN PWC-01, Sea&Sea WL0.56×/L, Epoque DCL-20 (Ikelite
W20), Inon UWL-100.
By David Knight
Given below are the results of tests carried out on a group
of wide-angle conversion lenses of the type that can be screwed to the
outside of an underwater camera housing. These lenses were all designed
to work without vignetting with cameras having a 35 mm-equivalent focal
length of ≥ 35 mm, i.e., they pre-date super-zoom
compact cameras (but that is not necessarily a bad thing). Note that
none of the lenses is a badged version of any other (i.e., they are of
diverse manufacture).
The purpose of the test, in each case,
is not to measure the magnification (which is in any case available
from the product literature) but to obtain a quantitave measure of
geometric distortion and chromatic aberration. The reason for choosing
this approach is that we want to know how well the lens corrects for
the water-air boundary (i.e., the flat port on the camera housing) and
for its own internal defects. The method used is to take test-card
pictures and apply radial corrections using the Panorama Tools 3
rd
party Photoshop filter plugin, the requisite information being obtained
by comparing the correction coefficients. The degree of improvement
obtained after correction might also encourge some readers to give
radial correction a try.
Note that the correction coefficients
obtained are not necessarily definitive. More time spent adjusting them
might have produced slightly better results. Note also that
there are easier ways to correct images than using the Pano Tools
software interface. Some programs, such as Adobe Photoshop Lightroom
offer simple slider controls that, although having a somewhat
restricted adjustment range, are adequate for most purposes (see
the
computer
lens
correction article).
Tests were carried out using two 3× zoom
compact cameras: an Olympus C-5050 (now discontinued), which
was pretty close in optical design to the lens maker's target camera;
and an Olympus μ410 miniature point-&-shoot (also
discontinued). The results obtained without the use of a supplementary
lens are given first for comparison purposes. Note that, in the case of
the C-5050, we also determine the characteristics of the camera in air,
since this information allows the effect of the flat port and the
performances of the supplementary lenses to be separated.
Olympus
C-5050 camera and PT-015 housing
4.9Mp 1/1.8" CCD,
2560 × 1920 pixel RAW output.
3× optical zoom, 7.1 - 21.3mm (35 mm equiv.: 35 - 105 mm)
Test conditions:
Zoom setting full wide. 35mm equiv. f = 35 mm
AF mode: Normal. Aperture = f/8 (aperture priority mode)
Underwater light source: Sea&Sea YS50TTL/N with Matthias
Heinrichs' DA-IR-N TTL converter (internal flash switched
off). |
|
Corner
details of photographs taken
with the C-5050 are cropped from the 2560 × 1920
full-size image.
C-5050 camera in air (no housing, natural
daylight)
Note: The purple tinge of the ruler is due to its high
reflectivity in the ultraviolet part of the spectrum (the picture was
taken outdoors with no UV filter). A UV filter is not necessary
underwater.
Radial Correction
PanoTools
radial correction
coefficients |
|
Camera and Housing (flat port, no
supplementary lenses or filters)
Note: The illumination hot-spot in the full picture is due to
the use of a torch as a focusing-assistance light.
Radial Correction
PanoTools
radial correction
coefficients |
|
Notice that the C-5050 zoom lens set at maximum-wide (equiv. f = 35 mm)
gives a small amount of barrel distortion in air, but this is almost
completely corrected underwater by the pincushion distortion introduced
by the flat port. This results in a very small third-order radial
correction coefficient ( b=0.005) to remove the residual pincushion
distortion. This happy coincidence, incidentally, is probably
accidental. Compact camera zoom lenses are not designed with underwater
ports in mind, and other cameras will not usually turn out to be
rectilinear underwater. The flat port moreover, does introduce some
chromatic aberration (as it must), and this combines with the residual
aberration of the zoom lens to require shifts in the first-order (d)
coefficients for red (-0.0005) and blue (+0.003) relative to the green
(reference) channel.
|
Olympus μ410
in ikelite 6138.4 housing
3.9Mp 1/2.5" CCD, 2272 × 1704 pixel JPEG output
3× optical zoom 5.8 - 17.4 mm
(35 mm equiv: 35 - 105 mm)
Test conditions:
Zoom setting full wide. 35 mm equiv. f = 35 mm.
Focusing mode: AF, normal. Aperture = f/3.1
|
Underwater lighting: ikelite DS-125 with 4100.6 slave sensor / EV
controller.
Corner
details of pictures taken
with the μ410 are cropped from the full-size
2272 × 1704 pixel image.
Flat port, underwater (corner detail has been WB corrected).
Radial correction
PanoTools
radial correction
coefficients |
|
The 6138.4 is an example of an ikelite miniature compact housing from
2004 (pre-dating the Ultra Compact). It had a front port body
of 2.2" outside diameter, but the 2004 version had no provision for the
attachment of wide-angle (or macro) supplementary lenses. Larger
ikelite compact housings of the time had a 3" diameter port with an M67
filter thread, but it was perhaps thought that people would be less
interested in using supplementary lenses with the then-new
high-performance miniature cameras. Olympus however, had no qualms
about providing a filter thread on its PT housings
for μ-series cameras, and the wide-angle results were
very good.
The μ410 was included
in these tests as a way of showing just how good the results could be
with a fully-automatic digital point-&-shoot. In other words,
it was an oblique way of telling people that it was time to ditch their
underwater film cameras and go digital. To that end, we might simply
have screwed an M46 wide converter to the port of the Olympus PT-016
housing, but we also wanted to persuade ikelite to make provision on
its miniature housings. A particular design issue at the time was that
an M67 thread offered the greatest choice of add-on lenses; but some
people also wanted to use the built-in flash, which would be
permanently blocked-out if a 3" port was used. The solution would be a
removable M67 port-adapter.
For the purpose of the test using the
ikelite housing, it was, of course, necessary to improvise a port
adapter. This was done by bolting the M67 lensholder ring from an
Epoque lens-mount adapter to an Ultralight AD-SS YS-mount adapter. This
gave a lensholder that could be attached to a short ball-joint arm,
which was mounted on the accessory shoe on the top of the housing. To
permit slave triggering of an external flash without light from the
internal flash leaking into the back of the supplementary lens, a black
rubber light baffle (cut from an old car-tyre inner tube) was fitted
around the housing port. The hole in the baffle was cut 2 mm smaller
than the port diameter, to make a stretch-fit so that the baffle would
adopt the shape of a funnel. Wide-lenses were tested with the
back-surface pressed against the port, although, since all of the
lenses were easily capable of covering the camera's FOV, a gap of a few
mm would have been tolerable.
Ike Brigham (late lamented founder of ikelite
Inc.) was suitably impressed by the results we obtained and
subsequently arranged for the design and manufacture of a range
of 2.2" port
adapters (allowing either Inon AD
or M67
lenses to be fitted). Post 2004 versions of the ikelite 2.2" port have
a groove for retention of the adapter, and also, where possible, an M46
filter thread.
The lensholder improvised for these
tests is incidentally nowadays available in the form of the Inon M67
lens arm (see right). The device is also particularly useful for making
LCD magnifiers. |
Inon lens arm |
|
UN
PWC-01
Depth rating: 4 0m
Magnification: ×0.58
Dimensions /mm: 102 Ø, 59
long. Weight: 650 g.
M67
filter thread |
Radial Correction
PanoTools
radial correction
coefficients |
|
|
Sea&Sea
Wide Lens 0.56×/L
#58070 (Discontinued in 2004).
Depth
rating: 60 m
Magnification: 0.56×
Max. FOV without
vignette: 100°.
Weight: 460 g in air, 280 g UW.
Dimensions
/mm: 93 Ø, 53 long. |
Radial Correction
PanoTools
radial correction
coefficients |
|
|
Epoque
DCL-20 (and ikelite W20 #6420)
Magnification: 0.56×
Max. depth: 60 m
Weight: 650 g in air / 400 g
UW.
Dimensions /mm: 99.4
Ø, 59.5 L.
Configuration: 5 elements, 4 groups,
AR coated.
Materials: Optical glass lens, Al alloy body. |
Radial Correction
PanoTools
radial correction
coefficients |
|
|
Inon
UWL-100 M67
Depth rating 60m.
Magnification: ×0.57077
Max. FOV without
vignetting: 100° UW,
158° in air
(when camera lens 35 mm equiv: focal length is
31.5 mm).
Configuration: 5 elements, 4 groups,
optical glass.
Afocal. Achromatic. AR coatings front & back.
Dimensions /mm:
100 Ø, 58.5 L. Weight: 625g in air, 370g UW. |
Radial Correction
PanoTools
radial correction
coefficients |
|
|
Inon
UWL-100 with dome port (128°)
Depth rating 60m.
Dome port forms
sealed air cavity in front of
UWL-100 Lens.
Max. coverage
without vignetting: 131°
underwater (when camera lens 35 mm equiv: focal length is 31.5 mm).
Materials: Optical glass dome with
internal multiple anti-reflection coating. Anodised Al body.
Dimensions /mm: 129.6 Ø.
L = 48.9 dome unit only, 85.5 with UWL-100.
Weight:
498 g dome unit only (air). 587 g UW with UWL-100. |
Olympus C-5050 camera. Zoom: 35mm. Aperture: f/8.
Note that, due to the 128° FOV of this lens, the camera is
extremely close to the test card. The apparent loss of edge resolution
is due to curvature of field, and will be largely compensated by depth
of field at greater distances.
Radial Correction
PanoTools
radial correction
coefficients |
|
Collected radial correction coefficients - 35M67 lenses
with C-5050 camera
Note that the coefficients are always adjusted so that a + b + c + d =
1 in the green (reference) channel. This ensures that the
image height remains unchanged by the correction.
Lenses are ranked in order of increasing coverage
Lens
system |
magnif
-ication |
FOV |
Geometry |
Aberration |
a |
b |
c |
a+b+c |
Δdred |
Δdblue |
Camera + flat port |
|
46.2° |
0 |
0.005 |
0 |
0.005 |
-0.0005 |
0.003 |
Camera |
1 |
63.4° |
0 |
-0.02 |
0 |
-0.02 |
0.001 |
0.0005 |
Port
(f = 35mm) |
|
|
0 |
0.025 |
0 |
0.025 |
-0.0015 |
0.0025 |
PWC-01 |
0.58 |
|
0.01 |
-0.06 |
-0.005 |
-0.055 |
0.0011 |
0 |
UWL-100 |
0.57 |
91.4° |
0 |
-0.06 |
-0.035 |
-0.095 |
0 |
0.002 |
DCL-20 |
0.56 |
|
0 |
-0.055 |
-0.04 |
-0.095 |
0 |
0.002 |
S&S
WL0.56 |
0.56 |
|
0.015 |
-0.05 |
0.005 |
-0.030 |
0 |
0.0015 |
UWL-100
+ dome |
|
127.9° |
0.02 |
-0.08 |
-0.16 |
-0.220 |
0.001 |
0 |
A measure of the image geometry is given by the figure a+b+c, where a
positive value indicates pincushion distortion, and a negative value
indicates barrel (fisheye) distortion. For the lenses, in
each case, note that the values given correspond to lens + camera, so
it is necessary to subtract the figure for the camera to get a figure
for the lens. From this, we can see that the Sea&Sea
lens (now sadly discontinued) is very nearly rectilinear, giving a
figure of -0.03, of which -0.02 is attributable to the camera lens.
The Inon UWL-100 with dome, on the other hand, is a serious
fisheye, as is to be expected for a lens with such an extreme FOV.
Chromatic aberration is indicated by
shifts in the first-order (d) correction coefficients for red and blue
relative to green. These shifts correspond to the adjustments in
overall magnification of the red and blue images, as needed to
superimpose them on the green image. Thus:
Δd
red
= d
red - d
green
Δd
blue
= d
blue
- d
green
An important function of any wide converter is to correct for the
efects of the water-air boundary. If the converter is successful in in
this respect; and assuming that it is of an achromatic design (so that
it corrects for dispersion in the glasses used to make it) then the
residual aberration should look like that of the camera. Note however,
that the coefficients obtained by working with a 5M pixel image have an
uncertainty of about ±0.0005. If the result comes
out better than the camera, then it is a happy coincidence, but it
won't be reproduced when using a different camera.
If we leave the Inon lens with the dome
aside; there is little difference in performance between the four
remaining lenses. The Sea&Sea lens showed the best rectilinear
correction, as mentioned above. The UN PWC-01 has the least chromatic
aberration, but all provide considerable improvement over the
uncorrected lens port, and the other lenses offer a greater angle of
coverage (lower magnification). The Epoque DCL-20 and Inon UWL-100 are
practically identical in behaviour, except that the Epoque lens gives
slightly greater coverage . There is no great reason to favour any one
of these 35M67 lens over another; except on the basis of price or
availability (In 2004, the Epoque lens was cheapest, the
Sea&Sea lens was
discontinued, and the UN lens was hard to get in the UK.); or (in the
case of the
Inon lens) for the ability to accept an add-on dome in order to achieve
super-wide coverage. Overall, all of the lenses provide substantial
improvement over the bare port and do their jobs well.
The results obtained using the μ410 camera do not give
a great deal of additional information about the lenses except to
indicate that there is a small penalty in corner image sharpness for
using an aperture of f/3.1. Worth noting however, is the
remarkable image quality that can be obtained from a fully automatic
digital camera.
A final point that should be kept in mind however, is that both of the
test cameras had only a 3× optical zoom range. In
recent years there has been a tendancy towards greater zoom range, and
from an underwater photography perspective, this is not optimal. The
problem is that super-zoom lenses change physical length radically as
the focal length is varied, and are usually shortest at wide-angle
settings. This forces the housing designer to place the port a long way
from the camera-lens at wide-angle settings in order to prevent the
lens from crashing into the port at telephoto settings. When using a
wide converter, the effect of having the entrance pupil a long way from
the port is to degrade optical performance even when no vignetting
occurs. The main effect is an increase in chromatic aberration, and
this will be visible in large reproductions. The solution, for compact
camera users who want to produce exhibition prints or displays, is to
apply computer radial correction. Users of interchangeable-lens cameras
who wish to use an external wide converter are advised to adopt lenses
of fixed focal length, or zoom lenses of limited range, and choose the
shortest allowable port body.
DWK
© David W Knight, 2004, 2012, 2018
Photographs by David W Knight.