Wide-angle zoom lenses – results and discussion.
I tested the wide-angle zoom lenses (DX 12-24mm and FX 16-35mm) only in the swimming pool, using the large dome port (Seacam Superdome) and corresponding extension rings. I have then compared the sharpness of land photos with the photos made under water. The sharpness was measured in the centre of the image and in a point close to the far corner for both extreme focal lengths.
The 10-24mm f/3.5-4.5DX lens at 10mm exhibited a good underwater sharpness in the centre compared to the sharpness of the land image. It was quite sharp even when the aperture was open, and the sharpness increased up to f/11 when it started to deteriorate due to diffraction. The sharpness in the far corner was noticeably lower than the land image sharpness. The sharpness with open aperture was quite low and has increased with aperture stopping down in an approximately linear fashion up to f/11, where it started to gradually fall. The optimal aperture for this lens combined with the Superdome at 10mm is f/11.
The situation for 24mm is very similar, only the central sharpness is somewhat higher even with the aperture open. The sharpness increases very slowly up to f/11 and then starts to decline due to diffraction. A very similar situation to 10mm in the far corner is obtained with 24mm, but with slightly larger values (better sharpness). f/11 is therefore the optimal aperture for 24mm, too.
The 16-35mm f/4 VR lens at 16 mm exhibited a significant reduction of sharpness in the centre compared to the land sharpness. The centre sharpness has increased very slowly, reached the peak at f/11 and then started to slowly decline. The sharpness in the corner was consistently increasing during aperture closing down and was significantly lower than the 10-24 sharpness at 10mm. The optimal aperture for 16-35 at 16 mm is somewhere between f/11 and f/22.
At 35mm focal length, the situation in the centre is very similar to 16mm, the sharpness peaking at f/16. Even at 35 mm, the sharpness in the corner was consistently increasing with the aperture stopping down, the actual values exceeding those measured for 16mm. The optimum f-stop value for the 16-35 lens at 35mm is f/16.
Comparing the performance of both lenses under water, we can see the 10-24mm f/3.5-4.5DX outperforms the 16-35mm f/4 VR for both extreme focal lengths, both in the centre and in the corners, as well as for all f-stop values up to including f/11 (at the short end, both lenses with open apertures produce equally bad images in the corners). Both lenses perform approximately equally at f/22 in the short end, and at f/16 at the long end. At f/22, the 16-35 lens manages to slightly surpass the 10-24 (due to stronger diffraction on a smaller sensor).
The results lead to the following conclusions: The 10-24mm f/3.5-4.5DX lens combined with a dome port is optically better than 16-35mm f/4 VR in nearly all parameters, consistent with the theory of dome ports.
Both lenses exhibit a persistent increase of sharpness as the aperture is stopped down, as well as a decrease in the sharpness for the largest f-stop values as a result of diffraction. The optimum aperture values are f/11 for 10-24 and f/16 for 16-35.
Fisheye lenses – results and discussion.
I have devoted a lot more attention to fisheye lenses than to zoom lenses because I use them a lot more often under water. As in the macro photography tests, I began by testing the difference in CA for fisheye lenses. The work was done in a swimming pool (using a large fisheye dome) and I must emphasize at this point that taking photos from behind a dome port (with the lens positioned correctly) introduces no additional CA, as it is the case with taking photos through a flat port. All CA originates from the lenses themselves and is completely equal on land and under water. I was surprised to discover that the newer 10.5mm f/2.8DX fisheye lens is burdened with substantially more CA than the much older 16mm f/2.8! The old lens was designed deeply in the era of film, when the images recorded on film could not be corrected. But the new lens is a child of the digital era and its designers obviously just didn’t bother to fiddle with CA, considering it can be removed quite easily with a suitable software application…
Making an objective evaluation of results obtained with fisheye lenses, having an almost 180° diagonal angle of view and heavy barrel distortion, was much more difficult. In addition, the ratio between the horizontal and vertical angles of view for fisheye lenses is not 1.5 as with all linear lenses, instead it is larger. This is why I used an additional test element with lines, placed to the right of the test chart, parallel to its vertical edge and at a distance that put the test element at the edge of images made with a fisheye lens. The sharpness was evaluated in the centre of the test chart and on the additional element, situated at the far right edge of the image.
Testing both lenses on land I have also noticed that the sharpness of the older 16mm f/2.8 lens does not lag significantly behind its “younger brother,” contrary to the reports on its optical obsolescence often found on the web. The major disadvantage of the old lens for an underwater photographer is its inability to focus closer than 25 cm to the sensor plane. The new 10.5mm lens is able to focus to 14 cm from the sensor plane, meaning as little as 4 cm from the front lens element. This makes it suitable for extreme close-up ultra-wide-angle images, which are lately becoming increasingly popular in underwater photography.
I tested both lenses in a swimming pool using the large and small fisheye ports. Using the small dome port on the old 16mm lens may be completely nonsensical, but I have nevertheless conducted the test if Nikon one day presents a new version of the lens, which is able to focus closer.
Consistent with the expectations and the dome port theory, the sharpness in the centre of the image obtained with the 10.5mm lens and the large dome is only slightly worse as the sharpness obtained on land. The small dome port produces even worse results. The sharpness in the centre of the image peaks at f/8 and f/11 for both dome ports. At f/16 and f/22, where the sharpness declines due to diffraction, the differences between underwater and land results are minimal.
A completely different situation emerges at the far edge of the photo. The sharpness obtained with both dome ports is significantly worse than land sharpness, although the latter is also far from impressive. The sharpness peaks at f/8 on land, at f/11 for the large dome port and at f/16 for the small dome. The sharpness is equal at f/22.
A general conclusion could be that the optimal f-stop value for a 10.5mm lens with a large dome port is f/11. The photos taken with the small dome port exhibit optimal sharpness at f/11 in the centre of the image and at f/16 at the edge. The sharpness of the image made with a small dome port at f/16 is more uniform, so I am favoring this aperture value.
The situation in the centre of the image made with the 16mm lens and the large dome port is similar to the 10.5mm image, but the sharpness is considerably lower. The sharpness in the centre of the image made with the large dome port slowly increases in relation to aperture closing down and peaks at f/16. The situation with the 16mm lens and the small dome port is completely different. The projection of objects at infinity made by the small dome port is so close that the lens is no longer able to focus (as discussed above in the section on dome port theory). I was only able to take the photos by switching to manual focusing. The images were never focused, but stopping the aperture down gradually made them sharper owing to the depth of field effect. The best sharpness was not obtained until f/22.
The situation at the edge of images made with the large dome port and the 16mm lens is similar to the results obtained at 10.5mm, but the images were once again noticeably fuzzier. The peak of sharpness occurs at f/16 and the sharpness is only slightly reduced at f/22. If a picture made with the small dome port is fuzzy in the centre, we can expect nothing but worse results at its edges. Even f/16 yields no satisfactory sharpness and a large jump in sharpness is observed only at f/22. We may draw a general conclusion that the optimal aperture value for the 16mm lens and the large dome is f/16 – in the centre and at the edges. The small dome port is useless for the 16mm lens and a satisfactory sharpness is only possible at the extreme aperture f/22.
The only reasonable comparison of both lenses can be made by using the large dome port. In the centre of the image, the 10.5mm lens outperforms the 16mm convincingly for all f-stops from f/2.8 to f/11, where the sharpness reaches its peak. The sharpness of 10.5mm lens starts to decrease at f/16 due to diffraction, while the 16mm lens reaches its peak at f/16, meaning the sharpness of both lenses is equal at that setting. At f/22, the sharpness in the centre is equal for both lenses and somewhat worse than at f/16.
The situation at the edge corresponds to the situation in the centre, with a large gap in sharpness observed at f/11. Here is where the 10.5mm lens reaches its peak and the sharpness of the 16mm lens is lower by approximately a half. The 16mm lens gains heavily on sharpness at f/16, almost equalling the performance of 10.5mm. The 16mm lens even manages to surpass the 10.5mm lens at f/22, considering the latter is more prone to diffraction owing to the D2x’s smaller pixels.
As discussed before, the optimum aperture value is f/11 for the 10.5mm lens and f/16 for the 16mm lens. Comparing the physical diameters of opening (focal length divided by the f-stop number), we can see that the diameter of the opening at optimal aperture is about 1 mm for both lenses. The 1 mm “pinhole” is obviously responsible for delivering the depth of field required to offset the dome port’s optical aberrations (field curvature), so that the sharpness is acceptable from the centre of the image to the edges.
The evaluation of far corners is even more problematic for fisheye lenses, so I didn’t find it sensible to measure the sharpness there. Instead of the element with test lines I used an element with a drawn scale (60 cm with 5 mm graduation) to assess the boundary between sharpness and fuzziness caused by the dome port. This scale could unfortunately not be used to evaluate the sharpness in real-life conditions, but it has served well to demonstrate sharpness and fuzziness.
The motivation behind the tests in the sea was to show the practical meaning of numerical values of sharpness, obtained by taking the photos of test charts in a swimming pool. Both fisheye lenses were tested in very real-life conditions – the tests were conducted in February in Piran, Slovenia, the water temperature was 9 °C and the maximum visibility was 5 m. I have chosen two different motifs of the seabed and took photos using both systems and all f-stop values. The first system consisted of Nikon D2x + 10.5 mm f/2.8DX fisheye lens + Seacam housing with a large fisheye port dome + two Seacam 150 flash units. The second system included Nikon D3 + 16 mm f/2.8 fisheye lens + Seacam housing with an identical fisheye dome port + two Seacam 250 flash units. I took two or three photos for each f-stop and picked the sharpest images. Problems with too strong ambient light were observed when the aperture opening was the largest at f/2.8 to f/5.6 (making the images more greenish), so I have lowered the ISO value to 100 on D3. For the two most stopped settings of aperture f/16 and f/22, I have raised the ISO value on the D2x to 200. The shutter speed was always 1/250s in order to eliminate the influence of fuzziness due to vibrations.
The sample images (100% crops) were taken at the longer edge of the image (closer to the centre, hence with maximum sharpness), at the shorter edge (far from the centre, hence with lower sharpness) and two times in the far corner (the worst sharpness). The edge crops show the 10.5mm lens having approximately one f-stop of advantage in sharpness over the 16mm lens. In the corner crops, the 10.5mm lens is leading the 16mm lens by almost two f-stops. This result is consistent with the curvature of field theory: if we press a flat rectangle (sensor) against a spherical surface (where the image is created) so that the centre of rectangle touches the sphere, the edges of the rectangle are closer to the spherical surface than its far corners. The smaller the sphere, the greater the difference of distances. This means that the smaller the dome port, the greater is the curvature of the field (smaller sphere) and the more difficult it is to obtain a sharp image in the far corners.
Finally, I have made a comparison between the JPG image obtained directly from the camera and the image converted from the NEF format in Capture NX. All crops originate from the corner of images taken with both lenses at f/16. The JPG image from Nikon D2x displays a quite strong CA, while the image processed in Capture NX exhibited no CA and was comparatively very sharp. Both images taken with Nikon D3 were practically identical, without CA, but noticeably fuzzier.
It must be noted that the sharpness from corner to corner is only required when the image includes a lot of details, e.g. when taking a photo of a coral reef and the corals can also be seen in the far corners of the image. But if the main subject is concentrated in the middle section of the image and there are no important details in the corners (e.g. a big fish on a blue background), we can afford less sharpness in the corners, because it will go unnoticed.
I also recommend the underwater photographers to preferably take their photos using the large dome port, which will give them optically better results than the smaller one. The small dome port should only serve as backup for special effects, i.e. for extreme close-up wide-angle photography.
Which sensor format is therefore better for underwater photography? I believe most photographers will agree with me that it is the DX (small) format. The small sensor is undoubtedly more suitable for macro photography, because it makes it easier to take very small crops in nature. A larger depth of field of small sensors unfortunately proved to be no advantage, considering the stronger diffraction as a result of smaller pixels on small sensors. The diffraction cancels out the positive effect of depth of field for (excessively) stopped down apertures.
Considering the use of dome ports for wide-angle underwater photography, there were initially some doubts as to which sensor is more suitable. The theory and the results of practical tests have once again confirmed the advantage of the small sensor, where the dome port aberrations (changed focus and curvature of field) are less evident. We must also take into account a significant fact that only the small format sensor enables technically successful extreme close-up wide-angle images to be taken with a small dome port and fisheye lenses, which are capable of focusing very closely (AF Nikkor 10.5mm f/2.8DX and Tokina 10-17mm f/3.5-4.5, a lens that is also very popular among underwater photographers).
Until recently, the superior characteristics of large FX sensors for photography in bad lighting conditions (low noise at high ISO values) put them at a large advantage over the small sensors. The large sensors were popular mainly with underwater photographers who worked predominantly in bad lighting conditions, e.g. taking photos of shipwrecks in great depths. But the recent launch of Nikon D7000 changed everything… This “big” small Nikon has proved that it is possible to take satisfactory photos with small sensor even when it is getting dark.
Considering all the listed facts, it should not be too hard to guess the successor to my old D2x, which has already made more than 50 000 underwater images. Without a doubt, it is going to be a “high-end” DX camera – the successor to D300s!
For help in conducting the tests I would like to thank:
Uroš Ilič (Syncomp) for making the test chart.
Gianni Pecchiar and Adriano Morettin (both underwater photographers from Trieste, Italy) for lending me additional Seacam equipment.
Igor Vrhovec and Marjan Makuc (Norik Sub group) for logistic help in conducting the tests in the swimming pool (Igor) and in the sea (Marjan).
Borut Furlan was born in 1959 in Postojna, Slovenia, and is now living in Ljubljana. He has a Master of Science (M.Sc.) degree in chemistry.
Borut started diving in 1977. He is a CMAS diving instructor level II, ANDI technical Nitrox diver and underwater photography instructor. He is diving all around the world from Norway on north to South Africa on south, from Columbia, Panama and Mexico on west to Indonesia, The Philippines, Malaysia and Solomon islands on east. However his main focus is his “home sea” Mediterranean and the Egyptian part of Red sea.
He entered the world of underwater photography in 1979, building his own underwater housing for his first camera. Since 1998 he uses Seacam underwater camera housings. Since 1980 he uses Nikon photographic equipment and from 2006 on he works in digital technique.
He is participating in underwater photography competitions since the beginning of his career. In ten years, from 1994 to 2003, when he quit competing in national contests, he won the title of the national champion in general underwater photography every year. The most important of his international awards is certainly the 3rd place on the World Championship in Underwater Photography in Egypt in 2000 and “Best of show” award in Chicago, USA in 2008.
Borut Furlan had many photo exhibitions, lectures and multivision slide shows in Slovenia and in neighboring countries (Italy, Croatia).
His work was published in many domestic and foreign publications, among other in magazines like National Geographic Slovenia and other brochures. He published some calendars and cooperated in several domestic and foreign book editions. In 2006 he presented the first Slovenian underwater photo monograph, a 296 pages coffee table book “A Journey into the Silent Worlds,” published by Didakta. For more information, please visit his website.