The purpose of a lens hood is to prevent flare, which can seriously degrade the image quality of photographic lenses. A longer hood offers better flare protection than a short hood, but when the hood exceeds a critical length vignetting sets in. Therefore, a lens hood needs to be carefully chosen. The optimum hood depends on the philosophy behind the hood, with the shape and size as the key parameters, and on the lens aperture and subject matter. Several lens hood considerations will pass in review in this article. I will not discuss issues such as choice of material - I leave it to the imagination of the reader that an effectively blackened hood is more useful than one whose interior shines as a mirror. Rather, I concentrate on size and shapes. Although I am the first to admit that the contents are partly of an academic nature, the material could be vital for those readers who strive after the best possible image quality.
Size of the hood
A wide, long hood is a better choice than a narrow, short hood. Figure 1 exemplifies this statement for the Sonnar 135/2.8. The imageforming light collected by the lens is indicated with the gray beam that strikes the front element at the lens field angle. Vignetting sets in when a lens hood penetrates into this beam. Both indicated lens hoods, the short, built-in hood and the larger Contax metal hood #5, just clear the front element, i.e. they do not obstruct the gray beam and hence do not cause vignetting. However, the small hood allows the slanting black ray to hit the front element whereas the large hood blocks this ray, preventing it from contributing to lens flare (indicated by the reflections in red).
Figure 1. Two lens hoods on the Sonnar 135/2.8. The larger hood is the better choice as it blocks more nonimageforming light.
Influence of hood length on vignetting
Vignetting by a lens hood is called mechanical vignetting. It is usually associated with black image corners and an abrupt transition from bright to black. However, a lens hood can also give rise to a gradual corner darkening in a similar fashion as natural and optical vignetting. The influence of a lens hood on vignetting will be illustrated with a series of sketches with an increasing hood length.
Figure 2 shows the Planar 50/1.4 as is, without lens hood. The green part denotes the rim surrounding the front element. The red bars mark the lens entrance pupil, which is the image of the aperture stop (the black bars) seen by an observer looking into the lens from the front. Finally, the purple bars correspond to the image of the rim around the rear element. Each of these elements is a circle and the clear aperture is given by their common area. The circle plot at the right of the lens is the situation relevant for the image corner. A corner object point at infinity sees the circles in these mutual positions and is confronted with a clear aperture marked by the orange area, viz., the common area of all circles. There are also more complicated lens designs which require the inclusion of internal rims to determine the clear aperture, but the principle of the common area remains the same. For the image center all circles are concentric and the smallest one defines the entrance pupil, which is round.
Figure 2. The Planar 50/1.4 without lens hood at full aperture.
At full aperture the lens accepts a broad light beam. For the image center this beam is colored yellow. For the image corner the accepted beam is narrower. The darker orange beam is the beam that would be accepted by the entrance pupil if the lens barrel weren't present and if the lens elements were larger. The brighter orange beam is the part of the darker orange beam that is not obstructed by the lens barrel. It is delimited by the common area of the green and red circles in the equivalent circle plot. Further clipping of the oblique beam occurs by the rear rim and what remains is a clear aperture that is substantially smaller than the aperture for the image center (optical vignetting).
Figure 3. The Planar 50/1.4 without hood at f/11.
When the lens is closed down to f/11, the entrance pupil becomes small. Figure 3 shows that optical vignetting is no longer a concern: the oblique beam accepted by the entrance pupil is narrow and no longer clipped by the lens barrel.
Note that figures 2-6 do not show refraction of the yellow and orange beams. In this regard it should be realized that the lens elements in the sketches merely serve as a guide to the eye. Refraction is indirectly taken into account by the position and size of the entrance pupil (Zeiss data) and the image of the rear rim (calculated). The beams are correctly drawn up to the point where they hit the front element, and should further only be considered relative to the colored circles. In passing, figure 3 does away with the myth that a lens employed at a small aperture uses only a small part of the front element. The intersection of the yellow pencil with the front element marks the section needed for the image center, the intersection of the orange pencil with the front element marks the section needed for the image corner. So although each image point uses only a small part of the front element at a small aperture, the image as a whole still relies on a large part of the front element.
The brochure lens hood for the Planar 50/1.4 is metal hood #4. It is a wide hood which just clears the front element: figure 4. Since the hood does not affect the clear aperture for obliquely incident light (cf. figure 2) it does not lead to vignetting. Flare protection is offered without the slightest compromise to the design whatsoever.
Figure 4. The Planar 50/1.4 equipped with Contax metal hood #4. Look here to examine the real thing.
If the length of the hood is increased by 15 mm, the situation of figure 5 is established. At full aperture the clear aperture for the oblique beam is reduced and the image corner receives less light than it would in the absence of the hood. Mechanical vignetting sets in. By contrast, the lens hood has no effect at f/11. At this aperture the light beams accepted by the lens are not hindered by the hood. Thus, figure 5 represents a situation where mechanical vignetting is cured by stopping down the lens. An alternative view is given in figure 7, which shows illumination curves for the scenarios sketched in figures 2-6. At f/1.4 mechanical vignetting manifests itself by a corner illumination that goes down from 30% to 20%. The decline is gradual however and may not even be noticed in real-life images. As a matter of fact, depending on the application a small amount of (additional) vignetting may even be tolerated in favor of a better flare prevention. The curve for f/11 is identical between the upper left and upper right plots in figure 7: the hood has no effect on the image illumination at small apertures.
Figure 5. The Planar 50/1.4 with a lens hood that is 15 mm longer than lens hood #4. Vignetting now sets in at f/1.4.
When the length of the hood in figure 4 is increased by 30 mm (which happens to correspond to metal hood #5), figure 6 results. At f/1.4 only a small area survives to illuminate the image corner. The corner is dark but not black as it does receive some light. However, at f/11 corner blackening is a fact. Here, the lens hood completely obscures the entrance pupil (the small circle in the circle plot) and no light is passed on to the image corner. Where a small aperture cured the vignetting in figure 5, it worsens the vignetting in figure 6. The lower left illumination charts in figure 7 corroborate the corner blackening at f/11. The sharp kink in the f/11 curve implies an abrupt brightness transition from the image towards the corners, which are completely black.
Figure 6. The Planar 50/1.4 with a lens hood that is 30 mm longer than lens hood #4. Vignetting is manifest at all apertures.
Still longer lens hoods lead to black corners at all apertures and the main effect of the f-stop is found in the abruptness of the transition. This is illustrated by the fourth graph in figure 7, which results from the addition of yet another 15 mm to the hood.
Figure 7. Calculated image illumination for the Planar 50/1.4 in combination with several lens hood lengths.
From figures 5 and 6 it appears that the optimum length of a lens hood depends on the aperture. When the chief ray (the ray that goes through the center of the aperture) is not obstructed by the hood, stopping down the lens cures mechanical vignetting. When the chief ray is clipped, a small aperture leads to black corners. Unfortunately it is quite cumbersome to put this knowledge in practice. A lens that is regularly used at various apertures requires hood adjustment each time another f-stop is chosen. Impractical, but it can be done with a continuously variable, compendium type lens hood. A nice description to figure out the optimum length, by inspection of the exit pupil rather than the entrance pupil, is available as a pdf file [1]. The author allows some 20% pupil area obscuration by the hood because he considers protection against flare more important than a small, gradual decrease in corner illumination which is not noticed in the image anyway.
A simple, practical approach to determine whether a certain hood (or filters, or a combination) causes vignetting on a certain lens consists of a series of test exposures. The subject should be an evenly illuminated object at a large distance. Vignetting is less of a problem at close range than it is at infinity, so when infinity poses no problems a nearby subject is also safe. A brick wall on an overcast day will do fine. Four exposures are required, two at the lens full aperture (with and without hood) and two at the smallest aperture (again with and without hood). Slide film is preferred because of its short dynamic range, which will reveal vignetting more convincingly than print film. If the slides taken with the extension(s) show no additional corner darkening in comparison with the slides taken without, you are completely safe. If there is a slightly increased, gradual corner darkening at full aperture, you probably won't notice the presence of the hood with other subjects than a brick wall or a blue sky and you are also safe. Black corners however are generally considered gruesome and the extension (or combination of extensions) is just not suited for the lens.
Hood shape
So far the discussion involved circular lens hoods. Indeed, a circular lens hood has the same rotational symmetry as the lens and aesthetically matches the round image formed by a photographic lens. However, the round image is not fully used as the presence of a field stop, a 36x24 mm mask in case of a 35-mm camera, crops the image to a rectangular section. This has important consequences for lens hood design and the optimum lens hood is not round. The accepted light cone that is used to illuminate the frame is pyramidal. At full aperture, going from the lens towards infinity, the cone starts out circular at the front element and converts to a rectangular cross section at some distance. At small apertures, and depending on the design, the cross section of the cone is already rectangular at the position of the front element. The pyramidal cone is illustrated in figure 8.
| |||||||||
In the illustrations that follow a variety of hood shapes pass in review, designed not to introduce additional vignetting. The length of the round hood in figure 9 is such that it touches the light cone at four corner points. Voids in the plane of intersection evidence the shortcoming of a round hood: there are gaps where nonimageforming light may enter the system and introduce flare. One method to fill these holes is to extend the round hood to create the hood in figure 10. This so-called tulip hood is shaped by the intersection of a cylinder with a pyramid. Occasionally the designation butterfly hood is encountered.
| |||||||||
Another strategy resorts to a rectangular shape. Figure 11 exhibits a rectangular hood with the same cross sectional area as the hoods in figures 9 and 10. Both the tulip hood and the rectangular hood are more effective than the round hood. Not only because they are longer, but also because their shape is matched to the pyramidal cone and leaves no holes. A rectangular hood reduced to the same length as the round hood in figure 9 would still be more effective.
| |||||||||
Although the tulip hood in figure 10 is very effective with respect to flare prevention, it won't win a compactness popularity poll. To sacrifice some effectiveness for convenience, the two longer butterfly wings may be clipped to yield the chopped tulip hood in figure 12, which requires significantly less space in the camera bag. Finally, some of the effectiveness may be regained by filling the two gaps that arose in the clipping procedure. The 'chopped and capped' hood in figure 13 is relatively compact and still offers an excellent protection against flare. Zoom lenses are often provided with a chopped tulip hood that offers reasonable protection at the wide end, but which is inadequate at the tele end. Nonetheless, it is better than nothing.
Final remarks
This lens hood page has some overlap with the flare and vignetting pages, which are recommended reading to recognize the usefulness of a lens hood as well as the danger of overdoing it.
By PA van Walree
