How to design the perfect lens hood (for 35mm format)?

[snusmumriken]

If I wanted to make the perfect lens hood for an existing lens, what is the best way to go about it? Is there any theory that will help, or is it a matter of cutting away material until there is no longer any discernible vignetting?

 

[loccdor]

I think theory could only help you if you had the lens loaded up in software that traces rays. I'd go for the easier, less technical approach which involves sticking your fingers in the image corners until you can just barely see them (with SLR/TLR or ground glass) to get the initial idea about the size and angle you can make the hood. You could also shoot a paper circle at the desired distance from the camera lens, if it's able to focus close enough.

But why does it need to be perfect? Light is a phenomenon that follows the inverse square law so "close enough" gets you pretty much there. For example I think there'd be very little practical use for a hood that's longer than its diameter, even though technically speaking the most optically perfect hood would extend to infinity.

 

[Don_ih]

Given the focal length of the lens and the coverage, the angles needed to design your lens hood should be known. They can basically follow the diagonal angle of view of that lens (a value that can be calculated - or looked up on this handy table from Wikipedia

 

[loccdor]

That could be close enough. Beware that some lenses are not precisely their stated focal length, and some lenses distort.

 

[OAPOli]

It is possible to make a hood that's specific to a lens and to any format. But it has to be tuned to the entrance pupil (i.e. to a given f/stop). And a "perfect" lens hood would have to be infinitely long.

Here's a sketch:

The clear rectangle is the lens barrel, the blue ellipse is the entrance pupil, the blue rectangles represent the lens hood (its opening and length shown with the arrows) and the blue lines represent the FOV. The red lines represent the stray light cone that still gets into the entrance pupil despite the lens hood. The lens hood has to follow the FOV lines to avoid vignetting and to minimise the stray light.

First you need to locate the pupil. Prop the lens horizontally then use a digital camera on a tripod + nodal slide to focus on the front of the lens barrel. Stop down the lens then move the nodal slide until the aperture is in focus. The distance moved is the distance of the pupil with respect to the front of the barrel (or whatever reference point you used).

The FOV a for each side of the image is tan(a/2) = s/2/f where f is the focal length and s is the image side (e.g. 24, 36, 56mm etc...). Using similar triangles plus the size of the entrance pupil (f/f-no) you can determine the size of a lens hood of a chosen length (w.r.t to the pupil) and for each side of the image. You can see how a hood made for the full pupil will have to be shorter than necessary when the aperture is stopped down. Or how a round hood made using the image diagonal will be less effective.

For extra nerd points, round the corners of your custom hood with the same radius as the pupil.

Edit: This works for infinity focus but in principle the FOV changes a *little* bit when focusing to a closer subject.

 

[MTGseattle]

Out of curiosity, does the manufacturer not have a dedicated hood for the lens in question?

 

[snusmumriken]

Many thanks for those really helpful pointers, everyone.

But why does it need to be perfect?

To my mind, 'perfect' is the best possible compromise between preventing flare from light sources outside the image area, and causing vignetting. It hadn't occurred to me that this varies with aperture (thank you, @OAPOli) - and also whether a filter is fitted, I suppose. Evidently unless the hood is expandable, 'perfect' is going to be more of a fudge than I imagined.

Out of curiosity, does the manufacturer not have a dedicated hood for the lens in question?

Yes it does. It is a built in collapsible design, which I find unsatisfactory. Although you can maybe guess which lens this is, I don't think it would be fruitful to get into that territory. I have not been able to find an after-market hood designed for this particular model. Even if I did, I wouldn't know how to judge whether it is optimally designed, without years of trial and error. Likewise, I don't know whether the generic hood I actually use is as good as it could be. Hence the theoretical considerations described above are very helpful.

Please humour me if you think I am being obsessive, but I know I'm not the first.

 

[Pieter12]

This might help.

 

[OAPOli]

@snusmumriken it would a be bit silly but not inconceivable to have different hoods for different f-numbers, especially if they are 3D-printed for a very small cost.

 

[snusmumriken]

@snusmumriken it would a be bit silly but not inconceivable to have different hoods for different f-numbers, especially if they are 3D-printed for a very small cost.

I think that would be extreme, but I could probably justify two. I need to follow through the calculations to see how different they would be.

Thinking about the reasoning you posted above, I now understand how the tulip-shaped hood design comes about. I will extend the spreadsheet approach (thank you, @Pieter12) to calculate hood length right around the optical axis and generate such a design directly. Maybe it will turn out to be impractical, but I am curious to know…

 

[cliveh]

The best lens hood I have ever seen is the werra: -

Metal Super LENS HOOD 30.5mm 30.5mm 30.5 screw for Werra pentax ? rollei 35 ? | eBay UK

Fully Functional. I always take a lot of pictures. ✧ ✧ - Usable. ✧ ✧ ✧ - Good. ✧ ✧ ✧ ✧ ✧ - Excellent.

www.ebay.co.uk

 

[OAPOli]

@snusmumriken I don't follow the posted spreadsheet. It starts with random hood dimensions then derives its length (from where is this length measured?). It should be the other way around: select the longest practical length for the hood then derive its dimensions. For a hood length l from the entrance pupil, the inside opening of the hood is t = s/f x l + f/f-no where s is the length of the image side.

 

[snusmumriken]

@snusmumriken I don't follow the posted spreadsheet. It starts with random hood dimensions then derives its length (from where is this length measured?). It should be the other way around: select the longest practical length for the hood then derive its dimensions. For a hood length l from the entrance pupil, the inside opening of the hood is t = s/f x l + f/f-no where s is the length of the image side.

Yes, that spreadsheet needs re-working, for sure. Using a spreadsheet is a good idea, but I'm starting again from scratch.

 

[Don_ih]

If the angle of your lens hood cone matches the diagonal of your angle of view, you can make the hood 20 miles long if you want. Practically, a lens hood that sticks past the front element by an inch and a half is likely all you'll need. Math is one thing - reality is another.

 

[xkaes]

That table doesn't take into account the filter diameter of the lens. A 24mm with a 49mm filter thread needs a different hood than a 24mm with a 62mm thread.

The table also doesn't take into account the use of filters.

 

[snusmumriken]

Can any of you comment on those rectangular hoods where the edges are canted inwards?

 

[MTGseattle]

I think I can guess which camera/lens, which does complicate things a bit. Have you tried a silly collapsible rubber hood screwed onto the filter threads while leaving the built-in hood retracted? It would be low-tech and low budget but give you some idea as to where you could go. I would also think measuring the protrusion of the factory hood would give a good baseline for a "not to exceed" protrusion measurement for any other hood?

 

[snusmumriken]

I think I can guess which camera/lens, which does complicate things a bit.

Have you tried a silly collapsible rubber hood screwed onto the filter threads while leaving the built-in hood retracted? It would be low-tech and low budget but give you some idea as to where you could go.

I have a generic no-name screw-in metal hood. It provides more mechanical protection than a rubber hood, and doesn’t cause vignetting even with a filter between the lens and the hood.

I would also think measuring the protrusion of the factory hood would give a good baseline for a "not to exceed" protrusion measurement for any other hood?

The built-in hood is straight-sided, and quite a bit shorter than my conical generic hood. But yes, I could work out the equivalent angle by making a scale drawing. The built-in hood protrudes the same amount from the front element of the lens, whether or not there is a filter in place over the latter. So it seems quite a conservative design, to use as a minimum yardstick, surely?

 

[MTGseattle]

I'll bow out to others with more experience and better math skills, but we know for sure that the factory hood does not vignette, so that's why I suggested it as a starting point.

 

[Sirius Glass]

If I wanted to make the perfect lens hood for an existing lens, what is the best way to go about it? Is there any theory that will help, or is it a matter of cutting away material until there is no longer any discernible vignetting?

To get the fastest and easiest solution, find the lens's lenshood made by the manufacturer.

 

[snusmumriken]

To get the fastest and easiest solution, find the lens's lenshood made by the manufacturer.

Thanks, but we’ve already covered that in earlier posts.

 

[xkaes]

Add petal hood to the mix, and you can really go nuts!

 

[OAPOli]

@snusmumriken I've 3D-printed lens hoods using the method I posted above. It's nice because you can design them into a rectangular shape and for a specific f-number for maximum effectiveness. Designing for a light friction fit of the hood onto the barrel will allow you to use filters and to properly phase the rectangle.

 

[snusmumriken]

@snusmumriken I've 3D-printed lens hoods using the method I posted above. It's nice because you can design them into a rectangular shape and for a specific f-number for maximum effectiveness. Designing for a light friction fit of the hood onto the barrel will allow you to use filters and to properly phase the rectangle.

Thanks for your encouragement. That’s the route I will go down.

I’m still wondering about those rectangular hoods with the edges reflexes inwards, eg like all recent Leica and Fujifilm hoods. The result is reminiscent of the way James Ravilious adapted hoods by trial and error using electrical tape. What are the advantages of that design (shorter for same effect?) and how would one calculate the dimensions?

 

[OAPOli]

Thanks for your encouragement. That’s the route I will go down.

I’m still wondering about those rectangular hoods with the edges reflexes inwards, eg like all recent Leica and Fujifilm hoods. The result is reminiscent of the way James Ravilious adapted hoods by trial and error using electrical tape. What are the advantages of that design (shorter for same effect?) and how would one calculate the dimensions?

I don't know what you mean by "edges reflexed inwards". Didn't Ravilious use tape to get a flat rectangular opening in his hoods? To make a rectangular hood, locate the entrance pupil, then compute the openings according to the chosen length and using 24 and 36mm as the image sides. Formula is in post #12.