I'm trying to record water droplet impacts with a high speed camera. I would like to record the impact from the side and below (through the glass slide that the droplet hits). Due to lab constraints, only one camera is available. In my current setup detailed in the photo below, the glass slide intersects the image with a mirror stationed just below. However, due to the shallow depth of field and differences in path length, I've been unable to have both views in focus.

The goal is somewhat hard to explain so I've provided some crude drawings to explain the situation. The top half is what my original setup was and the bottom is what I want to see in the photo/video:

I understand this is fairly ambitious, any help is appreciated. Thanks.


3 Answers 3


A few ideas:

More mirrors to make the path lengths match

The first is to use more mirrors such that the path lengths are much more similar, while hopefully not increasing the total by very much. Here's a sketch based on yours. The up and down parts of the extra path in the top beam are each half the lenth of the extra path distance in the bottom beam.

More mirrors

It is of course possible that you don't have space for this

Barlow lens

The second idea involves using a diverging lens in front of the camera lens for the mirror path only. In microscopy this is referred to as a Barlow lens (note that the term is used slightly differently when applied to telescopes).

This would need to be a weak negative lens, ideally an achromat but a meniscus lens might be enough. It would need to be mounted square to the optic axis quite near the existing lens and aligned in three dimensions, with a reasonable amount of travel in the focus direction. You may need to have the lens cut down to clear the path to the other image.

Commercial split field filter

A "split field filter" (link to a Cokin product but I suggest a google search) would seem like a good idea, but I can only see +1 to +3, which are probably too strong. This is a little like a manufactured version of what I describe, but without all the adjustment. Also the filtered bit would have to go in the shorter path. The lack of translation along the optic axis means that tyou have little control over the distance ratio that's in focus.

Tilt-shift lens

A tilt-shift (or just tilt) lens is designed to do things like this. By introducing a slight angle between the lens and the camera the magnification and focus plane change across the image. The effect on the magnification may be tolerably small or calibrated away. This adjustment is normally built into the lens. I've seen various projects to add tilt to an existing lens, they all at as extension tubes as well, reducing the maximum focal distance.

These lenses are expensive, and because the effect is continuous you may struggle to get enough depth of focus within a given half-image

  • 1
    \$\begingroup\$ "more mirrors" -- as a longtime engineer in various field of optics, I can attest that this is the way to match path lengths. \$\endgroup\$ Jun 2, 2016 at 11:39
  • 2
    \$\begingroup\$ @CarlWitthoft I came to this from the point of view of having designed and built laser systems, which is why I thought of it. The tricky bit is likely to be squeezing everything together given the mirror thicknesses. \$\endgroup\$
    – Chris H
    Jun 2, 2016 at 12:15
  • \$\begingroup\$ Thanks for the input. Considering the scale of my setup, I'm tending towards adding the mirrors to match the path lengths. I'm having trouble finding resources on the split field filter. I understand how it works, but I'm confused on the +x parameter. It seems like the higher the number, the smaller the ratio between the distances of the objects in focus. \$\endgroup\$
    – Sam
    Jun 3, 2016 at 18:14
  • \$\begingroup\$ I think it's the other way round on the filter. They're the same as a close up filter only cut in half, if that helps. \$\endgroup\$
    – Chris H
    Jun 3, 2016 at 19:08

Looks like you have a triangle. ABC where

  • A - distance from camera to drop
  • B - mirror to drop
  • C - mirror to camera.

A < B + C Always.

So you need to position so (B + C) - A < DOF distance. To achieve it you can:

  1. Go to higher F-stop number
  2. Move mirror as close to drop as possible
  3. Move camera further from drop.

And there is always one more way take 2 shots ;)

  • \$\begingroup\$ Unless you've got gain (which we normally describe in terms of ISO) to spare, a smaller aperture will mean a slower shutter speed which might not be a good idea here. Similarly increasing the distance alone will mean filling fewer pixels, so a longer focal length would be needed. \$\endgroup\$
    – Chris H
    Jun 2, 2016 at 5:52
  • \$\begingroup\$ @ChrisH It is just a limitation. You can also mention diffraction from high f-stop number. Longer lens would give you less DOF. Any solution including yours would be a hack of some sort. That would potentially degrade quality of result and make setup more complicated. And you forgot one more option 2 synced cameras might be a better and easier solution. \$\endgroup\$
    – E_p
    Jun 2, 2016 at 14:19
  • \$\begingroup\$ I didn't forget synced cameras, I accepted the stipulation in the question. \$\endgroup\$
    – Chris H
    Jun 2, 2016 at 15:23

I think you may be at the limit of diminishing returns. Short of getting a custom-ground meniscus lens to compensate for the different optical paths, the solution looks complex (read a few light-losing surfaces).

The way out of this box will be to use two cameras and synchronize the views electronically, after the shooting session. That will give you more flexibility for choosing optimal viewing and lighting angles for detail rendition. I realize this suggestion is a deal breaker; but, being realistic about getting the results you want doesn't preclude the possible necessity given your demands.

The sheet of glass separating the upper from lower angle will be problematic for lighting and detail obstruction of the impact surface because its leading edge will be closer to the lens than the subject and appear to be out of focus. The blur will obstruct the impact surface at the subject distance.

Avoiding spatters and spills can be easier with different camera angles because you do not need to compromise focus for accessability to the setup.

Synchronizing two high-speed cameras could also be a nightmare without some kind of a sync-generator. Some cameras have optional ability to be synchronized externally.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.