Test result in last post - Beam splitter cube vs plate
Moderators: rjlittlefield, ChrisR, Chris S., Pau
I am reviving this old thread because I had an idea which might be relevant. Beamsplitters or semi-reflective mirrors in the optical path are discussed above, but these cause aberrations. We could eliminate aberrations by using a 50% transmitting/50% reflecting front-surface mirror. Place the mirror between the light source and the subject at a 45 degree angle.Shine the light (through the mirror) directly on the subject, and aim the camera at the reflection of the subject (on the first-surface side of the mirror), not at the subject. No imaging rays pass through the glass, so this should be aberration-free. Should work, right?
- rjlittlefield
- Site Admin
- Posts: 23041
- Joined: Tue Aug 01, 2006 8:34 am
- Location: Richland, Washington State, USA
- Contact:
Yes, that sounds right.Lou Jost wrote:Should work, right?
With flat glass, some rays will get through the front surface, bounce off the rear, get through the front again, and become imaging rays with an offset and aberrations.
Anti-reflection on the rear would help, and that is commonly available.
Increasing the reflectivity would also help, for example 75% reflection will improve the ratio of first- to second-surface rays by 6X compared to 50%. [Working through that number... If the rear surface is 5% reflective, then a 50/50 split gives .5 front surface versus (1-0.5)*0.05*(1-0.5) second surface for a 40:1 ratio, while 75/25 split gives 0.75 versus (1-0.75)*0.05*(1-0.75) for a 240:1 ratio. Apply an anti-reflection coating for 1% on the back, and both ratios improve by 5X so the 75% reflectivity gives 1200:1, really not an issue.]
Conceptually, it would be even better if the mirror were formed on the slant surface of a prism, rather than on a flat piece of glass. That way the back surface would contribute no structured rays, only veiling glare.
Seems like a prism with a half-silvered face should be simple enough to manufacture, but on quick search I couldn't find one.
--Rik
- rjlittlefield
- Site Admin
- Posts: 23041
- Joined: Tue Aug 01, 2006 8:34 am
- Location: Richland, Washington State, USA
- Contact:
That sounds easier to implement than precisely positioning a mirror at 45 degrees, for me at least. Sorry, I cannot help, as I know nearly zero optical math/physics. But who else do you need, when Rik is hereLou Jost wrote:Could we glue a flat half-silvered mirror to an ordinary prism using lens cement?

Selling my Canon FD 200mm F/2.8 lens
Unless I'm missing something it doesn't have to be 45º?
It used to be snappy to do things like this in Word but they broke it in the version I have now......
Not to scale: Light (yellow arrow) passes through the part-silvered mirror towards the object (star).
The spacing of the ghost images, at (eg) 10%, 10% of 10%, etc strength will alter as you tilt the mirror.

It used to be snappy to do things like this in Word but they broke it in the version I have now......
Not to scale: Light (yellow arrow) passes through the part-silvered mirror towards the object (star).
The spacing of the ghost images, at (eg) 10%, 10% of 10%, etc strength will alter as you tilt the mirror.

Chris R
Yes, I guess the angle just changes the perspective, which could be very useful.
I've got some front-surface mirrors to enable use of a horizontal stacking set-up on vertical subjects like specimens in alcohol. Trouble is that most of these mirrors are not very flat (4-6 lambda for Edmund Scientific mirrors). Some from specialized stores are flatter. These guys have 1/8 lambda mirrors:
https://opticalmirror.com/eighth-wave-g ... ce-mirror/
Will try to get one of these. But these are fully silvered, transmitting only 4% of the light. They also sell inexpensive anti-reflection beamsplitters but only with reflectiveness <=50%.
Edit: Someone may wonder why I'd use a horizontal set-up for a subject that needs a vertical perspective. The reason is that horizontal set-ups on a rail can support heavier lenses. Some of my lenses weigh 20 pounds or more! Though I have springs (Enrico's suggestion) to lower the effective weight on a vertical set-up. Still haven't had time to build this though.
I've got some front-surface mirrors to enable use of a horizontal stacking set-up on vertical subjects like specimens in alcohol. Trouble is that most of these mirrors are not very flat (4-6 lambda for Edmund Scientific mirrors). Some from specialized stores are flatter. These guys have 1/8 lambda mirrors:
https://opticalmirror.com/eighth-wave-g ... ce-mirror/
Will try to get one of these. But these are fully silvered, transmitting only 4% of the light. They also sell inexpensive anti-reflection beamsplitters but only with reflectiveness <=50%.
Edit: Someone may wonder why I'd use a horizontal set-up for a subject that needs a vertical perspective. The reason is that horizontal set-ups on a rail can support heavier lenses. Some of my lenses weigh 20 pounds or more! Though I have springs (Enrico's suggestion) to lower the effective weight on a vertical set-up. Still haven't had time to build this though.
Sorry, I don't understand some points, could you clarify?
I'm not able to further elaborate it, but maybe polarized light could help as some half mirrors do polarize the light
What kind of prism? If the goal is to illuminate through it some prisms will decompose the light...Conceptually, it would be even better if the mirror were formed on the slant surface of a prism, rather than on a flat piece of glass. That way the back surface would contribute no structured rays, only veiling glare
Unless I'm missing something it doesn't have to be 45º?
This would not eliminate the ghost image, just to displace it, and transmission and reflection rates will be altered, being smaller in Chris's drawing.Yes, I guess the angle just changes the perspective, which could be very useful.
I'm not able to further elaborate it, but maybe polarized light could help as some half mirrors do polarize the light
Pau
- rjlittlefield
- Site Admin
- Posts: 23041
- Joined: Tue Aug 01, 2006 8:34 am
- Location: Richland, Washington State, USA
- Contact:
Pictures take a lot of time to draw, so let me try this in words...Pau wrote:Sorry, I don't understand some points, could you clarify?Conceptually, it would be even better if the mirror were formed on the slant surface of a prism, rather than on a flat piece of glass. That way the back surface would contribute no structured rays, only veiling glare
When the mirror is formed on a flat plate, then some light rays from the subject get through the front surface mirror and bounce off the back surface of the plate, headed toward the camera. Some of those will again get through the front surface mirror. It is these rays, originating at the subject, that will create a secondary image that may be offset and aberrated.
When the mirror is formed on the slanted surface of a prism, then when light rays from the subject get through the front surface mirror they just keep on going, headed back to the illumination source. The only rays that reach the camera from behind the mirror come from the back side of the prism, where they will probably be severely OOF and in any case will not correlate with anything on the subject.
A glass prism will surely bend different colors to different angles. But if the light source is broad, then even after bending, the center of the outgoing beam will still be well mixed, with the same spectrum as the incoming beam. Colors will appear on the edges of the outgoing beam, so you would have to make sure those don't fall on the area being imaged.What kind of prism? If the goal is to illuminate through it some prisms will decompose the light...
I read their spec as "96% reflective", and I bet that means 4% absorbed, 0% transmitted.Lou Jost wrote:These guys have 1/8 lambda mirrors:
https://opticalmirror.com/eighth-wave-g ... ce-mirror/
Will try to get one of these. But these are fully silvered, transmitting only 4% of the light.
Looking again at the numbers, I would be very inclined to try a 75% reflective mirror with anti-reflection coating on the back. Ask Edmund's technical support if they have one that is fractional lambda.
Probably also good to ask the folks at opticalmirror.com if they can do partial silvering instead of full.
--Rik
I would think custom jobs like this would be very expensive for just one item. Maybe if there is interest on the part of others, we could get something at a reasonable price per item. I think I should first try the 50% transmitting front-surface mirror glued to a stock anti-reflective prism with optical cement (wherever one might get optical cement from). It might be good enough.
- enricosavazzi
- Posts: 1449
- Joined: Sat Nov 21, 2009 2:41 pm
- Location: Västerås, Sweden
- Contact:
That is not a standard component, but a beamsplitter made by cementing the hypotenuse of two right-angle prisms (one of them typically 50% transmitting, 50% reflecting) is called "beamsplitter cube" and is a commonly used optical component.Lou Jost wrote:Could we glue a flat half-silvered mirror to an ordinary prism using lens cement?
--ES
Even without anti-reflective coating, 240:1 is about 7.9 stops, would it cause that much trouble?Looking again at the numbers, I would be very inclined to try a 75% reflective mirror with anti-reflection coating on the back. Ask Edmund's technical support if they have one that is fractional lambda.
I tried to do a coaxial light source but failed. However, maybe a commercial coaxial light device can be dis-assembled and re-purposed.
-
- Posts: 3236
- Joined: Sat Nov 20, 2010 10:40 am
- Location: Santa Clara, CA, USA
- Contact:
I have tried multiple times to get a decent coaxial lighting system together, but the problems of contrast and distortion have kept me from being fully successful. I just received a couple more thin 50/50 splitters and now have 3 to try when I get time. I have a few ideas about how to improve the system that need some testing, and hopefully they work well enough for coaxial to be my standard lighting system. We'll see...
Enrico, that design doesn't solve the aberration issue. The point of my design is that no image rays pass through any glass.
mjkzz, my design might solve the problems you encountered. Front-surface mirrors are available at lambda/8 flatness, which might work at least for the magnifications that Ray uses. Ray, do any of your beam-splitters work in the way that I proposed here? I would think someone makes these somewhere.
mjkzz, my design might solve the problems you encountered. Front-surface mirrors are available at lambda/8 flatness, which might work at least for the magnifications that Ray uses. Ray, do any of your beam-splitters work in the way that I proposed here? I would think someone makes these somewhere.