true MSX & thermal fusion (no parallax): possible?

[calel]

basically a visual + IR with 1 single lens, taking in 1 single image, where the sensor gets both the visual & IR image

this way no different angle problems & no parallax compensation necessary no matter how close the range
(would be perfect for examining PC components at close range eg. when overclocking)

in theory is that even feasible?

[Bill W]

Just about - but IMO unlikely to be worthwhile over either getting a decent thermal camera in the first place or dealing with the parallax in software.

You need a lens that is both IR and visually transparent and yet also with optical power (to be a lens)
So you might be looking at ZnSe (yellow, but transparent) salt (tricky when wet  ) or sapphire.

Then you need a beam splitter to direct the two images to the two detectors - probably achieveable with coatings.
That means a very long back focus lens design - adding cost

[calel]

problem with 2 separate lens is that even compensation not enough at close range cause it just "shifts" one image to superimpose it on the other right?

but suppose one of the lens is right above (perpendicular) to a hot square - then other lens won't see a perfect square (cause it's not right above the square) so the 2nd lens will be seeing a trapezius instead of a square with one side appearing larger than the opposite side. no amount of processing can compensate for that right?? it's impossible for a CPU to "guess" what an image would look like from another angle


btw I didn't know it was that hard to get this sort of "dual transparence". so ordinary glass like the glass on our windows, our glasses or telescopes, are not transparent to infrared light? 

[Fraser]

An interesting idea was implemented for long range non contact IR thermometers by companies like Minolta and Raytek. They wanted the optical viewfinder to look through a lens on the same axis as the IR temperature measurement lens for good spot measurement accuracy. The solution was a hybrid coaxial lens with a small visible light lens in the centre of a Germanium lens. The viewfinder operates via two prisms to move the visible light image out of the axis of the IR lens. The Germanium outer lens then passed the IR image to the thermal detector. Other approaches use a central visible light lens combined with a reflector type IR lens. There is an IR Thermometer that uses a CCD camera looking through the middle of a Germanium lens using a clear coaxial insert. I own examples of these IR thermometers, including the one with the integrated CCD camera

I have not seen a coaxial lens used for a thermal camera but such would be feasible, if expensive.

Fraser

[Fraser]

Some pictures of coaxial IR lenses

Fraser

[Bill W]

Given some of the sizes of visual cameras in phones etc, you could just stick the whole visual camera in the middle of a thermal lens !

Bill

[Vipitis]

I have mentioned it before somewhere, but I got a 'working' prototype. Using a hard carbon coated Ge window at an 45° angle. The LWIR camera almost looks straight through the window with very minor refraction.

And using the reflective nature of the uncoated rear of the Ge window and a visible light camera at 90° to the optical line of sight of the LWIR camera.

The lenses for each camera are independent because you focus on the reflection/the transmission and the Ge window works as a "cold mirror" to reflect the visible light.

The biggest issue is setting it both up and focussing well. The reflection loss is quite dramatic, but if you do this in daylight shooting at F/5.6 gives useable depth of field for example. The image needs to mirrored and Color corrected(or pushed into grey scale) due to the chromatic characteristics of the reflection.

I want to do some test shots where I show side by side and superimposed video, but it's quite difficult to set up in good fashion.


Similar optical designs using a hot mirror to reflect nIR but transmit visible are used in astronomy and even microscopes.

[DaJMasta]

I was looking at some similar cameras recently and came across JAI's Fusion line of cameras, though they have a few other varieties.  The idea with them is splitting the image with a prism/dichroic mirror and then digitizing on different sensors.  The limitation here is the passband characteristics of the material, though, and I think their offerings are limited to visual/NIR fields, as the longer wavelengths are normally absorbed by a lot of the substrates the mirror/prism is made out of, so I don't know if there's a material option that would work that well in the case of LWIR thermal cameras.

You could potentially just use a 50% reflective mirror to split the image for two sensors, but then each only gets half the light and loses out on dynamic range, plus there are probably some particular design challenges that come from that kind of splitting and handling the wide spectral bandwidth required to cover 400nm to 13um or whatever your use case is.

[Ultrapurple]

I picked up a ZnSe beamsplitter a while ago that is coated so it can hive off LWIR separately from M- and SWIR plus visible. It's (very) transmissive to LWIR and reflective for everything else. I'll test it with a MWIR camera when I next have one up and running.

One fine day I intend to make up a parallax-free dual-band imaging system of some kind - either LWIR + MWIR or LWIR + visible. I may even see if I can work out a way of going three ways - LWIR, MWIR and visible.

One fine day...

(Thermal image quickly snatched with a Seek Reveal Pro; hardly my finest-ever work).

[DaJMasta]

Ooh that could have some neat applications.  I'm not familiar with them, but do you know the coating/substrate/whatever is used to make it?  Could be neat to play around with if I can find one somewhere.

[calel]

I didnt know the whole LWIR/MWIR thing was important

what type of IR do commercial thermal cams (like the Flir ones) detect?

[Ultrapurple]

@DaJMasta - the substrate is ZnSe but I regret Ihave no more technical information beyond what's printed on the label.

I don't think optical coatings are the sort of thing that one can expect to be able to do at home. Keep your eyes open on eBay for 'beamsplitter' and check the specs: you'll see a ratio of >100:1 visible light vs LWIR beamsplitters (ie the ones we want are rare), but the characteristic orangey-yellow of ZnSe means you can just glance at the photos to dismiss the majority.

The beamsplitter in my photo came from the USA and I found it on ebay.com, if I remember rightly. I think I paid something like $100, which I'm sure is a fraction of its original cost.

In a quick search I managed to find one, just 1" diameter: eBay auction: #140453371372

[bap2703]

There are some attempts:
https://www.sciencedirect.com/science/article/pii/S2211379720317502