Infrared 3-band color images (780, 830, 1064 nm)

[Andy Perrin]

No, that's not far enough for thermal effects to be discernable near room temperature. You would have to get pretty warm. (The blackbody peak is at (2898 K-um)/(3um) = 966 K = 693 C...)

And I don't want to resort to a scanning photodiode unless I have no choice. Sloooow, and poor image quality. I would rather just find myself an appropriate sensor.

[Stefano]

You are outside the range for upconverting phosphors, right? Are there still electronic transitions? Photons are weak there.

[Andy Perrin]

No, electron transitions are not the main mechanism for light interaction, it's mostly phonons (vibrations) and rotational.

https://en.wikipedia...ed_spectroscopy

 

MCT cameras are a possibility.

 

You are outside the range for upconverting phosphors, right?

Yeah to heck with upconverting phosphors. I tried them (you linked me above!). They are awful. Low low low efficiency required light so intense I nearly set the samples on FIRE with my 600W halogen. Also, they absorb in a very non-linear way, so it's hard to be sure what wavelengths you're seeing.

[Stefano]

Look at what I found on the Wikipedia link you put:

 

https://en.m.wikipedia.org/wiki/File:Ethanol_near_IR_spectrum.png

 

it almost seems like that most things are purposely absorbing beyond the sensitivity of silicon sensors. Of course it's a coincidence, but it's pretty annoying that you have to use InGaAs or germanium or other means to see that. NIR up to 1100 nm is quite boring, the interesting stuff is in SWIR.

[Andy Perrin]

it almost seems like that most things are purposely absorbing beyond the sensitivity of silicon sensors. Of course it's a coincidence, but it's pretty annoying that you have to use InGaAs or germanium or other means to see that. NIR up to 1100 nm is quite boring, the interesting stuff is in SWIR.

It's personal, Stefano. The materials of the world all hate you. I'm sorry, I don't make the rules.

[dabateman]

 

It's personal, Stefano. The materials of the world all hate you. I'm sorry, I don't make the rules.

 

So proof there is a creator? Don't have to wait for the Babel fish.

[Andy Perrin]

Nah, Stefano is clearly an animist.

[Stefano]

Well, you have to admit that it is a pretty odd coincidence. Just like foliage, it turns white right when you step away from the visible spectrum, it is already reflective before 750 nm.

[dabateman]

Look at what I found on the Wikipedia link you put:

 

https://en.m.wikipedia.org/wiki/File:Ethanol_near_IR_spectrum.png

 

it almost seems like that most things are purposely absorbing beyond the sensitivity of silicon sensors. Of course it's a coincidence, but it's pretty annoying that you have to use InGaAs or germanium or other means to see that. NIR up to 1100 nm is quite boring, the interesting stuff is in SWIR.

 

Just read the link. Yes and works both ways. The alcohols also absorb just out of my reach into UVC, at 205nm to 180nm. IPA, I was using as control to see if I was seeing at 205nm, water comes in at 190nm.

 

Acetone does absorb well at 280nm though. So if you loose the sense of smell, you can still see it in UVc. But then you will loose your sense of vision, using UVC light. So then just left with how it melts plastic, before it causes damage to your hands.

 

You know best to stay away from acetone.

 

[bvf]

Well, some of the filters that I ordered have arrived. Not the filters that I'm intending to use and awaiting, but I felt I had enough to do at least a Proof of Concept.

 

The flters I used were:

 

Short wavelength: ZWB1 + R72 (blue channel)

Mid wavelength: Midwest Optical BP850 (green channel; this filter is the biggest problem, as it has a fat tail trailing into longer wavelengths)

Long Wavelengths: Zomei IR950 & Midwest Optical LP920 (blue channel; both filters resulted in virtually identical results).

 

Here are the curves for those filters - adjusted to include the effect of CMOS sensor sensitivity at the various wavelengths:

 

 

Here are the results. This was a very quick and dirty PoC: no refocussing (you can see the focus shift between the visible and IR images), no alignment (hence the various colour fringes). Final white balance was done against the PTFE tile (behind which is a 19% Grey target, which results in a slightly bluer end result if you White Balance aginst this).

 

The first image is standard visible.

 

The second is the IR Full Colour, using the same technique as I uaed for UV. There is some colour, but not a lot.

 

The third is the IT Full Colour with the saturation ramped up. So you can see there is some colour there. Perhaps I'll get a bit more colour when I get the other filters (Midwest Optical BP735, Midwest Optical BN850, Midwest Optical LP1000).

 

 

 

[Stefano]

Nice image

I notice one thing: apart from some blue, the only false color you have is yellow. Some VIS-absorbing NIR-reflective objects still absorb a little in the far red region (your blue channel), and so appear yellow.

You can note this with standard IR false colors too. If you white-balance a normal IR photo you get false blues for long wavelengths and false yellows for shorter ones (very similar to what happens in UV). This way your false yellow objects should appear false blue. It happens often to see false blue in IR, usually in flower centers which appear black in VIS (a Rudbeckia Hirta should appear with white petals and blue center when photographed with something like an Hoya R72).

[bvf]

Yes, I can see that a visually black fabric will continue absorbing as you get in the very near IR before it starts reflecting, and so you would get a depressed blue channel leading to a yellow colouring.

 

But the blue and green coloured objects are interesting to me. They (by definition) absorb red, and continue to absorb as you cross over into IR. But they too start reflecting again as you get a bit further into IR. So their reflectance spectra must be U-shaped, with the left-hand arm of the U being in the blue or green region and the right-hand arm of the U being in the 800+ IR region.

[Andy Perrin]

Nice result! You are definitely getting some variation.