MAybe we have someone who knows optics better than I (Peter?) here. I have a problem with our video remote viewing I'd like to really fix, and I'm almost there, but not quite. Near IR, even in CCD cameras that have the IR block filter, shows as purple in the image. It even looks like the D spectrum. This is a huge problem for remote monitoring of the fusor, as we are running the grid back end near the melting point of the copper feed through, and want to be able to judge temperature fairly accurately there - hard to do if everything in the picture is purple. (Edit, that was a high end webcam, and a single frame grab from a video, we can and do do a lot better)
So, I've been testing this and that, hoping to avoid paying $70 or so for an IR block that might or might not work well enough. Since it's close to the same temperature, and well, here where I'm working on a major raspberry pi project for streaming video (it has FAR less latency than the VLC/webcam -> network -> mplayer which is the fastest way to do a web cam latency wise), I'm working with a Mr Heater as my test source.
I had an old welding goggle glass, the kind that has a LCD "instant dark", which had gone bad, but I saved the many layer laminate for who knows what reason. Well, after some belt sanding around the edges to heat and remove some of the glue, I managed to pry off some small pieces of the thinner outer layer, which seems to be a pretty good IR block at the near end where it matters (the camera can't see my soldering iron cranked to 900f - that's far IR).
Doing this shattered the filter layer (that's what I assume it was since it works, kind of) into small pieces, but the pi camera lens is tiny...so I taped one over it. BTW, these are originally high rex pix that I severely cropped - the Mr Heater was 20 feet off.
At 10 feet, you can read the lables on packages with this thing, it's better than you'd think (around 4 megapix, though the glass probably isn't as good as that).
Note that at least on my monitors here (they tend to be the best you can get FWIW) that's still not quite right - the real thing is orange at the top of the grille, and dull red/orange at the bottom. There's more work to do here. I think I need either a harder cutoff, or one closer to visible. I'm mainly interested in viewing fusor focus (man, this is going to rock at that range, about 15") and how hot things are getting. I don't want to be shutting down or rreducing power because this is showing me things hotter than they really are. Just one more little project to eat up time.
I turned up the brightness and the contrast on the second picture in the pi interface to make them as close as possible - the filter has a couple stops of attenuation at visible too.