Infrared with the M system

Fedia Le Grill wrote:

Wow if this graph is true I want a canon 350D immediatly :

https://www.researchgate.net/figure/Spectral-sensitivity-of-unfiltered-Canon-350D-sensor_fig1_229046541

the peaks in red first, then green, then blue seem distinct enough to produce a real three-color image only with the 700 to 1000nm IR spectrum !

The thing is I have a canon 1000D full spectrum (wich is only 3 years more recent than the 350D, it has a different sensor though...) and I never noticed that they were any color variation in the tones it produced when I use it with a 720nm filter....

Does it mean that the that objects reflecting IR refelect all IR wavelengths the same way ? That it is impossible to extract Three primaries like for exemple 700nm, 800nm and 900nm since all those wavelengths are reflected the same way ?

I did notice though that when I use a 720nm filter on the camera and I add a GRB3 filter that cut the longer IR wavenlegths the image color slide towards Red. I suppose that If I added a 850nm filter that cut the shorter IR wavelength the image color would slide toward Blue or deep violet. I dont have this filter just now, but I ordered it.

It certainly doesn't look right to me.

I'd expect an unfiltered sensor to show some sensitivity below 400nm & quite a bit more in the 900-1000nm region...

The IR peak in the blue channel also looks out of proportion. I have seen subtle increases in blue for some wavelengths above 700nm (often showing up dyed hair) but I've not seen other spectra showing much more than 10% variation between the channels above 800nm.

Usually sensors have a distinct red bias between 700 & 800nm with relatively similar transmission in all three channels above that.

With the right WB there can be a reasonably amount of colour variation with 720nm shots, (not something I've seen with 860nm & 960nm filters). Some objects such as foliage tend to reflect all wavelengths of near infra red, but this is certainly not true of everything.