No clue- Sounds like you just changing values without understanding what they supposed to be doing. This software you talk about and the arbitrary values. Its perfect for building into projects - industrial, scientific, or just fun - and much more affordable than most thermal cameras. I have a Flir Lepton 3.5 on a PureThermal 2 Board and am trying to download radiometric images down to a Raspberry Pi 4B using the uvcradiometry.py program which in turn uses the libuvc tool. (They recommend 25Mhz) Did you provide that? A sophisticated, hackable, 32x24 pixel thermal camera breakout Use it to monitor the temperature of your CPU or coffee pot, or to build your own heat-seeking night vision camera. The FLiR sensor will not be able to see any heat signatures behind the acrylic. The FLiR sensor only is able to detect temperature gradients on the surface of the material. I suspect they did not include a crystal in the LEPTON to keep the size down and embedded devices usually have access to some kind of constant clock any way. I tested a case out using a clear acrylic, FLiR attached to Raspberry Pi, and a Peltier Cooler. Now we know Raspbian is not very great at generating consistent clock rates consistently so you may be experiencing problems with this - The same way may other FLIR'ers are experiencing. I assume you got it hooked up to SPI correctly but one of the problems seems to be the Input Clock rate. So looking around really I noticed the protocol used on the Lepton is VoSPI (Video over SPI) Frankly never heard of that before but its plausible because the frame size is 80 x 60, which is not a crazy amount of data with some kind of AGC applied format which is selectable (Did you select the correct one?)

Raspberry pi flir