NETD is Noise Equivalent Temperature Difference; basically this noise in the images you are looking at.
Simply put this is how well (at a defined set of parameters) the thermal sensor can discern between 2 temperatures.
It is measured in Millikelvin (mK). Typically from 25mK to 50mK. The smaller the figure the better the sensor can discern small temperature differences, a Lower Figure will generally result in a better image, with better contrast (BUT Read On)
Excellent NETD is a fundamental that must be present. One other big item to watch out for is the F number of the Lens, often cheaper and smaller lenses are used such as F1.2, this means there will be a loss of 20% so that 35mK sensor is really 42mK
The end result will also be affected by the quality of lens, F Number, The programming of the Core and the environmental conditions.
This is a image showing a lower NETD on the left, vs that of the right – you can see far more definition in the image on the left.
NOW here is the truth of it
You need a unit with a reasonable sensitivity - anything less than 35mk is fine
This is all about telling the difference between 2 different pixels, see the image below - its a great example of good (left) and poor (right) sensitivity. BUT look at the image, it fills some 70% of the screen, its a circuit board, right in front of the thermal camera.
1 You need a LOT of pixels on a target to see a difference.
2 You need to be close, the Enviromental's (Moisture, Pollution, Dust and distance) all reduce a cameras ability to see the target
Lets take a 1x1m target at 300m - and typical lens - you have just 9x9 pixels on the target, 81 pixels out of 110,592 ...
Realistically this is a BS marketed number and is relevant to an industrial or scientific use rather than a hunting; if you believe the marketing images of the pig that looks like these circuit boards below - you need to probably run - its RIGHT IN FRONT OF YOU