06/08/2026
With so many products on the market in the thermal realm now.
Knowing a bit about the technical side of the gear before you pull the trigger definitely helps.
https://www.facebook.com/share/p/192AvabjxU/
WARNING - TECHNICAL STUFF BELOW.
Tried to keep this as easy to read as possible, but its a technical topic and its one that you need to really understand when buying a thermal device. This is similar to the "640 sensors are better than 384 senrors"...... Anyone who tells you that a 640 sensor is better at longer distance doesn't understand what they are talking about. Anyway, back to NETD.....
Does a Lower NETD Automatically Mean a Better Thermal Image?
One of the most common questions we get is:
**“But doesn’t a lower NETD mean it will be better?”**
Not necessarily.
NETD is only one part of what makes a good thermal, but it has increasingly been turned into a marketing number that gets far more attention than it deserves.
NETD measures a sensor’s ability to separate very small temperature differences from its own noise. That is important, but it does not directly determine:
* Image sharpness
* Identification range
* Focus quality
* Fine detail
* Overall image quality
To put the numbers into perspective, NETD is measured in millikelvin, which are tiny fractions of a degree.
A 25 mK sensor can theoretically distinguish a temperature difference of around **0.025°C**, while an 18 mK sensor is around **0.018°C**.
On paper, 18 mK sounds much better. In reality, the difference at the sensor level is only **0.007°C**. That number alone does not tell you which device will produce the better image.
The Lens Matters Just as Much
We have been talking about lens quality for years because the lens, its aperture and the quality of its coatings can have just as much influence on the final result as the detector itself.
A thermal lens controls several things at once:
* How much infrared energy reaches the detector
* How sharply contrast is transferred to the sensor
* Field of view and base magnification
* Focus quality
* Distortion and edge performance
* The effective sensitivity of the complete system
The aperture of the lens, shown as its f-number, can also have a major effect on the NETD of the complete thermal.
A lower f-number generally allows more thermal energy to reach the sensor, but there is a trade-off: it reduces depth of field.
We have moved from using a lot of f/1.4 lenses to f/0.9 lenses. You may have noticed that newer devices often require more frequent focusing. That is not necessarily a fault with the device. It is partly because the depth of field is shallower.
A good real-world example is the FLIR E76. It is the same thermal platform, but FLIR publishes different NETD figures depending on which lens is fitted:
* 42° f/1.1 lens: less than 30 mK
* 24° f/1.3 lens: less than 40 mK
* 14° f/1.5 lens: less than 50 mK
The sensor has not changed. The lens has.
That alone shows why comparing two thermals using only the advertised NETD number can be misleading.
# # How Was the NETD Figure Measured?
Another problem is that manufacturers often do not publish exactly how their NETD figure was achieved.
To properly compare NETD between two products, you would need to know:
* Was denoising enabled or disabled?
* What lens and f-number were used?
* Was the test performed using a controlled blackbody or a real scene?
* What frame rate and integration time were used?
* What temperature measurement range was selected?
* What was the detector’s operating temperature?
* Was cooling applied to the detector?
* Was the result measured from the centre pixels or across the entire sensor?
The temperature used during testing also matters. Measuring NETD at 30°C may produce a different result from testing it at 50°C.
Unless both products were tested under the same conditions, the numbers are not necessarily comparable.
# # Then There Is the Software
Software is another enormous part of thermal image quality.
We regularly see firmware updates that improve image quality. If the final image was determined entirely by the detector’s NETD rating, those software updates would not make much difference.
Software controls things such as:
* Non-uniformity correction
* Bad-pixel replacement
* Temporal denoising
* Spatial denoising
* Automatic gain control
* Local contrast enhancement
* Edge enhancement
* Tone mapping
* Palette mapping
* Multi-frame processing
These systems can make two devices using similar detectors look completely different.
Good software can:
* Pull out weak contrast without filling the image with noise
* Preserve detail in fur, leaves, grass and terrain
* Keep the background visible without blowing out the animal
* Maintain stable contrast while panning
* Avoid bright halos around hot targets
* Reduce noise without creating noticeable lag
Poor or overly aggressive software can:
* Smear moving animals
* Remove fine texture
* Make the background look artificial
* Produce bright outlines and halos
* Pump the brightness as the scene changes
* Create an image that looks sharp only because of excessive edge enhancement
# # The Bottom Line
NETD matters, but it is not the final score for image quality.
A good thermal image comes from the complete system working together:
**The detector, lens, aperture, coatings, focus system, calibration, software and image processing all play a role.**
Buying a thermal device based only on the lowest advertised NETD is a bit like buying a car because it claims to make 500 kW without asking where that power was measured, how it gets to the ground or how the rest of the car performs.
The number might look impressive, but it does not tell you the whole story.
