Grinding requirements and sampling advice:
The Roast Vision requires a Turkish Coffee Grind (possibly coarser to a Very Fine Espresso Grind).
The small particle size is required as the device was designed and calibrated around this particle size, as well as the benefit in reducing sample variation due to large chunks that would be present in larger grind settings (like for a pour over or cupping grind).
Shown here is the effect of grind size on the Roast Vision measurement, where the coarser the grind, the darker the measurement appears for the same coffee bean:
Turkish Grind = 13 
Moka Grind = 11 
Pour Over Grind = 10 
French Press Grind = 9
However, if you are grinding extremely fine, and are experiencing significant clumping, this can also make the readings less accurate, so we recommend to slowly adjust the grind setting coarser (to a very fine espresso like grind) to minimize clumping.
Importantly, when taking multiple measurements, we recommend that you discard each sample between measurements, and to use a new spoonful of ground coffee to ensure that you are sampling a larger subset of the roasted beans than otherwise using the same sample multiple times.
Loading the coffee into the sample cup:
When loading coffee into the device, there may be a premature measurement displayed on the screen, when the coffee is detected, but has not yet been leveled by tapping the device down to settle the grounds.
These measurements that are displayed before the coffee sample is fully prepared are not valid and should be ignored.
In the above video, the measurement would be recorded as a 13, as the initial 8’s that appear on the screen are NOT valid, as the sample hasn’t been properly loaded into the sample cup yet.
Tapping the device for a valid measurement:
In order to settle the grounds for a valid measurement, the device must be tapped in a vertical motion, against a solid surface, with substantial, yet light force.
Demonstrated in the following video, this can be achieved by holding the device and tapping it down on a counter manually, or by lifting and dropping the device against the counter (1/2 inch max drop height) to settle the grounds with a consistent force.
In the video, the only valid measurements are the 13’s, as the other measurements displayed on the screen were from before the sample was properly loaded and the device had been tapped to settle the grounds. Even the 12, although close to the correct reading, is not valid as the device and sample had yet to be fully tapped to level before it registered at the correct 13 for this sample.