Digital scales are only readable to +/- half of the smallest unit. So 0.1g is 0.05g to 0.15g (assuming conventional rounding - the specs rarely say). It pays to have an extra step of display precision than you need for your project. Note that display precision and stability is not necessarily the same as accuracy or repeatability, or the precision of the weighing mechanism. Sensitivity of open pan balances is not usually a problem for air movement, but it can make the balance slower to respond. Stability/drift is usually a temperature issue. If you keep the balance at room temperature (in the room you weigh in) then only operational heat should be an issue.
It is not as though we are weighing grams to 5 decimal places for general photographic purposes. When I was doing analytical chemistry, the 'bucket' lab only weighed to 0.1g, while the fine analytical weighing was done to 5 decimals (reported at 4), in a temperature controlled room using balances with integral draft shields.
I have a small electronic unit. It is almost too small for photographic chemical volumes, so I usually have to tare off a pan. Think about this when you choose a model so that the display is readable. The specifications on these little things are usually bad - it is not always clear if taring comes out of the total capacity (assume it does). I have some domestic metric weights down to 5g, and some new coins, so I can test the unit from time to time. A good beam balance would be nice, but I don't have the need for it.
