The real world scenario you are describing as far as I understand it is a trader writing an OTC call option. His price has to reflect the current and expected future price/movement/volatility of the underlying, namely the commodity future.
Lets first assume that the likely buyer of that option is interested in a a) ATM/ITM option or at least not too OTM, otherwise the premium will be likely too low for the writer to even bother going ahead with the transaction, and b) in an expiry on or near enough an underlying future's maturity (that should not be a problem as most commodities future markets offer maturities for days/weeks or at least monthly ones if they are traded 3 years into the future).
Since there is no daily implied volatility published by a body like an exchange (e.g. LME, SGX, CBOE, etc) one could use as a substitute the historical (aka realised) volatility plus a (il)liquidity reserve. There are several ways to calculate historical volatility using the underlying future prices (Close on Close, OHLC, look up Parkinson, Garman-Klass, volatility cones, etc).
Finally, addressing b) above again: what if the option's maturity can't 'fit' against any liquid traded future (i.e. you dont have an exact/near enough underlying price or time to maturity) ? Not a scenario I have read about or experienced but my guess would be that the underlying future price would be interpolated linearly from known points of the forward curve. Not sure if we are talking contango/backwardation (or even if we are talking perishable or storable commodities here...) but I guess the curve shape (spot out to 3 years) should allow linear interpolation somehow.
Please note that I am also assuming from the question that we are talking 'vanilla' call option, therefore any 'standard' BS or Black76 or similar pricing method could apply assuming you have the necessary inputs described above.