As you’re aware, DFT calculations typically systematically under-estimate the electronic band gap. The accuracy is related to the choice of exchange-correlation functional used – such as LDA or PBE (a GGA functional). All band structures currently reported on Materials Project use PBE even though we know there is a systematic error. These calculations can still be useful, because they can accurately describe other features of the band structure and densities of states, even if the magnitude of the gap is too small.
However, this is a fundamental limitation as a result of the physics behind DFT theory and the PBE functional. Therefore it is not a question of a ‘good’ or ‘bad’ INCAR. Even a ‘good’ INCAR will give an under-estimated band gap if you use PBE.
The important question to ask is what you actually want to know. If you already know the experimental band gap, what new information are you hoping to get from a DFT calculation?
There are ways to improve calculation of band gaps from DFT by using a more computationally-expensive exchange-correlation functional. I’ve had good experiences with MBJ, a meta-GGA, which is not too expensive (they take 2-3x longer than PBE) but the calculations can be tricky to converge. For inorganic materials, HSE06 is currently one of the best functionals we can use and often gives very reliable band gaps, but these calculations are very computationally expensive, which is why we don’t run them on all materials on the Materials Project. The appropriate INCAR tags here are GGA, METAGGA and, for HSE06, LHFCALC and HFSCREEN.
Just a warning though, these aren’t easy calculations to start with if you’re new to VASP, but good luck, and I hope this helps!