Just to add a twist.. If on your first calculation (probably at the most limiting TOM so far calculated) you do not achieve the 35ft clearance on the net flight path then you have to reduce TOM. This has two effects, it improves the gradient but also it delivers a shorter TODR so the obstacle is further away from the end of the TODR, a point that used to be called reference zero. In olden days this was dealt with by making an arbitrary reduction in TOM, maybe 10,000 kg, replotting the net take-off flight path, observing more than the minimum clearance and then interpolating between the two weights to find an obstacle limited TOM. I don't know how the modern computer programs run the calculation but I have a suspicion that they don't move the end of the TODR as the weight reduces, but keep the obstacle distance fixed from the end of TODA. I would be interested to know if this is the way they calculate or not, if anyone knows?