Better Virtual Gear Modeling
Love virtual gear and had some discussion on this before and it generally works. However, the realism of virtual gear doesn't correlate to real gear shifting, for example in real gear I can do one shift (which might jump one or two teeth in a cog), but in virtual gear I might need to +/-3 (or more) to get similar resistance. Hence for pacing or TT, current virtual gear is ok but in a sprint, I still need to resort to physical gears so I can do a gear dump.
Can we do a better modeling of the gears so each +/- 1 virtual gear feels closer to a physical gear? In theory it is a lot of just data collection of someone riding around the velodrome with no wind and recording the power at a set speed and cadence at a specific physical gear then find the equivalent virtual gear that generate the same power at the same speed and cadence.
Jarvis
shared this idea
-
Andy
commented
I agree, it can take 10-20 seconds to change 6 virtual gears, this is far too slow and makes racing practically impossible,. In a race I was in yesterday I went from 90 rpm to 40 rpm when we hit the hill and it took over 20 seconds to get to the lowest possible virtual gears, by this time I was dropped
-
Jarvis
commented
Once data collection of physical gear to virtual gear is done, it opens up the opportunity for "auto calibration" regardless of which physical gear you are in. It can create the model of what flat road should be and therefore when someone is spinning at the start before an event or right at the start it can compare that to the flat model generated and change the multiplier such that the rider resistance, and power matches the model. This auto calibration and take it effect each time there is a "flat" section