initial impression

I apoligize, I have not yet taken the time to read all posts perhaps all this has already been covered. I promise to get caught up over the holidays.

Finally got a computer capable of running this. I did my first 5 laps two nights ago.


as in no other riders, no ai riders, nothing. I had the island to myself? as such it was incredibly boring as the laps are so short.

Also even though my head unit (cycleops gps) was picking up everything as normal the on screen speed was not close. I would be spinning it out downhill and my trainer would be showing 40+ mph but because “wattage” was low because of the downhill resistance zwift would be showing 20 mph approx…

addendum, conversely doing slow rpm grinding out the 10% climb had me going much faster than reality because my wattage was high…

when really wattage is also a function of cadence and that is what seems to be missing from the algorithm for me right now

It may be worth opening a support ticket in reguards to the empty world. I don’t think we’ve had any downtime with the AIs recently, and there may be something useful in your log files.

As for the speed, how fast you go on your trainer has little to do with how fast you’re moving in Zwift. The speeds will never match, nor should they. We take the watts your power meter or trainer tells us about and apply standard physics to that data to move your mass through our virtual world, up and down slopes and through wind. For example, somebody on low resistance rollers might actually have their wheel spinning at 50mph, but in Zwift you may be going 5mph up a 14% slope. Watts are all that matter here.

Yea this doesn’t seem right. I think you must be adjusting the load on the trainer when we get to climbing (using a KICKR) somewhat; it does get harder to pedal and the wattage generated does go up when pedalling at same cadence and gearing.

But the whole point of the smart trainers is that you can set what the load should be based on environmental factors INCLUDING speed measured by the trainer.

I do think the logic should be different, as you suggest, for conventional trainers and rollers. But the simulation can be much better when using smart trainers…