Beta Feedback (initial two weeks)

First off I would like to say thank you for allowing me to be a part of the ZWIFT Beta program. My overall experience has been great with only minor issues in several sessions and five plus hours of riding over two weeks. I have tried to use various hardware and versions of windows on just about every ride. I have tweaked the graphics properties once or twice as well as outlined in another thread. I only have static resistance traners and so far all of my rides have been on aluminum rollers with magnetic resistance (max is about 500W in top gear @ 120 cadence).

The biggest problem I have had has been with the power sensor not working at the start of several rides. In the initial pairing screen, the device shows up blue as being detected, but when I start riding, power=0 and the rider doesn’t move. My Garmin is paired to the same devices and shows power fine. I have tried using an extension cable on the ANT+ dongle to get it close to the bike, but that has no impact. In every case, closing Zwift completetly and then restarting has fixed the issue. The problem seems to be more likely to happen when an update takes place during the software initialization.

As others have mentioned, the AI could use some tweaks. I seem to get alot of Wheel sucking AI riders all piled up on top of each other, but none seem to want to pull me. They also tend to accellerate in sprint mode when they do decide to pass. Some smoothing on their pace changes would go a long way. It is far easier to draft real riders than trying to keep up with the AI. It would be nice to have a variety of AI styles (all out race, sprinters, recreational) and perhaps have some AI riders that use previous laps or complete sessions that I have done. (kinda like ghost riders) One question is do the AI riders live in the Zwift servers or are they generated on each user’s system?

The visuals have been immersive and everything I hoped they would be. On some of my first rides where I just wanted to get a quick light 10 minutes in, I found myself wanting to keep going and pushed myself harder as the AI passed. Much better than staring at a wall, bar graph, or poorly syncronized floating camera view of other training software I have used and beta tested. Being on static resistance, the speed change noticed on the hills is fairly well tuned. The changing scenery (santa, snowman, blimp, etc…) has kept the “same course” fresh and looking for what will come next time I ride. A nice to have would be for my user to pull to the side of the road when stopping rather than blocking up the middle of the road. As others have said, riders appear to float in the air on downhill sections. One really nitpicky thing noticed is the chain is not moving/animated in view 5. Really looking forward to additional routes, workouts, and scenery in the future.

The current scenerio for FTP/Power/Weight (Watts/Kg) gives the heavier riders an edge, especially on the climbs where they are not pulling their weight. Would be nice when the workouts are added to have an FTP test that would dial in what a user is capable of for leveling the field. That or using historic user ride data to generate a dynamic power profile. Would also be beneficial to have courses and challenges that use both the static profile (like current), a dynamic power profile, and profiles based on real user weight, power, ftp etc. Looking forward to see what directions you go with this.

One other issue is Sensor drops on Windows 10 Preview. I initially was having all sensors drop all over the place and ended up stopping and going over to one of my other systems to do my initial 40K challenge run last week. That was mostly resolved by updating some bad default W10 drivers. I had a good complete 40K session on this same system yeserday. After completing the 40K session and taking a brief rest, all sensors dropped and would not come back when jumping back on the bike. These issues could still just be W10 growing pains, but I thought it would be good to let you know that Windows 10 appears to be a viable, mostly reliable option. I will probably continue playing with this version the most for the time being.

Here is a brief summary of myself and my hardware used to date.

Zwift level: 5 (2 weeks)
Occupation: Computer Geek
Biking level: Recreational Road/gravel B+/A- pace
Weight: 165ish
Age: 47
Gear: Garmin ANT+ dongle, Powertap SL+, Duotrap Speed/cadence, Garmin HRM3, aluminum rollers with static mag resistance, Cycleops Fluid2.
Other training software used: Trainer road, Golden Cheetah, Spivi home edition beta

Hardware/OSes tested:

Lenovo Thinkpad T410
Core i5-M540 @2.53GHz, 8GB ram NVidia NV3100 graphics
Windows 7x64 Pro SP1
Low AVG framerate 5FPS (basic profile)

Intel NUC D54250WYK
Core i5-4250U @2.60GHz, 8GB DDR3-1600 ram, Intel HD5000 graphics
Windows 8.1x64 Pro
AVG Framerate 25FPS (basic profile)

Gigabyte EP45UD3P
Core2 Quad-Q9550 @3.4GHz, 8GB DDR2-1200 ram, AMD HD6850 graphics
Windows 7 Ultimate SP1
AVG Framerate 15-35 FPS (varies alot) (high profile selected automatically)

Intel NUC D34010WYK
Core i3-4010U @1.7GHz, 8GB DDR3-1600 ram, Intel HD4400 graphics
Windows 10x64 Preview
AVG Framerate 25FPS (basic profile)

RIDE ON!
Karl

Karl, wow, thanks for all of that. This is all pretty useful feedback. I agree that the ANT+ communications and searching needs a bit of work, and that is a pretty high priority fix for us. Hopefully we can address that soon.

Regarding weight, you’ll be able to change it this week I hope.

We occasionally test on windows 10 here too and haven’t seen the ANT+ issue, so it might be worth putting your USB ANT+ adapter onto an extension cord and try to ensure line of sight between the dongle and your power meter.

Jon
(ps, you might be the first to mention the chain not moving. Good eye…we were wondering when somebody would notice. :slight_smile: )

FYI, the framerate above was wrong for the NUC D54250WYK. Should have been 35FPS

Intel NUC D54250WYK
Core i5-4250U @2.60GHz, 8GB DDR3-1600 ram, Intel HD5000 graphics
Windows 8.1x64 Pro
AVG Framerate 35FPS (basic profile)