I upgraded the sim to VR3 last night and found, like several others, that I seemed to get generally better performance. I can run in HDR now with no significant hit compared to non-HDR, and it does seem slightly smoother.
Today, though, I watched the video below, made all of the changes suggested therein, and then had a truly remarkable 2:38 VR flight (yes, two hours, thirty-eight minutes). The sim was buttery smooth nearly the whole time, and this was WITH me streaming it and having the YouTube chat window in the cockpit via Oculus Dash. The flight started at KSPG Albert Whitted, which usually hits my frames pretty hard, but I was running at 45fps throughout. Across the flight I had no “grey flash” moments, judders, or image freezes, save one (when weather re-loaded). Toward the end of the flight I could tell the sim was probably running at 22.5fps for a time, but given that I’d been running it (and chewing through VRAM) for nearly three hours I wasn’t surprised. And after a bit, it improved. Truly a great flight, and I highly suggest taking the 40 minutes to watch his vid.
And for those who want to see the flight, here’s the final 2:00 (it seems YouTube only keeps two-hours worth of streaming video). The flight was a dress-rehearsal for the cross-country I’m set to fly tomorrow real-world. One of the fun things in this flight: about six miles out from my first landing a YouTube watcher suggested via the chat that the airport was closed via NOTAM and that I had to divert. I did, and it was a great example of what to expect from a real-world examiner.
Yesterday I flew my first scheduled live stream on YouTube, a PilotEdge flight from Grand Junction to Aspen. This is actually a continuation of my point-to-point flights, but I’ve switched to the JustFlight Turbo Arrow as the Saratoga isn’t a 3D cockpit suitable to VR (that Arrow is fantastic by the way – a really great X-Plane GA bird). The stream worked relatively well, although I’m still suffering from some upload speed issues. The flight goes relatively well. A weather update gives us some sudden IMC conditions, but we quickly get below the cloud deck and change our planned flight path to stay in the valleys. I also blow the approach into Aspen but the controller is good enough to get me back into the pattern. And last but not least, I crunch the landing hard enough to set off the ELT transmitter – I was using the Rift controllers and thought I had my hand on the virtual throttle in the flare but did not. It’s pretty funny to watch. Also, thinking about it now, I kept having to add rudder on the approach thinking it was a cross wind. It wasn’t. I’d used the Arrow’s rudder trim in flight and never trimmed it back. This is why we practice!
I have an older webcam and I decided to try hooking it up to the sim PC today to try a live stream on YouTube. It was a rough start with horrible PC performance, and as you’ll see in the video SPAD.neXt seemed to be the problem. It has never been a resource hog before, so I’m not sure what the issue was – perhaps its input / output conflicts with the video streaming software? But once I terminated it things worked much better, and I was able to stream using OBS as the video compiler and still have frame rates pretty close to where they would normally be.
I’d never done this before, and didn’t know quite what to expect. But it was actually very cool: we had 20-30 people online, and they represented Indonesia, Brazil, Ireland, England, Italy, Russia, France, and the States among other places. One viewer even jumped into his sim, loaded in at KSBP on PilotEdge, and made the flight with me. You can see him land near the end. Sadly the audio stream wasn’t capturing the PilotEdge transmissions, so you don’t hear him or ATC. I’ll get that figured out as soon as I can.
I had the stream and live chat running on my iPad, which worked reasonably well although it’s clear that one should not fly and participate in YouTube chats at the same time. Here’s the vid of the stream. You can clearly see and hear how the quality changes once I fiddle with things and determine SPAD is an issue. Thanks for watching.