+- (https://forums.vzfit.com)
+-- Forum: VZfit Forum (https://forums.vzfit.com/forumdisplay.php?fid=5)
+--- Forum: General Discussion (https://forums.vzfit.com/forumdisplay.php?fid=6)
+--- Thread: Our next update (/showthread.php?tid=39)
RE: Our next update - emalafeew - 08-01-2019
As an update, we released Explorer last night with some bug fixes and promised improvements.
Hoping to make a Play release today with the same, and also try to reduce the startup time that's been building up.
Beyond that we're still working on multiplayer Explorer mode and Play notifications. We also have an idea to fix a lot of the distortion you see in Explorer, to make it good enough to ride through cities, and more comfortable everywhere.
RE: Our next update - Superlith - 08-01-2019
You guys working hard, the key for success! I appreciate the working tirelessly! Looking forward for the improvements
Tank games already are very smooth and very well done.
You guys working hard, the key for success! I appreciate the working tirelessly! Looking forward for the improvements
RE: Our next update - spyder17 - 08-02-2019
Really look forward to any improvements you can make in Explorer to reduce the artifacts that come from Street View.
RE: Our next update - emalafeew - 08-23-2019
The Play release is out with those fixes and improvements. Took longer than hoped and also got some vacation, but can now focus on finishing Explorer multiplayer and Cali Rally for Play next. Artifact reduction and web editing of rides are in the queue!
RE: Our next update - mrphilipjoel - 08-24-2019
(08-01-2019, 03:48 PM)emalafeew Wrote: As an update, we released Explorer last night with some bug fixes and promised improvements.
Hoping to make a Play release today with the same, and also try to reduce the startup time that's been building up.
Beyond that we're still working on multiplayer Explorer mode and Play notifications. We also have an idea to fix a lot of the distortion you see in Explorer, to make it good enough to ride through cities, and more comfortable everywhere.
I was curious, how do you describe that distortion effect you do? When doing the colorado path, I thought you added your own 3D models or something along the road, to make the Google Photos changing less jarring. i was amazed at how well it worked (tried a previous app called VR Cardio that used Google Earth, and it was very jarring). Then, I tried a path in italy, in the woods, and I noticed how all the tress and stuff where part of that distorted texture that twisted when you rode. It does a good job of making it feel like the scenary is moving with me, and not just a slideshow of photos.
So, what do you call that?
RE: Our next update - emalafeew - 08-26-2019
Thanks for complimenting, we call it scene morphing! At each white dot along the road we have a 360 degree image that is projected onto a mesh corresponding to estimated distances from the dot to the road and objects in that scene. As you ride from dot to dot, you are riding over that mesh until you get closer to the next dot, when we blend into its mesh. The success of this morphing technique depends on the accuracy of those estimated distances, the number of "hidden surfaces" revealed between dots, and the accuracy of the location and rotation of each dot. Those estimated distances are currently best along routes with gentle hills or solid walls, and less good now in cities but we have ideas to improve. The hidden surfaces appear as "image stretching" as you move from dot to dot, and are a function of objects you are passing at different distances. The location & rotation at each dot is simply better along some routes than others; we haven't found a pattern but also have ideas to improve. We're currently adding multiplayer to Explorer but will be trying these improvements right after!
RE: Our next update - emalafeew - 10-03-2019
Just an update for everyone:
We released Explorer with multiplayer two weeks ago, and are currently working on the next Play build, which will include Cali Rally (a cycling level from our PC version), better framerate on the Go, and some requested improvements that we hope to be released in a week or two.
We're also doing R&D on distortion improvements for Explorer, with good initial results, but will take another month or two to fully get into the game and our servers.
A side note about Go framerate: something seems to have lowered it in both Play and Explorer in the past month, that doesn't seem related to the builds themselves because we've reproduced it with much older builds that were confirmed to run at 60 fps. It seems that both updating to latest Unity and Oculus plugin restores that framerate. We updated Explorer on Monday like this, and will have it in upcoming Play build.
RE: Our next update - emalafeew - 10-17-2019
The Play build update is out! Now we're focused on an Explorer build that addresses white dot loading issues. The distortion improvements are coming along but we'll want to get this build out before those are ready.
RE: Our next update - mjmangerhelena - 10-21-2019
By riding preset rotes in the same direction as the google car is going, distortions are minimal. Why not default each route to that direction?
RE: Our next update - emalafeew - 10-21-2019
Routes can be composed of multiple google car recordings, going different directions and on different dates, in order to reconstruct our paths going from A to B. So even if you can start out going the same direction as traffic, there's no guarantee it won't turn against you later, unless you follow the same path as the google car that day.
We agree that cars in the road are a very noticeable distortion. Finding rides to suggest with no traffic has been our ideal, but we could do a better job starting you out with traffic. Note we let you ride any route in reverse direction so would have a hard time fixing both ways!
The other kinds of distortions we're trying to improve are a) how well dots line up with the road, b) how closely image transitions match up, and c) how accurate the surface depths are in each image. We can theoretically fix these up manually per image, but need to find automated ways to make them feasible.