Talk:OSM2City 1st Worldbuild

From FlightGear wiki
Jump to navigation Jump to search

Asking for clarifications

I'd like to see some clarifications, where the following information is coming from:

There will be more rollouts.

and in relation to the above:

Performance will improve drastically in future rollouts

This unsigned comment was added by Merspieler (Talk | contribs) 17:54, 26 June 2021‎ (UTC)

For example the format of roads is being changed by StuartB for use in WS3 (see Feb newsletter). This will allow less CPU bound rendering, better LoD levels, and lower disk space/bandwidth (roads are stored in vector format). Pylons/cables can get instanced (either as a generic AC file, or as a custom format). OSM vegetation and most buildings are already instanced. That mostly leaves AC buildings which mainly drag down performance in busy city centers, and at least some of them can be done by shaders. Performance is a case of the next biggest bottleneck, which will be unrelated things like random scenery objects (which can be switched off, or instanced)..the scripts merging custom objects for WS3 will help in heavy areas, and various general improvements to CPU boundness with the new rendering should also reduce remaining bottleneck. So this CPU boundness isn't something that FG is stuck with, and the current advice (link) on purchasing hardware is to wait until the LTS with WS3 if at all possible. Vs (talk) 12:26, 27 June 2021 (UTC)
The road changes are in the base scenery, not osm2city.
Same does apply to the rebuild.
For changes like instanced pylon rendering, there would need to be a new osm2city build (afaik, that's not in the ws3 base scenery)
Same applies to making more buildings shader based...
As fas as I can see, there's no such osm2city build happening nor planed.
Merspieler (talk) 10:21, 2 July 2021 (UTC)
"The road changes are in the base scenery, not osm2city." It's based on OSM data, and it will improve performance as far as the end user sees, also replacing the Roads folder (Stuart also released the build scripts). The roads are the next major bottleneck in perfomance and bandwidth, as buildings are in pretty good shape.
"For changes like instanced pylon rendering, there would need to be a new osm2city build" All instanced rendering rendering needs is the pylon positions, nothing new. Pylons only have a minor performance hit, roads are the bigger bottlenck.
"As far as I can see, there's no such osm2city build happening nor planed." There will be changes at some point in future whether using the same infrastructure as the current build or not. For example using the work Volador is doing on building appearance to which Vanosten is reworking OSM2City scripts to integrate. The main thing is expectation management, pointing out the current state so people don't assume FG's approach can't be improved and start complaining/comparing. Better to have lower expectations and be pleasantly surprised. Also to act as a cue pointing out where things can be improved to invite contributions which is important for an opensource project.
"This is an experimental 1st rollout. There will be more rollouts. Expect to download everything again :)." -> expectation management. "Performance will improve drastically in future rollouts. Please don't buy new hardware yet based on early rollouts" -> also CPU bound performance will improve just due to rendering changes. Vs (talk) 03:17, 4 July 2021 (UTC)