Posted this on ML, but got no reply whatsoever.
On next, are LOD ignores values higher than ca. 50 km, so you can no longer
make scenery go all the way to the horizon:
https://imgur.com/a/S1FlcTP
Test:
$ fgfs --aircraft=ufo --altitude=61000.0 --disable-real-weather-fetch --timeofday=noon
next version info:
flightgear b73413d52 2020-11-22 Part 2: Work towards the option of interpolating between climate tiles by combining all climate related ground level parameters in one structure and calculating the sea level tile from the (not yet interpolated) ground level tile. Read the elevelation of the climate tile from the köppen database image which now contains the climate class in the red channel and the ground elevation in the green channel. (HEAD -> next, origin/next, origin/HEAD) [Erik Hofman] simgear 0a379865 2020-11-21 Use ref_ptr for ReaderWriterSPT (HEAD -> next, origin/next, origin/HEAD) [Stuart Buchanan] fgdata 580e1d485 2020-11-22 Keep the Köppen climate class in the red channel and add the ground eleveation in the green channel. (origin/next, origin/HEAD, next) [Erik Hofman]
2020.3.3 version info:
flightgear 782f3cad0 2020-11-21 Sentry: use context to report exception locations (HEAD -> release/2020.3, origin/release/2020.3) [Automatic Release Builder] simgear d95b1c04 2020-11-12 new version: 2020.3.3 (HEAD -> release/2020.3, tag: version/2020.3.3, origin/release/2020.3) [Automatic Release Builder] fgdata 1ca7b078c 2020-11-18 Joystick-config: fix ToggleButton, add AP-DISC (origin/release/2020.3, release/2020.3) [James Turner]
Seeing if Stuart can comment.
I think this was caused by either:
- photoscenery merge;
- CompositeViewer
Sorry, I was wrong.
Now I found the offending commit:
Specifically,
of the file that is now
src/Viewer/CameraGroup.cxx
Does anyone see this? At all?
Weird. I though that change would only affect the WS30 work. I'll take a look.
-Stuart
Reproduction:
$fgfs --aircraft=ufo --airport=LOWI --altitude=60000
and set cloud layers to "Clear" in weather dialog. Result: see photosSetting visibility seems to affect tile loading, but setting Bare LOD not anymore.
Last edit: Anonymous 2021-01-08
Maybe the black hole in the bottom of slybox is what causing this? Because if it wouldn't be there, then the horizon wouldn't be that noticeable
https://sourceforge.net/p/flightgear/codetickets/2507/ is related
This kludge partially reverts the offending commit and makes the tiles stop disappearing right under you (but they are still disappearing on the edge)
Last edit: Anonymous 2021-01-09
And the black hole in the bottom end of skybox indeed seems to be an Intel thing.
i see the same hole as noted in ticket #2507... it is also seen in one of the photos i posted there...
i don't know that LOD is the problem but i do see weirdness like this with some tiles occasionally... i've not been able to pin it down to anything specific...
i use use one of three different LOD sets...
75km horizon:
150km horizon:
370km horizon:
Starting from 30000 ft the "horizon" of displayed tiles starts narrowing until at around 60000 ft you get just one tile like on the screenshot I attached.
But increasing visibility (with basic weather) seems to get the scenery to load. This can not be done with Advanced weather, though.
Hi Michael,
Thanks for the analysis and apologies for the delay in looking at this.
I've had a look, and it seems this LODScale is used to multiply the apparent distance of every LOD node evaluation. So with 1.5 set, from a culling perspective, everything appears 1.5x as far.
That would mean that tiles would get culled as 2/3 the fogging range, which means they'll be unloaded too early.
I'll remove the setting of LODScale.
Thanks,
-Stuart
Hi,
I just tried experimenting with the value, and putting 0.5 to 0.75 instead of 1.5 seems like the safe choice -- then terrain smoothly disappears into the fog without any hard edge.
Seems to have been fixed.