|
Yeah
#15287 posted by metlslime on 2015/08/22 22:52:27
map format already supports larger map sizes, it's the network protocol that is the limiting factor.
I didn't add this feature to protocol 666 because I couldn't think of a good way that wouldn't make packet sizes bigger for normal-size maps. This means you'd get a packet overflow on some older maps that wouldn't overflow on protocol 15.
Fifth
#15288 posted by SleepwalkR on 2015/08/22 23:58:35
The empty map will contain an initial room at some point in the future. If you want to testdrive the new / old create brush tool, I can upload a new build that contains it.
For Those Who Care About Software Render
#15289 posted by adib on 2015/08/23 17:03:35
Mankrip, da man behind Makaku and Retroquad engines, runned my jam 6 level and reported two bugs that pop in software render.
The mipmaps of my fixed rock_brown texture were wrong. I used Wally to edit the wads. Apparently the default settings for Remip Deluxe doesn't work well with Quake software render. I had to go to View / Options / Remip and click the "Use Quake Lookalike Settings" button.
Also, liquid textures bigger than 64x64 get distorted by software render. I was using *lava4a. For full compatibility, I'm changing to *lava1.
DLC will have both fixes.
Mce.bsp By Negke
#15290 posted by ptoing on 2015/08/23 20:56:14
How are those custom ammo boxes done, given that this is just a bsp file?
#15291 posted by - on 2015/08/24 07:23:49
Do Fitz/QS engines have a command to turn off .lit support? Something to easily see what a map looks like without colored lighting (even if you have to reload the level)
#15292 posted by ptoing on 2015/08/24 15:27:42
Hm, just went through all the commands and such in QS, not Fitz though, but QS does not seem to have this. Would be a cool option. Something like r_uselit.
The fastest way now is to rename the lit file to something like lit~ and then restart the map in QS. A bit tedious, but it works. Also damn, some maps look horrible without the lit file, like Skacky's Jam5 map. Super blown out. I think it has to do with the thing that ericw mentioned a while back somewhere about using dark lightcolours.
#15293 posted by Spirit on 2015/08/24 15:41:00
Unless you are using BSP2, reQuiem has gl_loadlitfiles 0/1 (also available in the menus).
#15294 posted by metlslime on 2015/08/24 20:23:04
Also damn, some maps look horrible without the lit file, like Skacky's Jam5 map. Super blown out. I think it has to do with the thing that ericw mentioned a while back somewhere about using dark lightcolours.
That's suprising, i would have assumed that the non-colored version of the lighting would just desaturate the color while keeping the apparent brightness.
#15295 posted by ericw on 2015/08/24 20:34:26
It could have to do with how .lit support was added to tyrutils, the bsp lighting didn't use the _color keys at all. I changed it, for the next release of my fork, to average the RGB components of the ilt sample to get the greyscale value for the bsp.
Ericw
#15296 posted by ptoing on 2015/08/24 20:39:27
That's a very good solution for sure. And great work on the compile tools in general.
RGB Is Terrible For Almost Everything.
#15297 posted by Spirit on 2015/08/24 21:03:37
Careful with averaging RGB, that would give you some weird results. You rather want to calculate the (perceived) lightness of the color. I would need to go through my docs but if you can't find a good solution give me a nudge.
I wonder if the weird green tint on explosions/dynamic lights when playing games with (reddish?) lit files is also due to this. Noticed it a lot in the jam6 maps with QS.
Argh
#15298 posted by ericw on 2015/08/24 21:22:36
Actually it looks like averaging won't work..
The constraint is the LightNormalize function that MarkV is using (Baker mentioned that he took it from some QW engines). It scales the lit data based on the bsp file lighting, supposedly to prevent cheating in MP by using a fullbright lit file.
I looked at it again and, to prevent MarkV from scaling the lit data, the bsp needs to contain max(r,g,b). This kind of sucks because it forces your bsp to have overblown greyscale lighting, or else have potentially distorted colors in MarkV.
#15299 posted by metlslime on 2015/08/24 21:24:14
1. as Spirit says, don't just do a staight average of RGB. Try instead using one of the formulas from here: http://stackoverflow.com/questions/596216/formula-to-determine-brightness-of-rgb-color
2. Spirit, regarding the green tint on explosions, this happens because the explosion is blowing out the lightmap to white. At the edges of the explosion, only certain color channels are blown out, so you get different colored rings around the explosion based on the color of the explosion and the color of the pre-existing lightmap. To preserve the color better, probably an engine would have to calculate the correct color first for each lightmap point, then clamp all 3 channels to different values to preserve it.
Also Spirit
#15300 posted by ericw on 2015/08/24 21:29:06
I wonder if the weird green tint on explosions/dynamic lights when playing games with (reddish?) lit files is also due to this. Noticed it a lot in the jam6 maps with QS.
I think I see this with the rocks in jam6_daz, pretty sure it is completely normal (the rocks have a bit of green in them, the explosions are white so the orange tinted lighting briefly changes to white)
Yeah.
#15301 posted by ptoing on 2015/08/24 21:34:53
I think it is more of an optical illusion as well. Partly the rock colours, partly the amount of red on screen getting your retinas to get cyan/greenish afterimages a bit.
If you look at this shot with it looks a bit greenish on the rim of the light sphere.
https://dl.dropboxusercontent.com/u/15588722/post/func_msg/lighting_1.png
When looking at it with lightmap only it is a lot more yellow looking, and there really are no greens as such, but some dirty yellows.
https://dl.dropboxusercontent.com/u/15588722/post/func_msg/lighting_2.png
How Cool Is That
#15302 posted by Spirit on 2015/08/24 22:03:48
Contrast/complement effect from red tinted faces? Someone should test if we see red if the lights are rather green.
#15303 posted by ptoing on 2015/08/24 23:17:27
Just made a quick test room with greenish cyan light.
Plop in there and shoot some rockets at the walls. Esp the blue wall looks weirdly purple-ish. So it seems like it does work at least a bit.
https://dl.dropboxusercontent.com/u/15588722/post/func_msg/complementtest.bsp
Missing Lit File Please
#15304 posted by Spirit on 2015/08/25 16:35:51
Derp.
#15305 posted by ptoing on 2015/08/25 16:57:13
Beyond 4096
#15306 posted by necros on 2015/08/28 17:33:45
which engines support both bsp2 and > 4096 boundary protocol?
i checked the usual suspects: DP and RMQE but neither supports BSP2.
Prefer a stock-alike engine if possible.
#15307 posted by Spirit on 2015/08/28 18:12:16
Fteqw should do, you can make it very stock like too. Quakeforge might do it too, not sure though. DP does support bsp2 though, did you use an old version.
#15308 posted by necros on 2015/08/28 19:14:29
thanks spirit, what is quakeforge?? will need to google that shizzle.
i thought DP did bsp2?? but it said unsupported type. although, now I look at the build, it says it's from june 2011. so yeah, probably an old one.
grabbed the 20140513 build, loads the map, but geometry flickers like mad: http://shoresofnis.com/temp/jam620150828131103-00.jpg :(
#15309 posted by Joel B on 2015/08/28 19:34:36
BSP2 support (allegedly): DarkPlaces, QuakeSpasm, Fitzquake Mark V, qbism Super8, latest automated build of FTE, TyrQuake from 2013 or later, and QuakeForge.
DP Flickering Bug
#15310 posted by ericw on 2015/08/28 19:35:36
try "r_useportalculling 0"
#15311 posted by Spirit on 2015/08/28 20:03:03
ouch, same here. It also spammed more than 500 times "Mod_Q1BSP_RecursiveNodePortals: WARNING: new portal was clipped away".
Quakeforge is pretty awesome. Very non-standard though in terms of getting it properly configured (at least on Linux). It deserves a lot more attention.
|
|
You must be logged in to post in this thread.
|
Website copyright © 2002-2024 John Fitzgibbons. All posts are copyright their respective authors.
|
|