Unvanquished Development activity

Events from:

Unvanquished
Unvanquished
Wednesday, 08 April 2020
@illwieckz
illwieckz commented on issue Unvanquished/Unvanquished#1165 Apr 8, 2020

I experienced it, this is a release blocking bug. We have to fix it or revert the change.

@illwieckz
illwieckz commented on issue DaemonEngine/Daemon#169 Apr 8, 2020

Do someone has something against using an higher arbitrary value as workaround? Like 256. At least we know the 64 limit was set for another code th…

@illwieckz
illwieckz commented on issue DaemonEngine/Daemon#76 Apr 8, 2020

This is confirmed that jota issue was the collision of two bugs, lightmaps spilling out (probably due to lossy webp compression) that would not be …

@illwieckz
illwieckz closed an issue in DaemonEngine/Daemon Apr 8, 2020
pure white line artifact on edges #76

This is a weird behavior we can see on some different use cases: lines because two edges are rendered very near even if they are on different dept…

7 comments
@illwieckz
illwieckz merged a pull request in DaemonEngine/Daemon Apr 8, 2020
glsl/lightMapping: increase lower bound in computation to restore original light color, fix #299 #306

glsl/lightMapping: increase lower bound in computation to restore original light color, fix #299

+18 -1
2 comments
@illwieckz
illwieckz closed an issue in DaemonEngine/Daemon Apr 8, 2020
deluxemap overflow #299

Reported there before: UnvanquishedAssets/map-station15_src.dpkdir#3 I'm now sure it's an engine issue, and I now know the line causing this: ligh…

7 comments
@illwieckz
illwieckz closed an issue in DaemonEngine/Daemon Apr 8, 2020
deluxemap overflow #299

Reported there before: UnvanquishedAssets/map-station15_src.dpkdir#3 I'm now sure it's an engine issue, and I now know the line causing this: ligh…

7 comments
@illwieckz
illwieckz pushed to DaemonEngine/Daemon Apr 8, 2020
1 commit to master
  • @illwieckz a4a2e7e
    glsl/lightMapping: increase lower bound in computation to restore ori…
@illwieckz
illwieckz commented on pull request DaemonEngine/Daemon#306 Apr 8, 2020

Reminding it was looking like this: Without the line in that glsl file it looks like this (feature being disabled): With the line being edited by…

Daemon Engine
Daemon Engine
Tuesday, 07 April 2020
@illwieckz
illwieckz commented on pull request DaemonEngine/Daemon#307 Apr 7, 2020

This implementation or another one is required for Unvanquished 0.52 because Vega normal maps have to be fixed in a way we don't have to edit mater…

@illwieckz
illwieckz opened a pull request in DaemonEngine/Daemon Apr 7, 2020
renderer/tr_shader: introduce engine-agnostic normalFormat material stage keyword #307

Introduce engine-agnostic normalFormat material stage keyword. There is already a normalScale keyword which allows to revert channel by using a neg…

+189 -51
@Nickrrau

An FPS/RTS hybrid game powered by the Daemon engine (a combination of ioq3 and XreaL)

C++ 516 Updated Apr 7

@illwieckz
illwieckz opened a pull request in DaemonEngine/Daemon Apr 7, 2020
glsl/lightMapping: increase lower bound in computation to restore original light color, fix #299 #306

glsl/lightMapping: increase lower bound in computation to restore original light color, fix #299

+18 -1
Daemon Engine
Daemon Engine
Monday, 06 April 2020
@dimhotepus
dimhotepus opened an issue in DaemonEngine/Daemon Apr 6, 2020
Unneeded AES256 decryption while performing AES256 encryption #305

We decrypt AES256 encrypted message and discard result in Aes256Encrypt. Looks like it is the artifact of the developer who was testing encryption,…

Thomas Debesse (8201f9b8) at 06 Apr 00:19

easy-builder: make it compatible with Ubuntu 18.04/CMake 3.10

Unvanquished
Unvanquished
Sunday, 05 April 2020
@slipher
slipher commented on issue Unvanquished/Unvanquished#1168 Apr 5, 2020

We will make special efforts to not redo the mistake next time. The release validation script I created recently checks that the md5sums are prese…

@illwieckz
illwieckz closed an issue in Unvanquished/Unvanquished Apr 5, 2020
Can't download 0.51.1 assets via download-paks - md5sums is empty #1168

Try to download assets for the latest version (0.51.1) from http://cdn.unvanquished.net/unvanquished_0.51.1/pkg/ via ./download-paks: Empty file: …

1 comment
@illwieckz
illwieckz commented on issue Unvanquished/Unvanquished#1168 Apr 5, 2020

Hi, its is known, see #1100. 😃 Unfortunately we cannot modify the md5sums after release because it is part of the torrent and this server is also a…

@dimhotepus
dimhotepus opened an issue in Unvanquished/Unvanquished Apr 5, 2020
Can't download 0.51.1 assets via download-paks - md5sums is empty #1168

Try to download assets for the latest version (0.51.1) from http://cdn.unvanquished.net/unvanquished_0.51.1/pkg/ via ./download-paks: Empty file: …

@slipher
slipher opened an issue in DaemonEngine/Daemon Apr 5, 2020
Condump filesystem issue with OneDrive #304

When OneDrive storage is enabled for the My Documents folder on Windows 10, the condump command fails with the message Failed to open 'condump/' fo…

@slipher
slipher pushed to Unvanquished/Unvanquished Apr 5, 2020
1 commit to master
  • @slipher b2e7c51
    Port "fix large hex letters parsing" from Daemon

Thomas Debesse (366e4cb1) at 05 Apr 03:59

readme: update

... and 7 more commits

NetRadiant
NetRadiant
Saturday, 04 April 2020

Thomas Debesse (800aafc8) at 04 Apr 09:12

docs: set index.html to redirect

Thomas Debesse (932d21b4) at 04 Apr 08:59

contrib: update urls and names

... and 22 more commits

@illwieckz
illwieckz pushed to DaemonEngine/daemonmap Apr 4, 2020
2 commits to master
  • @illwieckz ed295c5
    drop RADIANT_VERSION, compute git version from current dir (for when …
  • @illwieckz e49c24d
    rename Q3MAP_MOTD as DAEMONMAP_MOTD
NetRadiant
NetRadiant
Wednesday, 01 April 2020

Thomas Debesse (7d2c6f1a) at 01 Apr 00:04

bundle: make daemonmap buildable and bundlable like other tools as ...

This is only made to make easy to build NetRadiant packages for Windows.

Windows library management has a design issue: instead of patching binaries with rpath to set an extra library directory to look for, we have to both patch binaries with assembly information using .rc files and distributes an .xml sidecar file that lists every library to look for in that library directory.

So, making a bundle from two repositories (i.e. merging two packages) requires to merge XML files.

So I had two alternatives:

  1. write an helper to merge XML files
  2. add a submodule and extend CMakeLists.txt

I obviously chosen the second option, I don't want to spend time writing software to workaround operating system design flaw. Workaround for operating-systems better live in toolchain.

Daemon Engine
Daemon Engine
Tuesday, 31 March 2020
@illwieckz
illwieckz pushed to DaemonEngine/daemonmap Mar 31, 2020
1 commit to master
  • @illwieckz 9e64c06
    easy-builder: fix submodule checking for daemonmap
@illwieckz
illwieckz pushed to DaemonEngine/daemonmap Mar 31, 2020
2 commits to master

Thomas Debesse (b8fc2520) at 31 Mar 23:39

easy-builder: make possible to check for more than one submodule

@illwieckz
illwieckz pushed to DaemonEngine/daemonmap Mar 31, 2020
2 commits to master
  • @illwieckz 2a6b801
    warning: avoid comparison of integer expressions of different signedness
  • @illwieckz 0f069f7
    warning: make sure to not write a terminating nul past the end of the…
  • 4 more commits »

Thomas Debesse (36e32cba) at 31 Mar 16:50

easy-builder: strip release binaries, support older cmake

@themuffinator

The Dæmon game engine. With some bits of ioq3 and XreaL.

C++ 110 2 issues need help Updated Apr 8

@illwieckz
illwieckz pushed to Unvanquished/unvanquished-mapeditor-support Mar 31, 2020
2 commits to master
  • @illwieckz 1a7fa81
    mkeditorpacks/buildmenu: fix tool detection
  • @illwieckz 9a97cf5
    mkeditorpacks/buildmenu: read q3map2 game keyword from game.yaml
@illwieckz
illwieckz commented on issue DaemonEngine/Daemon#299 Mar 31, 2020

Vega is ok with .2 but station15 requires at least .3. With .3, it's possible recognize a slight artifact on chasm, but that's hard to notice witho…

@illwieckz
illwieckz commented on issue DaemonEngine/Daemon#299 Mar 31, 2020

Using .1 is not enough: This looks better, but it's not enough.

Thomas Debesse (350fed8a) at 31 Mar 00:45

readme: use cmake -S instead of undocumented -H

Thomas Debesse (00b3b32a) at 31 Mar 00:43

easy-builder: improvements

Daemon Engine
Daemon Engine
Monday, 30 March 2020
@slipher
slipher commented on issue DaemonEngine/Daemon#299 Mar 30, 2020

The concept of a light direction for something other than a point source is pretty mathematically bogus (I mean, even more so than most rendering m…

@gimhael
gimhael commented on issue DaemonEngine/Daemon#299 Mar 30, 2020

The reason why this line exists, is because the lightmaps generated by q3map2 don't store the raw light value, but they store light premultiplied w…

@slipher
slipher commented on pull request DaemonEngine/Daemon#300 Mar 30, 2020

Maybe we should start indenting things in #ifdef, since it's too hard to see this stuff.

@illwieckz
illwieckz commented on pull request DaemonEngine/Daemon#300 Mar 30, 2020

See else on line 98, this part of the code never run with PBR.

@slipher
slipher commented on pull request DaemonEngine/Daemon#300 Mar 30, 2020

Why is !defined(USE_PHYSICAL_MAPPING) removed?

@illwieckz
illwieckz commented on pull request DaemonEngine/Daemon#303 Mar 30, 2020

Thank you @dimhotepus! That's very precious. 👍 I remember Daemon#24, Unvanquished#1013 and Unvanquished#1014, 😉. Good job! That's nice to see you a…

@slipher
slipher pushed to DaemonEngine/Daemon Mar 30, 2020
2 commits to master
@slipher
slipher merged a pull request in DaemonEngine/Daemon Mar 30, 2020
Static analyzer (PVS Studio) fixes #303

Some fixes from PVS Studio static analyzer. Product page: https://www.viva64.com/en/pvs-studio/ How to use for free: https://www.viva64.com/en/b/0600/

+46 -36
1 comment

Also, this code is meant to be deleted one day: the long term goal is to build navigation meshes in game. Since I don't know when that would be done on game-side, I need an easy way to ship this tool on short term.

Note that this is completely optional at build time: if the submodule is not there, the tool is not built.

The repository is still entirely buildable without cloning any submodule (that just disables .crn file support (see !104 (merged)) and does not compile the daemonmap tool).

This is only made to make easy to build NetRadiant packages for Windows.

Windows library management has a design issue: instead of patching binaries with rpath to set an extra library directory to look for, we have to both patch binaries with assembly information using .rc files and distributes an .xml sidecar file that lists every library to look for in that library directory.

So, making a bundle from two repositories (i.e. merging two packages) requires to merge XML files.

So I had two alternatives:

  1. write an helper to merge XML files
  2. add a submodule and extend CMakeLists.txt

I obviously chosen the second option, I don't want to spend time writing software to workaround operating system design flaw. Workaround for operating-systems better live in toolchain.

@illwieckz
illwieckz pushed to DaemonEngine/daemonmap Mar 30, 2020
1 commit to master
  • @illwieckz f64be32
    make daemonmap buildable as netradiant submodule
Daemon Engine
Daemon Engine
Sunday, 29 March 2020
@illwieckz
illwieckz pushed to DaemonEngine/daemonmap Mar 29, 2020
1 commit to master
  • @illwieckz fe8bf86
    update recast submodule url
@illwieckz
illwieckz pushed to DaemonEngine/daemonmap Mar 29, 2020
2 commits to master

Thomas Debesse (0bd48db4) at 29 Mar 21:27

bikeshedding: minor readme fix

@illwieckz
illwieckz pushed to DaemonEngine/daemonmap Mar 29, 2020
2 commits to master
@illwieckz
illwieckz commented on pull request DaemonEngine/Daemon#300 Mar 29, 2020

Yes, in my agenda the next factorisation step is to reduce the number of places calling computeLight. Basically the whole file containing this code…

@slipher
slipher commented on pull request Unvanquished/Unvanquished#1158 Mar 29, 2020

It might be a bit visually cluttered when fighting. I'd have to try it and see.

@slipher
slipher commented on pull request DaemonEngine/Daemon#300 Mar 29, 2020

computeLight is called in several places, but this code only exists here. Is that an intentional change?

@slipher
slipher commented on issue DaemonEngine/Daemon#299 Mar 29, 2020

I don't understand why this line is needed, but if there are artifacts, maybe they could be caused by bad values in the normalmap, for example if a…

@illwieckz
illwieckz pushed to DaemonEngine/daemonmap Mar 29, 2020
2 commits to master

Thomas Debesse (27e8ce20) at 29 Mar 18:58

q3map2/smokinguns: fix .tex writing, <3 @vloup

@dimhotepus
dimhotepus commented on pull request DaemonEngine/Daemon#303 Mar 29, 2020

Done. Can you, please, review fixes and added commits?

@dimhotepus
dimhotepus commented on pull request DaemonEngine/Daemon#303 Mar 29, 2020

Restored to original, thanks!

@slipher
slipher commented on pull request DaemonEngine/Daemon#303 Mar 29, 2020

Thanks for the fixes. I think I ran into that first [overrun] one recently while working on WASM support.

@slipher
slipher commented on pull request DaemonEngine/Daemon#303 Mar 29, 2020

I find this change makes the intent less clear. How about adding a TYPE_BEGIN alias for the first enumerator?

@slipher
slipher commented on pull request DaemonEngine/Daemon#303 Mar 29, 2020

So there's this wacky multi-mode error handling thing in these functions... If &err equals a magic value, then errors are reported by throwing exce…

@slipher
slipher commented on pull request DaemonEngine/Daemon#303 Mar 29, 2020

This use a formatting library (tinyformat) based on C++ templates, not va_args. So there is no problem here - all integer format specifiers are tre…

@dimhotepus
dimhotepus opened a pull request in DaemonEngine/Daemon Mar 29, 2020
Static analyzer (PVS Studio) fixes #303

Some fixes from PVS Studio static analyzer. Product page: https://www.viva64.com/en/pvs-studio/ How to use for free: https://www.viva64.com/en/b/0600/

+24 -16
@themuffinator

An FPS/RTS hybrid game powered by the Daemon engine (a combination of ioq3 and XreaL)

C++ 516 Updated Apr 7

Thomas Debesse (dd26dc48) at 29 Mar 01:05

bikeshedding: reduce diff noise with Garux tree

Yes I'm running renderer2, I reproduce the bug with legacy renderer.

Thomas Debesse (5b33d0e7) at 29 Mar 00:30

q3map2: pad with zero, not with null pointers, ref #160

good catch. thx

NetRadiant
NetRadiant
Saturday, 28 March 2020

Note that I then added a new commit (96a0d4ae) that replaces .[ExecutableType] by [ExecutableExt] before variable parsing. The later variable is using a per-OS define set in libs/globaldefs.h like other per-system defines. This makes variable parsing less hacky and less prone to errors.

Also the code previously used to set the ExecutableType is now deleted from Cmake and Makefile. If needed (I highly doubt about it), ExecutableType is still available as the executable extension minus the leading dot, if executable extension exists.

Using [ExecutableExt] in gamepacks is discouraged by the way, since it would make those gamepacks incompatible with other netradiant fork living out there, and with older builds.

I have tried tGridlightSun_net.bsp in xonotic, and it worked, as expected.
So lightgrid data is there, but not in form, understandable by some engines (vanilla q3, older ioq3, q3e github.com/ec-/Quake3e ).

Can you share the compiled map to check on my end? Screenshots look like renderer2, which may not or utilize lightgrid differently.

Note: I built map with both Linux and Windows build and get the same result as my previous screenshots whatever the OS.

I just noticed your example map, so I built it and I got this:

I don't see anything wrong, but maybe I'm missing something?

The weapon may look brighter (especially on the penultimate, but that's also because the light is not behind me, that looks better when I go from brighter zone to dark zone), but at least there is a going brighter and brighter as I walk into the bright zone.

This is well seen there:

Do you have ways to reproduce this bug?

On my end I compiled q3dm17sample map with q3 build profile and got this (using q3map2 master from this repository), then I loaded the map in ioquake3:

So, the lightgrid seems to work on my end.

Thomas Debesse (96a0d4ae) at 28 Mar 02:09

radiant/buildmenu: rename ".[ExecutableType]" as "[ExecutableExt]"

Solved in my netradiant-1.5.0-20200328git build

A million thanks, folks ;)

Thomas Debesse (ae00fba5) at 28 Mar 00:53

radiant/buildmenu: fixup, ref #138

For posterity, what was extra confusing is that the bug wasn't actually in q3map2, but in Radiant's code that launches q3map2.

So, well done, that fix!

HAHA. Nice bug.

I can confirm that f4c0af26 fixes the bug. The code introduced in 25379b2c had indeed a bug in Radiant's state machine that parses [Variables] in command line strings, in particular in the new code to strip .[ExecutableType] on non-Windows.

NetRadiant
NetRadiant
Friday, 27 March 2020

netradiant-1.5.0-20200327git:

@illwieckz
illwieckz pushed to Unvanquished/unvanquished-mapeditor-support Mar 27, 2020
1 commit to master
  • @illwieckz 5fce04f
    buildmenu: add support for daemonmap in netradiant
@illwieckz
illwieckz pushed to Unvanquished/unvanquished-mapeditor-support Mar 27, 2020
2 commits to master

@julius__ current master may be fixed (I'm not able to reproduce anymore on my end).

Thomas Debesse (f4c0af26) at 27 Mar 17:05

radiant/buildmenu: better executable extension strip, ref #138

My build of Netradiant is: netradiant-1.5.0-20200206git

Light stage parameters: -light -lightmapsize 1024 -lightmapsearchpower 4 -deluxe -patchshadows -samples 4 -randomsamples -fast -fastbounce -bounce 4 -bouncegrid -dirty -dirtdepth 128 -dirtscale 1.0 -fill

Netradiant's console output:

...

Dirtmapping enabled

Dirtmapping depth set to 128.0

Dirtmapping scale set to 10.0

...

Thomas Debesse (179a7525) at 27 Mar 15:46

bikeshedding: some touchup on library-bundler

Bug seen while investigating #138.

Before:

$ q3map2 -light
2.5.17n-git-f41ca5cd
threads: 8
Q3Map         - v1.0r (c) 1999 Id Software Inc.
Q3Map (ydnar) - v2.5.17n-git-f41ca5cd
NetRadiant    - v1.5.0 Jan 30 2020 20:27:00
Your map saw the pretty lights from q3map2's BFG
VFS Init: /home/illwieckz/.q3a/baseq3/
VFS Init: /home/illwieckz/dev/netradiant/install/baseq3/

--- Light ---
--- ProcessGameSpecific ---
 lightning model: quake3
 lightmap size: 128 x 128 pixels
 lightning gamma: 1.000000
 lightmap colorspace: linear
 texture colorspace: linear
 _color colorspace: linear
 lightning compensation: 1.000000
 lightning exposure: 0.000000
 lightgrid scale: 1.000000
 lightgrid ambient scale: 1.000000
 shader lightstyles hack: enabled
 patch shadows: disabled
 deluxemapping: disabled
--- ProcessCommandLine ---
Erreur de segmentation

After:

$ q3map2 -light
2.5.17n-git-d0ddc5a3
threads: 8
Q3Map         - v1.0r (c) 1999 Id Software Inc.
Q3Map (ydnar) - v2.5.17n-git-f41ca5cd
NetRadiant    - v1.5.0 Mar 27 2020 01:39:21
Your map saw the pretty lights from q3map2's BFG
VFS Init: /home/illwieckz/.q3a/baseq3/
VFS Init: /home/illwieckz/dev/netradiant/install/baseq3/

--- Light ---
--- ProcessGameSpecific ---
 lightning model: quake3
 lightmap size: 128 x 128 pixels
 lightning gamma: 1.000000
 lightmap colorspace: linear
 texture colorspace: linear
 _color colorspace: linear
 lightning compensation: 1.000000
 lightning exposure: 0.000000
 lightgrid scale: 1.000000
 lightgrid ambient scale: 1.000000
 shader lightstyles hack: enabled
 patch shadows: disabled
 deluxemapping: disabled
--- ProcessCommandLine ---
************ ERROR ************
usage: q3map -light [options] bspfile

Thomas Debesse (e88113e0) at 27 Mar 15:43

Merge branch 'missingarg' into 'master'

... and 1 more commit

Illwieckz:

[q3map2] -light -lightmapsize 1024 -lightmapsearchpower 4 -deluxe -patchshadows -samples 4 -randomsamples -fast -dirty -dirtscale 1.2345 -fill "[MapFile]"

Dirtmapping enabled

Dirtmapping scale set to 12345.0

[q3map2] -light -lightmapsize 1024 -lightmapsearchpower 4 -deluxe -patchshadows -samples 4 -randomsamples -fast -dirty -dirtscale 0.001 -fill "[MapFile]"

Dirtmapping enabled

Dirtmapping scale set to 1.0

It is not multiplying by 10 as I said before, it is simply interpreting the values as integers, as if there was no "."

Garux seems right.

Related to "gtkradiant-1.5.0-20200326git", it was only a mistake from my part when making the directory to clone into. I have cloned the TTimo repo.

But it seems, if I'm not wrong, that GtkRadiant requires Scons for Python2. It is a package for Python3 here in Slackware and I'm having errors when trying to compile.

Bug seen while investigating #138.

Before:

$ q3map2 -light
2.5.17n-git-f41ca5cd
threads: 8
Q3Map         - v1.0r (c) 1999 Id Software Inc.
Q3Map (ydnar) - v2.5.17n-git-f41ca5cd
NetRadiant    - v1.5.0 Jan 30 2020 20:27:00
Your map saw the pretty lights from q3map2's BFG
VFS Init: /home/illwieckz/.q3a/baseq3/
VFS Init: /home/illwieckz/dev/netradiant/install/baseq3/

--- Light ---
--- ProcessGameSpecific ---
 lightning model: quake3
 lightmap size: 128 x 128 pixels
 lightning gamma: 1.000000
 lightmap colorspace: linear
 texture colorspace: linear
 _color colorspace: linear
 lightning compensation: 1.000000
 lightning exposure: 0.000000
 lightgrid scale: 1.000000
 lightgrid ambient scale: 1.000000
 shader lightstyles hack: enabled
 patch shadows: disabled
 deluxemapping: disabled
--- ProcessCommandLine ---
Erreur de segmentation

After:

$ q3map2 -light
2.5.17n-git-d0ddc5a3
threads: 8
Q3Map         - v1.0r (c) 1999 Id Software Inc.
Q3Map (ydnar) - v2.5.17n-git-f41ca5cd
NetRadiant    - v1.5.0 Mar 27 2020 01:39:21
Your map saw the pretty lights from q3map2's BFG
VFS Init: /home/illwieckz/.q3a/baseq3/
VFS Init: /home/illwieckz/dev/netradiant/install/baseq3/

--- Light ---
--- ProcessGameSpecific ---
 lightning model: quake3
 lightmap size: 128 x 128 pixels
 lightning gamma: 1.000000
 lightmap colorspace: linear
 texture colorspace: linear
 _color colorspace: linear
 lightning compensation: 1.000000
 lightning exposure: 0.000000
 lightgrid scale: 1.000000
 lightgrid ambient scale: 1.000000
 shader lightstyles hack: enabled
 patch shadows: disabled
 deluxemapping: disabled
--- ProcessCommandLine ---
************ ERROR ************
usage: q3map -light [options] bspfile

On my end (Linux), they both seem to work as expected:

illwieckz:~/dev/netradiant/install $ unbuffer ./q3map2 -light -dirtscale 0.2 | egrep '^Dirtmapping scale'
Dirtmapping scale set to 0.2
illwieckz:~/dev/gtkradiant/install $ unbuffer ./q3map2 -light -dirtscale 0.2 | egrep '^Dirtmapping scale'
Dirtmapping scale set to 0.2

(unbuffer from expect package is only used to be sure everything is printed out even if something fails later, I'm experiencing a segfault on my end, but that's another issue to fix).

cd /opt/gtkradiant-1.5.0-20200326git

Where do you find that 1.5 GtkRadiant? GtkRadiant 1.5 is maybe dead since a decade or something like that. TTimo repository (github.com/TTimo/GtkRadiant) provides GtkRadiant 1.6 which is a fork of 1.4 (from before 1.5 came to life, ignoring 1.5 entirely). The only GtkRadiant 1.5 that is alive is… NetRadiant (and its childs).

My first feeling is that it may be an issue in argument parsing, I was going to ask you to try to pass -dirtscale 1.2345 and -dirtscale 0.001 to see how it behaves…

It looks like @Garux is thinking the same.