ReLite

[cosmos10040]

0.6.3a runs great I got a crash with violent Doom and legion monsters on doom 2 when finishing first map, endgame was enabled and crashed then I disabled and same crash after completing the map again. Running doom complete


MAP02 - Underhalls

VM execution aborted: array access out of bounds. Size = 15, current index = 115

Called from hd_png.get_color at relite_0.6.3a.pk3:zscript/hd_image.zs, line 301
Called from hd_png.Init at relite_0.6.3a.pk3:zscript/hd_image.zs, line 184
Called from hd_relite_Events.add_textures at relite_0.6.3a.pk3:zscript/hd_relite_events.zs, line 452
Called from hd_relite_Events.find_textures at relite_0.6.3a.pk3:zscript/hd_relite_events.zs, line 512
Called from hd_relite_Events.WorldLoaded at relite_0.6.3a.pk3:zscript/hd_relite_events.zs, line 34
]
]

I don't get any crash running Violent Doom, Legion, and Doom complete. Not sure what is different.

Is "legion" this?

So I got the same error typing next map, also using a different mod voyager infernal without monsters and same results. Still looking for source.

[cosmos10040]

0.6.3a runs great I got a crash with violent Doom and legion monsters on doom 2 when finishing first map, endgame was enabled and crashed then I disabled and same crash after completing the map again. Running doom complete


MAP02 - Underhalls

VM execution aborted: array access out of bounds. Size = 15, current index = 115

Called from hd_png.get_color at relite_0.6.3a.pk3:zscript/hd_image.zs, line 301
Called from hd_png.Init at relite_0.6.3a.pk3:zscript/hd_image.zs, line 184
Called from hd_relite_Events.add_textures at relite_0.6.3a.pk3:zscript/hd_relite_events.zs, line 452
Called from hd_relite_Events.find_textures at relite_0.6.3a.pk3:zscript/hd_relite_events.zs, line 512
Called from hd_relite_Events.WorldLoaded at relite_0.6.3a.pk3:zscript/hd_relite_events.zs, line 34
]
]

I don't get any crash running Violent Doom, Legion, and Doom complete. Not sure what is different.

Is "legion" this?

Yes that is correct mod, argh. I will see if it's another graphic mod like the PBR 256 I had to remove. Will test some today.

I don't get the crashes with aracnicide, could be violent Doom causing the crashes, I also used the monster pack with no issues.

[Hey Doomer_]

Progress 0.6.4a

My converted & modified regex engine works well but is understandably slower than regex engines written in C used by most languages. Still it returns all matches in a few seconds. After minifying, tokenizing ("BRIGHTMAP FLAT" & "BRIGHTMAP TEXTURE" to "@@"), and stripping text to reduce search length this does the lifting:

Code: Select all

	foreach (def: defs) re.match("@@ [ A-Z0-9\"./_]+", def, matches);

The engine searches forward for one or more characters matching those in brackets (in ZScript double quotes are escaped as \") up to a character not in the set (a closing curly brace). The utility of this is it works regardless of text formatting, comments, order of attributes, etc. 0.6.4 works with bmplustextures, Sbrightmaps, Brutal Doom (which appears to embed Sbrightmaps), and anything else defining FLAT and TEXTURE brightmaps. Although I note these are oddly a rarity.

I'll also add support for brightmaps/auto files in 0.6.4, but from my limited observations most of these appear to be sprites.

Release "soon."

[Hey Doomer_]

Progress 0.6.4a

Code: Select all

0.6.4
11/05 - added blend function to texture colorization
11/05 - added mString class to mixins, includes minify functions
11/06 - changed hd_regex.match to return array of matches
11/09 - added hd_brightmaps with regex searching, file checking, bug fixes
11/09 - added pixel reconstruction for scanline filter type > 0 (https://pyokagan.name/blog/2019-10-14-png/)

Regex searching works with bmplus_textures and SBrightmaps. Note many lights will be missing using bmplus because some definitions are commented out. The new code ignores comments.

I've added code to reconstruct pixel data. While unzipped pixel data is always arranged in horizontal scanlines, depending on the encoder scanlines are filtered using simple algorithms to improve compression ratios; this process must be reversed to get original data. (Most - not all - bmplus scanlines are unfiltered.) Now both mods return the same data from different PNG types (SBrightmap scanlines are all filtered):

---

bm_plus

Found TEKGREN5 BRIGHTMAPS/TEXTURES/DOOM2/TEKGREN5.PNG percent 5.371094 avg 248
Found SW2BRCOM BRIGHTMAPS/TEXTURES/DOOM/SW2BRCOM.PNG percent 0.305176 avg 162
Found FWATER4 BRIGHTMAPS/TEXTURES/DOOM/FLATS/FWATER4.PNG percent 0.000000 avg 0
Found SW1BRCOM BRIGHTMAPS/TEXTURES/DOOM/SW1BRCOM.PNG percent 0.488281 avg 145
Found LITE5 BRIGHTMAPS/TEXTURES/DOOM/LITE5.PNG percent 50.781250 avg 137
Found EXITDOOR BRIGHTMAPS/TEXTURES/DOOM/EXITDOOR.PNG percent 7.052951 avg 193
Found EXITSIGN BRIGHTMAPS/TEXTURES/DOOM/EXITSIGN.PNG percent 9.570313 avg 125
BRIGHTMAPS/TEXTURES/DOOM2/SPACEW3.PNG Only bit depth 8 supported
Found SPACEW3 BRIGHTMAPS/TEXTURES/DOOM2/SPACEW3.PNG percent 0.000000 avg 0
Found TEKGREN3 BRIGHTMAPS/TEXTURES/DOOM2/TEKGREN3.PNG percent 0.024414 avg 213
Found TLITE6_4 BRIGHTMAPS/TEXTURES/DOOM/FLATS/TLITE6_1.PNG percent 10.253906 avg 226
Found TEKWALL6 BRIGHTMAPS/TEXTURES/DOOM2/TEKWALL6.PNG percent 0.006104 avg 107
Found TEKGREN1 BRIGHTMAPS/TEXTURES/DOOM2/TEKGREN1.PNG percent 0.048828 avg 181

sbrightmaps

Found GRNLITE1 BRIGHTMAPS/DOOM/GRNLITE1.PNG percent 19.726563 avg 177
Found TEKGREN5 BRIGHTMAPS/DOOM/TEKGREN5.PNG percent 5.371094 avg 248
inflate_block_no_compression
inflate_block_no_compression done
Found SW2BRCOM BRIGHTMAPS/DOOM/SW2BRCOM.PNG percent 0.305176 avg 162
Found FWATER4 BRIGHTMAPS/DOOM/FWATER4.PNG percent 0.000000 avg 0
inflate_block_no_compression
inflate_block_no_compression done
Found SW1BRCOM BRIGHTMAPS/DOOM/SW1BRCOM.PNG percent 0.488281 avg 145
Found CEIL1_3 BRIGHTMAPS/DOOM/CEIL1_3.PNG percent 13.012695 avg 215
Found CEIL1_2 BRIGHTMAPS/DOOM/CEIL1_2.PNG percent 13.427734 avg 218
Found LITE5 BRIGHTMAPS/DOOM/LITE5.PNG percent 50.781250 avg 137
Found EXITDOOR BRIGHTMAPS/DOOM/EXITDOOR.PNG percent 7.052951 avg 193
Found EXITSIGN BRIGHTMAPS/DOOM/EXITSIGN.PNG percent 9.570313 avg 125
Found SPACEW3 BRIGHTMAPS/DOOM/SPACEW3.PNG percent 0.341797 avg 255
Found FLAT2 BRIGHTMAPS/DOOM/FLAT2.PNG percent 21.630859 avg 165
Found TEKGREN3 BRIGHTMAPS/DOOM/TEKGREN3.PNG percent 0.024414 avg 213
Found TLITE6_4 BRIGHTMAPS/DOOM/TLITE6_1.PNG percent 10.253906 avg 226
inflate_block_no_compression
inflate_block_no_compression done
Found TEKWALL6 BRIGHTMAPS/DOOM/TEKWALL6.PNG percent 0.006104 avg 107
Found TEKGREN1 BRIGHTMAPS/DOOM/TEKGREN1.PNG percent 0.048828 avg 181

Differences are differences between these mods. "Percent" and "avg" are brightness metrics used by ReLite. This is now much closer to a universal solution.

Release "soon."

[Hey Doomer_]

Released 0.6.4a

Testing is OK with this. This now works with bmplus_textures or SBrightmaps. The latter is in Brutal Doom already and so doesn't need to be added. Since the regex engine is written in ZScript and not C it's a bit slower and load times will be a few seconds. However, this also adds file checks that were not in 0.6.3a and should be more compatible with more mods.

[cosmos10040]

I am not able to get new version to load up with the same settings and mods as previous version 0.6.3. However I looked into the log and found this:

Script warning, "relite_0.6.4a.pk3:zscript/hd_mixin.zs" line 55:
Comparison between signed and unsigned value
Script warning, "relite_0.6.4a.pk3:zscript/hd_mixin.zs" line 55:
Comparison between signed and unsigned value
Script warning, "relite_0.6.4a.pk3:zscript/hd_png.zs" line 23:
Comparison between signed and unsigned value
Script warning, "relite_0.6.4a.pk3:zscript/hd_regex.zs" line 39:
Comparison between signed and unsigned value
Script warning, "relite_0.6.4a.pk3:zscript/hd_regex.zs" line 106:

[Hey Doomer_]

I am not able to get new version to load up with the same settings and mods as previous version 0.6.3. However I looked into the log and found this:

Script warning, "relite_0.6.4a.pk3:zscript/hd_mixin.zs" line 55:
Comparison between signed and unsigned value
Script warning, "relite_0.6.4a.pk3:zscript/hd_mixin.zs" line 55:
Comparison between signed and unsigned value
Script warning, "relite_0.6.4a.pk3:zscript/hd_png.zs" line 23:
Comparison between signed and unsigned value
Script warning, "relite_0.6.4a.pk3:zscript/hd_regex.zs" line 39:
Comparison between signed and unsigned value
Script warning, "relite_0.6.4a.pk3:zscript/hd_regex.zs" line 106:

Those warnings have no effect.

[cosmos10040]

I am not able to get new version to load up with the same settings and mods as previous version 0.6.3. However I looked into the log and found this:

Script warning, "relite_0.6.4a.pk3:zscript/hd_mixin.zs" line 55:
Comparison between signed and unsigned value
Script warning, "relite_0.6.4a.pk3:zscript/hd_mixin.zs" line 55:
Comparison between signed and unsigned value
Script warning, "relite_0.6.4a.pk3:zscript/hd_png.zs" line 23:
Comparison between signed and unsigned value
Script warning, "relite_0.6.4a.pk3:zscript/hd_regex.zs" line 39:
Comparison between signed and unsigned value
Script warning, "relite_0.6.4a.pk3:zscript/hd_regex.zs" line 106:

Those warnings have no effect.

Interesting, I can run it with minimal mods, now I have to see which one is causing the hangs lol works great!

Update: when i removed "more lights 2.5 " I was able to load up maps. Not sure why that would be that cause, then again I wouldn't really need that anymore correct?

[Hey Doomer_]

Interesting, I can run it with minimal mods, now I have to see which one is causing the hangs lol works great!

Update: when i removed "more lights 2.5 " I was able to load up maps. Not sure why that would be that cause, then again I wouldn't really need that anymore correct?

I think more lights adds glows to flats. I'll have a look to see if that's causing a problem.

Edit

Problem is in the more lights gldefs file there is a syntax error where a block comment terminator is *. instead of */ - I've fixed that for the next version.

[cosmos10040]

Excellent!

[Hey Doomer_]

I'm experimenting with shadow mapping. It's possible to get "fuzzy" shadow maps. Note these comparison images:

e1m5 - fuzzy
e1m5 - sharp
map10 - fuzzy
map10 - sharp

The only difference is swapping spotlights for pointlights attached to textures.

[DarkkOne]

I like the sharp better, honestly. I think I'd like the fuzzy better, except they seem to... not expand past the corner/edge. It makes the lit areas look like they're narrowing as they move further in, which just feels really odd.

[Hey Doomer_]

I like the sharp better, honestly. I think I'd like the fuzzy better, except they seem to... not expand past the corner/edge. It makes the lit areas look like they're narrowing as they move further in, which just feels really odd.

Thanks for the feedback and understood. I've increased the spoti parameter (these are just flattened spotlights) to improve this and added as a proper option to the menu for users to check out.

Fuzzy shadow maps reduce banding artifacts and don't light the texture or flat generating the light.

[Hey Doomer_]

Progress 0.6.5a

Code: Select all

0.6.5
11/11 - added check in minify() for no terminal block comment marker
11/13 - added fuzzy shadow maps for flats and textures
11/15 - added fuzzy dispersion patterns

I've finally decided against the "fuzzy" menu option and looked at varying dispersion patterns based on textures, flats, and if a texture is in a window. Previously all light dispersion patterns were 360 degrees because everything was a point light. However, for flattened spotlights that is no longer true. I have made adjustments and placed the light and type of light depending on the sector. This leads to interesting effects such as this:

e2m2 - "fuzzy"

This rather astonishing effect is caused by a flattened spotlight on the floor of the texture with a -90 pitch, because the floor difference is greater than the ceiling difference. Still testing.

[Skrell]

LOVE it!