[Mblackwell]

 Gambini, on 21 June 2013 - 06:03 AM, said:

He has the lightsource but he has surfaces with their own shading attributes. The only way a smoothing method could work is by generating continuity between walls of the same shade and the same texture. Like, if five connected walls share shade and tilenumber, then they are automatically added to a smoothing group. Mappers then should avoid using the same shade value on perpendicular walls. Very complicated IMO.

The sector walls shouldn't be shaded at all except by a light. Otherwise you're stuck with the gradient. If it looks broken up define a normal map for the texture.

[Plagman]

Even with uniform shading and just a light you'll get sharp edges like this, because as someone mentioned there's no smooth shading for walls at the moment. If we had a good way to define smooth groups of walls in Mapster then the rendering part would be pretty simple to handle, but as-is it's not possible to do automatically, as it would smooth some cases where you don't want it smoothed.

[Diaz]

 Gambini, on 21 June 2013 - 06:03 AM, said:

He has the lightsource but he has surfaces with their own shading attributes. The only way a smoothing method could work is by generating continuity between walls of the same shade and the same texture. Like, if five connected walls share shade and tilenumber, then they are automatically added to a smoothing group. Mappers then should avoid using the same shade value on perpendicular walls. Very complicated IMO.

Ehm, not really. Shade and polymer lighting are two completely different things. You could have smooth lighting and still shade walls manually. I don't think using shade as a basis for smoothing would be useful at all.

[Gambini]

But if you don´t shade the walls they´re fullbright. How´s light going to affect a surface that is already lit? I´m thinking in the confines of Eduke32, where despite having polymer lights, surfaces are still shaded manually.

[Diaz]

 Gambini, on 22 June 2013 - 07:12 AM, said:

But if you don´t shade the walls they´re fullbright. How´s light going to affect a surface that is already lit? I´m thinking in the confines of Eduke32, where despite having polymer lights, surfaces are still shaded manually.

Yeah but on 100% Polymer stuff shade is basically the same on the whole map (pitch black, or just a slight ambient level). So using it for smoothing would be kinda doomed.

In Source, Quake, etc. compiling tools, you can define a maximum angle at which faces meet to apply smoothing on the resulting light maps, perhaps an adjustable setting for that in Polymer would work?

[Plagman]

But then would the rule be, angle inferior to threshold AND same tile, or would it apply to different tiles as well? It seems hard to imagine either of the two would do what you want for the whole map.

[Fox]

Using tile as a reference really wouldn't work, especially because there are arts composed of multiple tiles.

Won't it work using the the angle difference as a reference? I.e. if the angle is 90º or more there won't be any smoothness at all, and if the angle is 0º then it will be 100% smooth.

[Gambini]

That doesn´t sound optimal either, there are times when you have small angles and still want sharp borders. I´m not sure of how much data the smoothing group could gather but what about: same tile, same shading, aligned to each other, less than 30 degrees of difference. The only part that sounds difficult to determine is the angle.

[Diaz]

Well with Quake-based engines, the default angle threshold for smoothing (don't recall what it is) didn't really seem to be much of a problem. With Source, stuff could be added to smoothing groups, and the rules were that texture needed to be the same in all faces and PERFECTLY aligned. It was hard to get them right.

I think under most circumstances we would prefer to have smooth shaded faces on angles that are like this: \_

and no smooth faces on angles that are like this: /_ |_

But some adjustability would be desired because we might want an angle that is just slightly above 90º to have sharp borders. In this case, either an adjustable threshold, or having the smooth shading only apply if textures are perfectly aligned; then they could maybe be manually mis-aligned by one texel so the smoothing would not occur. I understand this is far from optimal though. And of course this would just affect Polymer lights, not classic shading.

[Gambini]

Diaz, on 22 June 2013 - 03:24 PM, said:

With Source, stuff could be added to smoothing groups, and the rules were that texture needed to be the same in all faces and PERFECTLY aligned. It was hard to get them right.

yeah it sucks.

Quote

they could maybe be manually mis-aligned by one texel so the smoothing would not occur.

Exactly, you have then a controllable method. Not that bad.

[Tea Monster]

Is there any way of having a heat haze effect on a model in polymer?

[Fox]

Is it possible to add two wall cstat bitfields to inherit the top / bottom ROR texture alignment?

[Gambini]

Tea Monster, on 23 June 2013 - 03:37 AM, said:

Is there any way of having a heat haze effect on a model in polymer?

Those heat affects can be replicated with animated dudv textures. Not sure if Polymer supports dudv shaders (according to those old promotional shots, it does).

[Hendricks266]

Fox, on 24 June 2013 - 03:26 PM, said:

Is it possible to add two wall cstat bitfields to inherit the top / bottom ROR texture alignment?

Can't you just do that yourself? Crawl the sector structure, find the relevant alignment, and compute it yourself. No need to add hacks everywhere in places they don't belong.

[Hendricks266]

Fox, on 21 June 2013 - 10:09 AM, said:

We need a new starttrack command that can specify the volume. The setuserdef[].volume_number trick doesn't work in multiplayer because it must be set per-player, and CON doesn't have power to change the userdef structure of each player.

On IRC, Mblackwell suggested using a system with a global var or two so that all players will set the userdef and start the track independently as a simple yet effective workaround.

I just threw this together off the top of my head:

gamevar musicVolume -1 0
gamevar musicLevel -1 0

onevent EVENT_GAME
// whenever you need to set a track
    {
        setvar musicVolume 3
        setvar musicLevel 4
    }
endevent

// There might be a better event or circumstance but it may have to be unsynchronized. I'm not sure.
onevent EVENT_DISPLAYREST
    ifvarg musicVolume -1
    {
        setuserdef[THISACTOR].volume_number musicVolume 
        setvar musicVolume -1
    }

    ifvarg musicLevel -1
    {
        starttrackvar musicLevel
        setvar musicLevel -1
    }
endevent

[Fox]

I got another issue now. Basically I want rewrite the end of level code, so that the screen slowly fades before the level ends. The problem is that I do it by checking if the .gm of the players has been set to MODE_EOL, removing the bitfield and using startlevel some time later. The problem is that when .gm is set, it also set userdef[].level_number and userdef[].volume_number to that of the next level, which until now I was just setting it back to what it was before.

The problem is that now I realized that you can't set the userdef for each player (and the hack above won't serve because it must be synchronized). On theory this will have some side-effects in multiplayer, such as displaying a different episode and level names in the Automap for a moment, etc.

Would it be too much to ask for an EVENT_ENDOFLEVEL which I could prevent the level from ending by setting RETURN to -1? I know that as far Eduke32 goes this is an theoretical bug since Multiplayer is unsupported.

[Tea Monster]

Gambini, on 24 June 2013 - 04:06 PM, said:

Those heat affects can be replicated with animated dudv textures. Not sure if Polymer supports dudv shaders (according to those old promotional shots, it does).

Cheers!

[NightFright]

EDuke32 r3912:
I get "Invalid Map Error" with Mapster32 when trying to open a map. Doesn't happen with r3905 or earlier, so it must have been broken in the latest revisions.

*EDIT, July 1*
Fixed in latest snapshot. Another thing, has anyone had a chance to take a look at the Polymer issues I posted over here?

[fgsfds]

Well, now I really need eduke with lua, so I went back to the attempts to build it.

Helixhorned, on 16 June 2013 - 02:50 AM, said:

You'll also need a statically linked library of LPeg 0.12 (its makefile doesn't have an option to build one, but I intend to publish the necessary changes on github sometime).

If I can't build it from sources, where can I get it then?

[Fox]

Hendricks266, on 24 June 2013 - 06:19 PM, said:

On IRC, Mblackwell suggested using a system with a global var or two so that all players will set the userdef and start the track independently as a simple yet effective workaround.

I just threw this together off the top of my head:

Nevermind that, I just realized that it's better to have it unsynchronized (assuming it's safe and won't cause sync errors). That way you can have a music for when a player gets to certain area of the level, or music for the menus, etc.

(still we need a way to play the main menu / briefing / loading tracks and stop the track)

[Jblade]

This doesn't really constitute as a bug, but I thought I'd mention it incase it's not intentional. When map shade preview is enabled, if you have any SEs with a lotag of 3 and a spritepal of 0 it will set everything in that sector to pal 0 as well - it works fine ingame, but I thought i'd mention it since it did confuse me a bit at first (That feature is incredibly handy though)

EDIT: I just used a lightswitch SE and realised that even if it's pal is set to 0 it will still set everything in it's sector to pal0 as well - was there a way to bypass this?

[Fox]

So nocompress, nodownsize and alphacut doesn't work for model textures?

[Jblade]

Did a method of getting a tile's size ever get added to the game? I'm working on something that I need to get the art tile's size for so I can apply an offset to it (otherwise it works fine but as soon as it switches to a frame smaller in physical size it floats away from the ground)

[Fox]

Yeah, there is tilesizx[] and tilesizy[] which work like gamevars. But it seems they may be replaced by something else in some time because Lua CON doesn't handle it well.

[Jblade]

Cool, that's perfect for what I need. thanks!

[Hendricks266]

Fox, on 08 July 2013 - 07:24 AM, said:

Yeah, there is tilesizx[] and tilesizy[] which work like gamevars. But it seems they may be replaced by something else in some time because Lua CON doesn't handle it well.

Thinking out loud to Helix here: Would we replace these with something like gettile[tilenum].sizx? That would make life easy to add the picanm animation structures.

[Fox]

Tile structure members seems interesting, but I think that a series of regular commands could do the job, Like gettilesizx <tilenum> <gamevar>.

[Helixhorned]

Fox, on 08 July 2013 - 07:24 AM, said:

Yeah, there is tilesizx[] and tilesizy[] which work like gamevars. But it seems they may be replaced by something else in some time because Lua CON doesn't handle it well.

LunaCON handles it just fine, but some operations were simply not implemented for system gamearrays prior to r3940. Now, the only thing that remains is that one can't writearraytofile the contents of a system array, but it's unlikely to have any practical relevance.

Hendricks266, on 08 July 2013 - 08:01 AM, said:

Thinking out loud to Helix here: Would we replace these with something like gettile[tilenum].sizx? That would make life easy to add the picanm animation structures.

That would probably have been the preferred syntax from the start, but there ought to be a reason why you went the array route, no? In m32script, light access is kind of mixed with tsprite, so I guess it has something to do with requiring to duplicate/specialize a good bunch of code for a new pseudo-structure? In LunaCON, adding it would be easy, but I don't know (and don't care, except that the two should stay reasonably in sync) about C-CON.

Fox, on 08 July 2013 - 08:52 AM, said:

Tile structure members seems interesting, but I think that a series of regular commands could do the job, Like gettilesizx <tilenum> <gamevar>.

I'm very reluctant of adding new commands because they change the language by introducing a new keyword. Most programming languages, and CON is no exception, forbid keywords as identifiers, even though they're lexically valid names otherwise.

Edit: you might counter "but gettile would be a keyword, too". Yes, but at least it would subsume multiple items, so one feels a little more at ease with adding it.

[Skulldog]

Check the shadow out.

[Fox]

Is it possible for fogpal or makepalookup to remap the black fog? By default a value of 0 0 0 use the lookups shade table, however I had like the game to generate it's own black fog table. Perhaps it could use a value that is not in the 0-63 range.

===//===//===

I should ask... why does the messages and inventory bar are not affected by the status bar scale option?

===//===//===

Is it possible to add a precache-like command that can be used in actor / events code? For example, it could be used to load all tiles used in a screen sequence.

===//===//===

Is it possible to check whenever a texture has a model (or HUD model) definition? That seems to be necessary to make a proper weapon display code.