AVP2 support

[x3k30c-azalia]

Dear Rene Kjellerup,

I tried to mail you earlier.
Is there any chance to remake/support AVP2? It would be great to see a linux port of these lithtech games anyway as they work horrible under wine.

Actually, I tried compiling it (AVP2) myself then saw the difficulties, the source code of the Talon engine is simply unavailable. Then I saw the GPL release and your fork and I though about something similar, maybe it would be possible to port the game to work on the GPL engine.

What I would also like to ask that would it be possible to somehow cooperate with you guys? I am definitely not at your skills but maybe I could do a few things. I saw that AVP2 is not on your support list anyway..
Actually, I gave it a try with gcc, many parts of the GPL engine could be compiled with some changes, I know that this is far from enough. On top of that, AVP2 relies on missing source code like lithtrackmgr which is simply missing from the GPL engine and it's quite a big file. Maybe I could try to reverse engineer it.

Starting from zero (with the 2012 release - I thought I give it a try in my spare time) - for me - I think it would be impossible to do all the required changes and finish it, and I don't have the time either hence I am asking it.

Some background: I haven't worked with games, except AVP1. I've made some changes on the source code of AVP1 (detail textures, texture enlargements, FOV fixes, EAX patches), a bit more playable, that's all (the Game wasn't rewritten by me of course, that's the Icculus guy).

Regards,
Akos

[Katana-Steel]

Can't make any promises at this time. Thank you for the suggestion, and I'd have a look, time permitting.

[oopsallnaps]

I've done some work in this area for NOLF 1 models. AVP2 works as well. (Vertex animations aren't supported yet.) There's some issues with bone weighting since this engine only supports 3 weights per vertex (since it's early hardware accelerated transforms), and nolf1/avp2 can have quite a bit more (since it's done in software.)

The code is an absolute mess as it basically converts the model format on the fly (by filling in the newer mesh structs, I haven't touched the rendering code.) But it's a proof of concept. I'm working on a version of this for the level format as well. After the level code is working I'll be hooking up the old API layer for Lithtech 2 games onto this engine. (It'll be ifdef'd for now.)

I've attached some preview images of a test in-game model viewer. You can see the weighting issue on Cate in the second photo.

[x3k30c-azalia]

Thanks for the response.
(The newer GPL engine allows 3 weights?) is the model/level code/format you are working on related to this issue (game assets) "Figure out the difference between formats and rework the engine to support both. (very unlikely)"?

Besides this, I can't ask or comment any more as I am not familiar with the structure of the engine anyway, client vs server etc. but by diff-ing some of the code files, there are some differences for sure but in many cases it's just the addition and/or the lack of functions or class members between the old and the new engine.

At least there is some hope as there aren't many people who would volunteer to do something like this.

I usually work on other projects. If I have some time, I will try to understand the structure and mechanism of the engine first.
I am also available at x3k30c@tutanota.com

[MelissaAutumn]

Hello! I've switched accounts to my main and finally made that code public:
https://github.com/MelissaAutumn/lithtech/commit/4a488f338623904c20ce54884c8e56f0af4303ad
https://github.com/MelissaAutumn/lithtech/issues/1

I can create a patch to upstream this if folks would like. It's a touch buggy and uhhh...probably poorly coded. (My c++ skills were never good lol.)

[x3k30c-azalia]

Thanks MelissaAutumn for this!
Good to see some progess is going on!
Personally, I'm very busy now, but the last time I saw the source code of the GPL engine I was baffled with those missing parts like lithtrackmgr. Can you comment something on this?
Regards, Akos

[MelissaAutumn]

Two more patches to finish off abcv9-v12 support
https://github.com/MelissaAutumn/lithtech/commit/4fbccdd710c47f7c42dad344ac89596327303f40
https://github.com/MelissaAutumn/lithtech/commit/3e3419f9aa8a0b183dc9769a9cfa2acdf0bf8296

First one fixes an issue where any vertices on node (bone) 0 were ignored, and the second one fixes an issue on some meshes where they don't have node weights that add up to 1.0. So a portion would be unweighted and stretch out into the great unknown. (As seen with Cate's eyelids in the screenshots above.)

Regarding lithtrackmgr, right now it seems to just be the Lithtech branch for NOLF2 / Tron. So it wouldn't have any components for NOLF1, or AVP2. I'm planning on adding that support back in, but I still have the world file format to support (mapped out, just need to find a good way to shove it into the existing renderer/game engine), and create a back-compat version of the sdk. (So the original game source codes can be recompiled against the new sdk with minimal changes.)

I don't have plans to explicitly release a patch for avp2, I think the master server folks would probably be better to take that on, but NOLF 1 and 2 is on the roadmap.

The only other area that needs work is a direct music re-implementation. But I'm also investigating and submitting patches to https://github.com/GothicKit/dmusic/.

[x3k30c-azalia]

Maybe it's better to ignore AVP2 due to its dubious status (I read that "The source code WAS released a few years ago but was revoked and made "illegal" to post.")
I didn't check whether the source code of all the other games were released or leaked. AFAIK, NOLF1 and NOLF2 are ok.
But it's "probably" not a problem if one plays with the source code and building at home without releasing a working version to the public.
But even if one tries to do this, the missing components may be a problem. lithtrackmgr might be a camera tracking, interpolating or whatever component in that particular game and the other missing components may or may not be that important - I don't know as I haven't spent that amount of time by analysing the source code, that's why I tried to ask it..
So the game could work to some degree without those components. Who knows?

The current lack of motivation towards games like AVP2 goes back in time. I read that several programmers tried to get in touch with the original game makers and they even offered to give the game a proper remake but monolith did not care or maybe theirs hands were cuffed (indeterminable legal status or whatever).

Since it is not possible (as far as I interpret) to modify the source code of AVP2, especially not to share it, it might be possible to taylor the GPL engine to AVP2 such way that those folks who want to play this game (or other lithtech based games that have a similiar status) could build the engine/game at home so that it could work.

I wonder how you guys see this..?

"(So the original game source codes can be recompiled against the new sdk with minimal changes.)" - This is probably what you thought as well Melissa!

So my main question would be, how crucial those missing components are? Because I'd prefer not to waste time by trying to implement something that can never be done.

Thanks for the reaction Melissa, as soon as time permits, I will try to go through the source code of AVP2 and the GPL engine again to try to see the picture even better, and if I can, try to contribute something in regards to Directmusic or something.

[x3k30c-azalia]

Dear Rene, Dear Melissa

I'm not sure I'm not doing anything wrong by writing this here, but if it's sensitive, you can delete it.
I don't want to pressurize you with this, you're probably busy or have better things to do than this.

First, NOLF2 works beautifully in Linux, excellent work!

I got a bit farther. I managed to compile the game code with Jupiter's headers. Changes were needed in the game, and added the new changes to the engine by #ifdef TALON - #endif directives. Now I can load the .so files with Lithtech but the game freezes at the very beginning in ButeMgr. If I comment out some initialization (Weapons etc.) the SDL window will appear but nothing really works (a lot of missing functions etc.).

I cannot write the exact steps in a few lines, maybe in a nutshell, but I prepared a repository in case if someone is interested. I'm probably still lagging behind a lot.

First, changes were needed for the game to compile with gcc (>=10) and with the old headers.

Next, I tried to isolate only the necessary functions from the engine, but the game still uses too many.

Then I tried to compile it with Jupiter's headers, even more changes needed here. Implementation of missing functions is to be done, animations are commented out etc. The amount of functions added to Jupiter is not catastrophic though.
Maybe it's possible to write a wrapper that translates the calls from the old engine to the new and vice versa, but this way seemed easier.

About the missing components/functions: trackmgr might have been used only by the game developers. The other missing components were recreated with minimal functionality, here it means that they do nothing and throw a "not implemented" message. That is also thrown for missing functions in the engine.

If these worked properly, how difficult would it be for the new engine to work with the old levels/models?

ps. I also have gdb logs if someone is interested.

Regards,