Speeder

 

 

@Minime891 sorry it might take some time since I'm rewriting the tool from scratch (with missing file format support, like .uva and .pmi). :D

@TDUZoqqer I didn't give a thought to it yet. :D Using a blur mask should be good enough (the whole procedural wheel blur they use in Forza is probably way overkill).

TDU2 probably use the same system as TDU1, so the distant sector (here the moutains) uses the 'Area-x-x-x-x' geometry file instead of the 'Sector-x-x-x-x'.

 

To put it short:

- An Area folder contains several sectors files + a area bnk, which is used as the highest LOD (level of detail) on the map screen and in game (for the farest sector that's not loaded into memory yet)

- A sector acts as the lowest LOD and is (afaik) only used in game. It also contains multiple LOD for sector specific geometry (e.g. lighthouse, ...) as well as an heightmap for the terrain (which is splitted into several patches for tesselation)

 

There might also be a way to edit the way the game streams the sectors (by adding more sectors to be loaded into memory) if it's not hardcoded into the game code. :geek:

I have run into a problem with the .shk files. Even when just importing into 3ds max and exporting whilst doing no editing results in a messed up .shk.

As you can see in the picture the .shk seems to fall short of where it was and this subsequently messes the whole .shk up. This also screws the shadows up too it seems.

Not holding much how as you've moved on but in the smallest chance i thought i would at least post what I've found.

Don't expect too much from the .shk generation, as far as I remember the MOPP (the bytecode used by the physics engine) generated is totally wrong.

I guess it could be cool to give it some time and rewrite a tool for map edition (research papers on MOPP emerged those last years). :D

@Ryzza5 Yep that's exactly how it works :D

Since I'm currently working on a virtual file system (GitHub - ProjectHorsepower/horsepower at modern_io), I thought it would be nice to write a small post about mods (well at least how I plan to implement these). :)

 

Mods are split in two categories:

• Local Mods: mods applied locally; meaning others won't see the mod. This is typically anything that doesn't interact with gameplay (3D models, sounds, textures, graphics effects, ...).
  
• Global Mods: mods applied globally. This kind of mod would be stored on the server, and shared to the connected clients (either direct download via HTTP or simply ask the user to download the mods from an external link). This kind of mods cover everything the local mods don't (3d levels, physics settings, gamemode, time of day, etc.).
  

What does it have to do with the virtual file system thingy? :D

 

Thanks to a virtual file system (vfs), you can virtually build a directory from different locations (zip archives, disk folders, network content, ...). And that's where it gets interesting.

With that setup, you can easilly extend the game content without touching your base data. It basically means mods can be shared as zip archives, drag 'n dropped to the mods folder and that's all. If the mod doesn't match your expectations, and you want to uninstall it, you simply remove the archive from the mods folder.

Another interesting feature (not implemented yet, but shouldn't take too long) is stackable mods, which let you override other mods content without needing to disable/remove these.

 

Example:

-Mod A introduces new car model CarModelA, with its own set of texture, 3d model, sound, ...

-Mod B replaces CarModelA texture set, and introduces new textures for environment

-Mod C replaces CarModelA sound

-Mod D replaces CarModelA texture set, overriding Mod B (but doesn't disable the new textures introduced by mod B)

 

This would work either automatically (based on the archive date) or manually (introducing a priority index or something like that).

The folder hiearchy would look something like this:

| Game.zip (base game data that should not be modified)

| Game/ (folder containing permanent changes, if you want to override base game data)

| Mods/ (folder containing mod archives, which can be added/removed on the fly)

| Game.exe

 

Last week progress (warning: contains geeky mumbo jumbo :p):

• Implemented GameObject binary serialization/deserialization
  
• Better GameObject API (e.g. component can be added in a template fashion, 'gameObject->AddComponent<Mesh>( "Tree.mesh" )')
  
• Better unicode support (which should allow to port the code easilly to *nix or MacOS later)
  

I fail to see how both project are incompatible.

Plus your work on UI looks pretty great, it would be a waste to throw that away don't you think? :)

@Diablo: roger that :D I have setup a git repo with the current codebase: [URL='https://github.com/ProjectHorsepower/horsepower']GitHub - ProjectHorsepower/horsepower: Horsepower - MMO Racing Game[/URL] The code should be compilable (as long as you follow the Readme). There is not code guideline/contribution guide whatsoever yet, if people are willing to contribute I'll make one. :) I've also made a [URL='https://trello.com/b/XgBYgiT3/']Trello[/URL] to manage the project's roadmap

@TDUZoqqer: right now it roughly takes 3.5ms per frame (with temporal reprojection enabled) at 1080p, which is expensive I have to admit. It should be possible to go below 1.5ms ([URL='http://advances.realtimerendering.com/s2017/Nubis%20-%20Authoring%20Realtime%20Volumetric%20Cloudscapes%20with%20the%20Decima%20Engine%20-%20Final%20.pdf']as suggested by Guerrilla Games presentation[/URL]) with some optimization. I guess UE4 volumetric clouds implementation is based on this paper as well, so it should be viable to use these for high quality settings. :) Raycasting; as you already mentioned it's really hard to get stable constraint-based physics (well unless you write your own physics engine as the guys behind BeamNG did :D). My implementation is really basic right now (e.g. I only apply a single force on the hull, I don't support gearbox yet, ...) but is stable even at low framerate.

What is the creation of the game world like? If it is fairly simple and intuitive, I can help you with huge desire.

I hope something as simple as a sandbox editor (like TDU challenge editor or Halo's Forge), in which you'd simply drag and drop entities and voila! :D

 

Nice to see another attempt of a car game! I think you are far more skilled than I am, so I guess there will be some actual progress to show compared to me ( :lol: )

 

Good luck with the project :) I would love to help, but I can't code C++.

Yeah that's the problem with C++. I'd love to add node-based scripting to make contribution easier, but I don't have enough free time to do so. :p

 

I should setup a repo quite soon (on github or bitbucket) so that people can have a look at the code and (eventually) contribute to the project.

 

Anyway some recent (visible) progress: volumetric clouds (with unfinished lighting) and anti aliasing (which still need some work)! I've also started reworking physics (which still need loots of work).

Well technically the client could send the hash of the AiConfig file to the server, the server would check wether or not the hash matches with the "official" AiConfig file, and then allow or reject the connection. :) (and eventually display a prompt asking to download the official patch; imo doing a silent download is not right (the player should always know what's going on behind the scene)).

 

But still, I'm not really sure if messing with players' files feel right (and on top of that, I don't think I'd have time to implement such feature :p).

@BeSweeet: Physics wise or graphics wise? :D

Does UE4's physics engine supports softbodies?

It could definitely improve physics realism (and also gives nice car damages, like BeamNG did).

 

Anyway good luck with your project, you seem to be already on a good path. :)

 

(btw imo you should reduce bloom intensity at night, or interpolate the camera exposure to avoid moon über-shininess :D)

Need for Script? :D

 

Anyway the gameplay and physics look as meh as the previous NFS. :hmmm:

I definitely don't want to let your hopes down, but from what I can recall, we reached a dead-end since we did not have enough material to work on multiplayer races/clubs. :(

 

I am also really busy with work and working on another hobby project as well (to put it short: a free open source moddable game dedicated to mmo racing games, might create a thread on the board when we reach a presentable state :D).

Sounds like it's the same deal.

 

I haven't launched TDU on Windows 10 yet but I am pretty sure they use the same kind of injection. :)

 

@lean: if you read this, could you pm me back the irc address? HDD crash w/o backups happened :D

You should install it (without uninstalling anything!), DirectX version doesn't matter in this situation. :DAlso I've heard that Windows10 has some Xbox PVR stuff pre-enabled so you might have to disable it.

Well, tempted guy me upgraded to Windows 10 and reinstalled TDU with my legal copy. I could start the game fine, but when i add the d3d9 file , i can't even boot up the game. Crashes before the logos even show up. Yes, I'm using compatibility mode and the game is basically a clean version. Patched to 1.68b like my old game. But it boots and plays fine without the d3d9 file. Any clue what's going on or is it just windows itself?

DirectX12 or Windows10 are probably incompatible with the mod. But it's too new to tell at the moment. :)

You can sometimes get disconnected from the server because you stayed inactive for too long (by inactive I mean not playing online :D). Simply restart the game after you've done your missions as Hypercycle said.