[e107 thread] Tech Demo Thread

A forum about everything related to the Cafu World Editor CaWE: mapping, editing, scripting, making GUIs and fonts, using the BSP, PVS and Light compilers.
Post Reply
User avatar
Site Admin
Posts: 2169
Joined: 2004-08-19, 13:46
Location: Germany

[e107 thread] Tech Demo Thread

Post by Carsten » 2005-09-19, 21:40

Thrawn wrote:Somehow I get the feeling that the idea IRC -> Teamspeak doesn't work! Then let's make a big thread collecting all the ideas and then let's start working!! Right now, Carsten is buzy with other stuff but untill we finished the discusion and have collected all the ideas, some time has gone by and maybe then we can start straight forward! Let's get the thing done!

All right, post your suggestions here!
Scott wrote: Has anybody designed anything at all for it? Has there been anykind of start at all?
Carsten wrote:Yes and no. There was a fruitful discussion in the old forum months ago, inclusive sketches and long feature lists and "theme" descriptions.
I'll re-post the former features lists and descriptions soon here at this thread.

However, indeed nothing "solid" had been modelled so far.
This is problematic, because the TechDemo map (.cmap) file must be modelled by a single person, as such maps cannot easily be worked on by several individuals.
I'll also start to code additional GPU effects and new entities soon.

On the other hand, anybody can create separate artwork for integration with the TechDemo. E.g. entity code, models and prefabs can be created separately and later be merged with the TechDemo.
(Prefabs are normal cmap files that are copied into the TechDemo map by copy'n'paste.)
levaravel wrote: Just a little concept sketch i drew about 2 days ago. Nothing special yet but maybe that will give you guys some ideas.
here another sketch...let me make a 3D model to show you exactly, what i mean.
nothing exciting, no details snf no real texturing

just another idea..
A Creature
more renders..
Scott wrote:I like your idea for the layout, and those domes look good as well.
Whats the polycount on your creature?
levaravel wrote:the creature is still a sub-patch model, i didn't add any detail yet... the thing is, i do like 100'000+ poly characters, and render it out into a texture and after that, i apply it onto a low-poly character, 1 - 3K polys
maybe there should be an underground floor, with a arched corridor, and on both sides, there is one room after the other, and in every room, there's a feature of the engine beeing shown. maybe with some text in the UI, that describes the feature, when you enter the room?
Corridors Idea
posting the image didn't work, sorry.
a few ideas for the rooms:

Skeletal Deformation
- a few characters walking / fighting / doing other actions.

Lighting Room
- demonstrating soft shadows, per-pixel lighting. and more lighting stuff

particle rooms
- a car driving in a circle on a muddy / sandy underground, spinning around dust particles.
- shooting range: a few weapons ligned up, shooting in like 1-second intervals, showing the particle engine being used for weapons.

Reflections / Refractions
- chandelier reflecting and refracting himself and the environment (pre-rendered reflection maps, but frame-buffer distortion for the refraction effect.
Carsten wrote: Levaravel, thanks for all your suggestions!

Well, to be honest, the monster is a bit too much like from a comic for my taste. E.g. there are round and smooth features at its head and fingers. The atmosphere in the underground lab is thoroughly aimed and planned towards horror, though...

However I like how you integrated the 7-sided buildings with the terrain.

UI text: Yes, that's what I'm planning. The player walks into a new room, and then a small info text pops up that briefly describes the effect.

Ideas for the rooms: Thanks. I've a big list somewhere, too (it used to be in the old forums, iirc). Will re-post it here soon. In the meanwhile, here is the link to the old thread. Recommended reading!
levaravel wrote:thanks carsten!

Does the Engine support models to be used for terrains? it would be alot easier that way. modelling actual terrain, than using heigh maps.
should the quality level be at unreal3-engine level? or PS3 graphics level?


that's what i'd expect to be the quality level we're going for...or no?

i'm sorry

" ...when we still talked about a "museum":
> It should, for example, contain a long aisle that forks into separate rooms... "

haven't read the old forum yet will do that now though, thank you.
[PS3 graphics right there. ]..should be possible with the Ca3D-Engine as well! right?
Kai wrote:Hi , nice to see someone who is also interested in working with Ca3D

First of all it would be nice if you could reduce you "euphoric" way of posting, there is an edit button you could use instead of posting again and again just one or two scentences

The level of quality is depending on the artist who is contributing the stuff like textures, model and things like that. Unreal 3 Engine is way above Ca3D no doubt about that, but it also requires the latest GFX Card to look like that, an so far some people not even have a DX9 capable card with PS 2.0 specs or above.
Lets say Ca3D can look like Doom3 or Quake4 and better (or different, since we also have static lights and very huge terrain support)
So if you want to let Ca3d look like the quality of Unreal 3 its up to you if you can do that
And again this quality is somehow limited to the gfx card of the user, and it is not useful to create a techdemo that looks stunning but only on 5% of all user.

Those killzone screenshots are a fake, the are not using realtime hardware since it is announced that Sony still has no devloperkit. And i heard that a guy who worked at this viedo said this was a promo video for Killzone 3 (and those shots are promoted as Killzone2 ^^)

Don't get me wrong, i would love to see a game like that but so far this is just a nice looking prerendered video, that might run like that on the theoretic Ps3 hardware.

So i dont think we can do something like that since no engine or gfx Card is capable of such a massive poly (did you find any polygon on a silhuette those must be extremly high tessalated ? ) and effect count.

My tip: Those models are the highres version for the Nmap that will be used on a much lower polygon
levaravel wrote: hi MDKai,

yes, i will reduce the posting..i just got kinda excited today with ideas..uhm yeah. okay

It won't be a problem to produce this high quality content for me. creating levels that look like 100'000 polys, but that are in fact only like 10k. baking the textures and using regular bump maps, it would run on older systems as well. I would like to start modelling and texturing that stuff, but i'm not sure about what carsten wants.

well if you want, i can make a sample room in the final quality?

what experience do you guys have with real-time 3d and modelling / texturing / animating?

UPDATE 23. July
2 Room Screenshots
Carsten wrote: Hi Levaravel,

while indeed the "specs" of the TechDemo indoor building are not quite fixed yet, I hope that we together will be able to do so soon.

The Unreal 3 Screenshots, as well as those Doom3-alikes that I sent you recently (Kai posted the link in the chat recently, but I'm currently on the wrong computer and don't remeber the URL), plus the "old" specs at the link to the old forum that I posted above (http://www.ca3d-engine.de/phpBB2/viewtopic.php?t=75) certainly indicate the right direction for the indoor part.

About your question about models for terrains: I'm not quite sure if I understand your question correctly. In any case, using model-meshes for terrains is not possible. It's also not reasonable, as the great algorithms for LoD terrain rendering require terrain definitions via heightmaps, not meshes.

As I don't know how much work is involved in making a room (preview) in full quality, I can hardly ask to you do one.

Please see my other thread about the TechDemo about what might be done next.

My personal experience with professional 3D modelling (3DS, Maya, LightWave, etc.) is little to nonexistant, but I feel comfortable with CaWE editing.
levaravel wrote: Hi Carsten,

That's a good idea, we should write down a todo list now and assign jobs to all the guys that are helping out.

the terrain-question... okay, i thought it's possible to have an actual mesh as landscape, which also uses an LOD algorithem, to optimize it.

... "As I don't know how much work is involved in making a room (preview) in full quality, I can hardly ask to you do one. "

- allready working on one in final quality. don't be shocked. this is just the high poly model to render it into a texture later. currently it's also not textured, just raw geometry that i'm modelling. texturing comes after that. allright, here the two pages

Room Pictures 1

Room Pictures 2

i hope to hear some comments or suggestions for the room. any ideas for some stuff on the walls? (i thought about 3 lamps on each side, lighting at the room)

*back to work*

Stuff to be done:

-Concept Design / Artwork
-Rough Layout
-Modelling / Texturing / Mesh Animations
-Level Assembly

Do you have an list of active artists, that are available, to start this project very soon?

There should be an FTP server where every artist can map a drivename to, so everybody is working from the same files. with strict filenaming conventions like:

modelname_001.mdl, modelname_002.mdl ..and ALWAYS, the latest revision is the model to be used. that way everybody knows what is going on, and how stuff is being organized.same thing for the world edit.

techdemolevel_001 techdemolevel_002..and so on. organization is a very important thing, otherwise, after a few weeks, maybe even days, if there are so many people working on this, there's just gonna be a chaos. everybody will use different filenamingsconventions, different foldernamings and we would get confused, and would spend more time trying to figure out which file to open, instead of working on the actual project...

just a thought.

*recommending NetDrive for drive mapping*
Carsten wrote: Hi Levaravel,

don't worry about the terrains: The LoD algorithm for heightmaps that I employ does a much better job for terrains than any generic mesh LoD algorithm could do.

I like the two shots of your room, although I think that the big chain across the room looks a bit old-fashioned, not like something you'd expect in a high-tech environment.
However, I too would be interested in and wait for comments from others, who probably have a less biased view than I have.

Levaravel wrote: ...
-Concept Design / Artwork
-Rough Layout
-Modelling / Texturing / Mesh Animations
-Level Assembly

Sure sounds good.
Levaravel wrote: ...
Do you have an list of active artists, that are available, to start this project very soon?

Those who help are all volunteers, and this is also why everything is pretty loosely organized. Kai has provided great artwork and help in the past, but I can only ask you, him, and others to join.
From my point of view, best situation is if somebody says "I'd like to do XY, and I'll do it."

Levaravel wrote: ...
There should be an FTP server where every artist can map a drivename to, so everybody is working from the same files. with strict filenaming conventions like:

What you describe is a sort of version control.
One problem with your approach is: What happens if two people work on the same file? (And its a binary file that cannot be merged with the file by the other person?)

Now, there are some good solutions for this. The best, imho, is Subversion.
The Subversion version control system essentially solves all the problems that you mentioned above. It can even handle situations when many people modify the same file, but only under the assumption that the file is a text file (e.g. C++ source code), not binary.
Working with many persons on binary files requires that the person that currently edits the file locks the file, so that nobody else can modify it at the same time (or at least is noticed about the lock). When the 1st person has finished editing, he submits his changes for the others, and releases the lock. Then somebody else may lock.

I would certainly be happy if you all used Subversion, see http://subversion.tigris.org/ and http://svnbook.red-bean.com/ for details. Btw, I use Subversion internally for all my development work. However, I was really hoping that we could avoid this for creating the TechDemo (except for you all fall in love with subversion now ). We might rather employ simple organisatoric measures, like "Carsten does the Terrain, Levaravel the Concept Art, Kai the Textures, etc.". This way there would be no problems either.
levaravel wrote: Hi everyone.

Last night, when i was watching a movie, i had another idea for the demo: 2 Demos in one. 1 Demo like a cinematic action movie, with automatic camera movement and no user control, just eyecandy. And the other demo, the one with full user controls, like in the game. (the actual tech-demo).

I do like the thing coty wrote in the other thread (direct importers for maya / 3ds / lwo files, for use as terrains, characters, entitys. or even as whole levels. since in maya / 3dsmax and lightwave you can make the whole scene complete with lighting, all the characters in position and so on...

even though it would be a huge amount of work to code that into the engine, it would be woth it in every way. Artists are then able to create theyr levels and design them in theyr favorite program, import it into Ca3DE and play it!

Would there be time to programm that carsten?

Thanks for the suggestions so far, carsten.

yeah orginazation should be like that. whoever is best at something should do it (and what he likes or course).

The idea with the ftp server was, that every artist has mapped a drive on his pc, like T: (which is actually a location on a remote server). so everybody is working from the same "network drive".
Kai wrote:Hi

I really like the idea of a webserver for all the artwork content.. im also interesting in using subseven, however i forgot how to use it XD (used it quite a while ago for a project with my friends)
But with an organized work we can provide some sort of a pool that holds exclusive maps ,materials, objects and more just for the developer team.
So maybe we can come up with a good solution how to and who host such a system ?

Oh yeah i have many years of experience with Max and Photoshop, i also had some time using Maya, C4D and Lightwave..
But my personal favorit still is Max.
Carsten wrote:Great to hear that. I'm already operating a public Subversion server, too. It's just a home PC with DSL connection and DynDNS service, currently hosting the public CaWE development.

I hope that its internet connection is good (fast) enough for the TechDemo binaries, and will prepare it for access for you.
I may also check-in the latest CaWE and Ca3DE executables, so that you could always easily get the latest.
Thrawn wrote:Hi levaravel,
it's very nice to see a new face in the community that it that active and talented.

I like your ideas and your work, but there is one point I wonder about.

The levels in Ca3DE should be build with the mapping editor,CaWE itself, and not with models, thats not the way Ca3DE normally works (although it might be possible!? Static Detail Models have no collision but planting clip brushes in the editor could solve it).

So if your idea works - then maybe you can Kai would have to model the techdemo level in your modelling editors - if not, somebody has to map it with CaWE and both of you have to do content (models, textures, concepts etc).

Just wanted to mention it before you model a whole level - We first need to know how to do it - anyways, your map models are at least great concept arts...
Coty wrote: levaravel wrote: ...
I do like the thing coty wrote in the other thread (direct importers for maya / 3ds / lwo files, for use as terrains, characters, entitys. or even as whole levels. since in maya / 3dsmax and lightwave you can make the whole scene complete with lighting, all the characters in position and so on...


The problem lies in the SUBTRACTION (REMOVAL) of unseen brushes when processor is building the BSP. Why do we need this? Why not tell the processor NOT to remove any brushes of detail level 1, or detail TRUE, or in ID editors, "func_detail" brushes ( I think). Now any brush marked for detail would need to have the unseen faces removed by the level artist and not the BSP compiler. The other problem is processor trying to SNAP brushes to the grid, or snap brushes together to form a solid block or something. Obviously this wouldn't work with single sided brushes. We need to tell processor not to try to snap the faces of brushes together that are Marked "no snap" in their properties. The detail 0, or detail FALSE, or func_wall in "ID Software" (I think) can still be made to do their thing normally by sealing the level and creating vis blockers (hulls for vis blocking). And the detail TRUE, NoSnap TRUE brushes can be you geometry imported from your modeling program. I think you do this somewhat already by allowing actors, or models, or something like that, which you have made in your modeling program that you have binded to some sort of object entity. But I'm getting a little off topic and don't mean to confuse you.

Let's say I add a command to the BSP settings: -nosnapnocsg and processor doesn't try to snap or remove any brushes that have detail TRUE and NoSnap TRUE in their properties. Remove unseen faces of detail 0 (func_wall) that seal the map and block vis (the normal way), but ignore the other brushes as far as snapping and removing go. Calculate the pvs on the detail 0 (func_wall) and ignore detail 1 brushes for blocking vis. You do this anyway.

LoL, then you would have to redo your editor to allow moving a single vertice. Make a new "Mode" in CaWE, geometry mode, where you can select single vertices and move them instead of that face thing you got in brush mode, where moving one vertice moves the other one on the opposite end to keep the brush planar. You don't need this with sigle sided geometry, especially triangles. CTRL-E on a selected brush will split the non-planar face into a planar triangle.

All this is in older versions (Nolf1 AvP2)of DEdit, Lithtech's map editor.

Anyway, something to think about...

Here is a link from my friend Modmaker on this type of level building using DEdit in AvP2: http://www.planetavp.com/modmaker/modmaker50.html
Giant Bit wrote:Hallo all,
I am wondering who is doing what and if at all, since the last post in here is almost a month old.

I just wanted to let you know, that I can make some models or textures. I don't want to make any concept art or even spend my thoughts on what has to be done, but if you decided what
you want and need some additional artist to help, let me know.
And I'll see if I can do it.

(If it is not guns and the like boring things)

I have to add, that if I make a model, someone will have to convert it in ase-format.
Best regards,
Giant Bit
Posts: 1
Joined: 2005-09-21, 17:47

Post by Giant Bit » 2005-09-21, 17:57


Have you given up on it?

(Or is it because I wrote:"---guns and the like boring things---")?

Well, my offer is still valid.

(And then I also want to thank you for the nice welcoming words)

User avatar
Posts: 302
Joined: 2004-08-30, 10:38
Location: NRW, Germany

Post by Thrawn » 2005-09-21, 18:28

Of course, we need everybody and everything we can get. But at the moment there is chaos. But I'm sure that soon everything will be more clear and we then know what we need etc

Please be patient :?
User avatar
Site Admin
Posts: 2169
Joined: 2004-08-19, 13:46
Location: Germany

Post by Carsten » 2005-09-21, 19:11

no, not given up on anything, never.
It's just so that I'm permanently working on the code that I think will be required for the TechDemo, and that leaves little time to work on the TechDemo itself.
So don't worry, things are good, and they're developing good, too. They've also not gotten out of focus.
In fact, there is some good amount of TechDemo-related work going on behind the scenes... (e.g. in the chat, private tests, etc.).
Best regards,
Post Reply

Who is online

Users browsing this forum: No registered users and 1 guest