This discussion started in the "H.A. Suggestions Box" thread.

The idea is to try something new: There are game engines out there for creating real-time 3D games and I'd like to evaluate at least one for a small sexy game. The first try should be of a limited scope so that the project can be done in reasonable time and effort. Ideally the result should be an expandable engine which can be used for more complex games and evolve with each game.
I'm posting this in Shark's lagoon because I hope to excite Shark and others. In case we can create something usable I hope Shark's lagoon could be the place to offer those game(s) to the public. I cannot make any promises what so ever about the success because I never tried anything with that level of complexity in regard to animations. Typical real-time 3D are less focused on details and offer a small set of animations. On the other side we can simplify things because we don't need a large virtual world, terrain or networking.

What exactly would be needed for such an engine?
-everything you expect from a 3D game engine (scene graph, input event system, easy to use model and animation support, collision detection, ray casting for mouse selection, sound support)
-scrïpting engine for fast development, event driven approach
-models with either lots of animation sequences or even something like animation blending à la cal3d
-mount points on models to attach other objects like clothes
-some AI scrïpts to make models walk in the scene
-a basic game scrïpt that supports global states to create a story line
-object models and scenes (terrain is not so important, concentrate on indoor environment)

The result may not be the high end of current virtual 3D experience but typically free games point in new ideas, niche gaming fields and user support. Without commercial pressure there is more room for experiments.

I think Panda3D could be the base of such a beast. The original site is down at the moment but there is a limited mirror at http://g33xnexus.com/panda3d/index.html

Edit: The original site is up again: http://panda3d.org/

The critical point is a collection of animated models. To evaluate the effort that has to be put into that I plan to get a very small test program out which can load a model and animate it. I have created some static objects and terrain in the past and got some experience with Python scrïpting but I cannot help with organic objects like human characters, fauna or flora. I tried several times before and failed miserably. It just doesn't look right whatever I try.

Python?
Battlefield 2 relies heavily on python scrïpts so if you end up using that I could help proofreading the code. An extra pair of eyes on the code is often all it takes to detect that silly spelling error that breaks the whole scrïpt.
Beyond that I can only stand and cheer you on... Sorry, my skills are simply not good enough with either 3DSmax or Maya.
Heh, in fact I failed making the tutorial cube for both of those! LOL

What about Sharks existing models? Can't they be mounted and rigged for animation?

Shark already gave us the answer: The models he used for the flash games are too highpoly for a 3D engine based game.
In short that means that models with very much detail and thus have a high polygon count are perfect for still rendering, e.g. render to a picture where it doesn't matter how long the render task takes. But a 3D game renders objects in real time several times a second. Typically one should get at least 30 frames per second which is roughly the threshold where a human eye is tricked to see a smooth motion instead of single pictures. That means that you have to limit the number of surfaces and complexity to accomplish this performance. For a real-time 3D textures matter more than vertex count.
In this case I'm not yet sure how the effect will be because I think the environment will be very limited, i.e. indoor and thus no large terrain. It might be that the problem of high poly model is less recognizable in this case. From what I've seen in other approaches the terrain takes up the majority of the vertices in a scene.

So the characters would be low poly , and simplwe terrains mostly indoor or could we cheat with simple outdoor rooms that offer little interaction with the environs. Maybe a rock or a tree or a river/pool with the rest being a backdrop?

Maddog, to be honest I don't know yet. My previous tests with different engines had a huge terrain so this is the first time for me targeting a very small environment. I'll first try without any tricks and see how far we can get.

This is a very first test digging into the Panda3d engine:
http://www.duffy1.franken.de/test1.rar
Beware: This is not a game yet - far from that. It should just give an impression how Panda3d can be scrïpted. The test1.rar just contains the scrïpts and two meshes. This not (yet) a complete deployment. To actually run it you need the Panda3d engine from http://www.panda3d.org/ installed and then run: python test1.py in a command window.
Controls are: cursor keys to move, mouse wheel to zoom the camera and ESC to quit.

I made some progress in regard to lights, animation and mouse picking.
Could somebody please try to use the Panda3d exporter with one of the 3ds models and convert it into the "egg" format?
An exporter for Max3d is included in the Panda3d release and can be found at http://www.panda3d.org/download.php

Nobody wants to try the exporter with shark's models? Or maybe nobody has a Max3D license? Or nobody understands what I'm talking about?

well i know a few things about 3d and games

sharks models are high poly ( but the textures are actually fairly low ) .. some tricks games use are as follows
low poly models with high qual textures - this tricks the eye , but you do end up with invisible bounding boxes and sometimes 'clunky' movement.
baked in - or pre rendered scenes , this gives high quality , but not allot of player interaction or if the player has decent interaction then it's very limited.

the best is of course med poly models with high textures and semi pre-rendered scenes .. if you know who illusion games is , you know what type of result that can do ( artificial girl is one of their games ).
Craban i don't have max3d to look into what your talking about but i do work in blender , poser and carrara regularly , so obj files i can play with , it's the bones and bvh files that get weird from program to program.

Just found a price re 3DMAX, I think on an Australian outlet, $3495!

Until recently I hadn't realised that Shark was not using Anim8or, but rather using it as a referral to view his models.

Unless somebody gives me a copy of 3DMAX I'm going to try to understand Animator and leave Shark, RKS, Pusooy, etc to develop whatever ideas we are to offer them.

Anim8or is very good software to animate 3d models. 3dsmax allows more realistic pictures but that is not necessary to make a flash game

lucindrea: Would you volunteer to create a test model? There is an art pipeline from Blender to Panda3D: http://www.panda3d.org/wiki/index.php/C ... om_Blender
It doesn't need to be a masterpiece - a simple one with maybe one or two animations is good enough for the first tests. Maybe you have something laying around waiting to be put into the above mentioned test scenario.

Re two posts up:
One could not take one of your models into Anim8or, impose 'bones' and make a movie? (Other factors considered of course.)

I notice that your "poly" structure seems to be different from the Hand tutorial at that site, or is that because of the level of "sophistication" that the programs offer?

From some advents with AutoDesk I realise that Anim8or may well never compete

This may be a little off topic, but this link is for free 3D software called Alice from Carnegie Melon in Pittsburgh, PA, USA. I read a little about it and I think it was designed to make 3D environments and games easier to create. The link is this:
http://www.alice.org/

Hi! I working with Poser.
If extract the Poser scene to 3DS Max format, then can i importing in Panda3D or Alice?