Jump to content

flachdrache

Members
  • Posts

    397
  • Joined

  • Last visited

Posts posted by flachdrache

  1. Well, that might was poorly worded - rick suggests pure object orientation with the camera being an object, the "only" other option would be self aware/containing objects ( with a own camera using disable old/ enable new ). Now, in le2 framework I "must" use the later because the ears ( listener ) are at the main cameras position "always" - fobaring positional audio e.g. light humming etc ...

     

    that's was what I wanted to point out - if you want to have it extensible ... or as a base etc.

     

    for beginner : sorry'bout all the "" ... le2 programmer know ( or should remember ) all the little issues we had ... and still have tbh.

  2. That's correct - its just about interaction and defining rules. The mover just follows a path along an origin = the axis center point. The mover does not need collisions because the lift comes in as complete, extern, model with parent child relation.

     

    However - You ask for and josh suggests so called "brush models" ... which is valid too.

     

    PS: such mover is just a point in space like 1x1 or how small you can go ... but in such case the extern model needs to be positioned to such point correctly and might have an offset.

  3. Well its a depth based shader AFAIK and has been (tried) done before but the terrain was too smooth to look good.

    Some user might revive it, if you status update for it ... but tbh its well known and there are free and legal versions 'out there'.

     

    However I am not familiar with current state of post processing and the foliage seams to be a 'special case' e.g. second pass render.

  4. All levels can't be big an blocky like HL.
    

     

    should not but must still biggrin.png ... you cant expect that random sticks and stones do give you a desired result i am afraid. Navmesh extractors aint that good (currently afaik). Thats an issue i had with eki one and tree models ... and clouds of navmesh in the leaves:d. Rather pointless to build AI without a final map if you ask me.

     

    For "Le3" i think it would be good to suggest josh - to have a special material for CSG which can be used to block-out your level. Its nothing more then a collider for your AI agents at the end and details just slowing and bugging the navmesh extractor. Having ai-collider materials is what q3a uses for its area-awareness-system.

     

    PS: not just an id fanboy - the terrain in my xna engine used a terrain processor with a virtual material ( no texture ) for its collision mesh too.

  5. There always used to be better, free alternatives to the "standard" - don't recall the name though ... however, if you skim through the 3.x forum over at pixologic (the older, mate painting topics) you'll see that this was always possible with zbrush.

  6. No, no - I am a fan of this ... however

     

    Self shadow bump map - usually a height map gets ray marched or texture3d is used e.g. might be a user case

    PRT - same e.g. user case

    GI and physics - currently the holy grail of high end engines ... for a reason ( toasting client pc`s )

    Reflections - all tech has different tradeoffs ... usually projection is "enough"

  7. Basically 2 sets of tools - one for 3d models and one for 2d e.g. uvw mapping ... the rest is edges, vertices and faces you use the tool with. Learn the shortcuts like r, f, p for 3d window navigation and u to switch to UV window. Setting up the side pannels to show the tools you'll need also helps e.g. draw style ( flat, smooth with texture or not etc), scale, move etc. etc.

  8. Beside the driver issue with 2.5x I had a very strange behavior lately.

     

    I created a portal mirror ( like the q3a one ) - I create a camera on surface ( its a plane ) which is looking from a orgin towards a target ( works as mirror and portal cam ).

     

    I need to switch main camera with a portal camera because of the listener ( which would be re positioned if I just would move it in framework )... plus I have a camera render buffer to paint the camera image on surface.

     

    The buffer is local to object and created like CreateBuffer(..., 1+2+4) iirc.

     

    I had no issues with that - I changed to only use color buffer.

     

    Yesterday it failed because of that ... so, how can such ever happen ?

     

    Please note - the issue is not that I messed up my code - the issue is that the engine doesn't realized the fact. Does structures be keept in memory even if the host gets "destroid" ? How to clean/overwrite memory ( and that is hilarious ) if such can possible be the case ?

  9. Hmmm - to that point we are pretty much even in progress I like however that we differ in game design ( I use small terrains with local hubs and a pretty nasty environment ).

    If its not a completely dead world I suggest to use some events as eye opener like distant jets passing, stars falling ( guess you know what I mean ) etc.

    The little rockwall looks out of place and I had the intention that 2.5 is capable of vertex painting the road ... they look to clean IMHO. Looking forward to see your ai - I didn't even really started :D.

     

    congrats so fahr - quit an archivement :)

×
×
  • Create New...