Re: question to Alon or Flyby - XCC ForumRegister | Login | Search
History | Home | Messages

ok i got the compilerViPr12:48 20-05-2002
question to AlonViPr07:21 21-06-2002
Re: question to AlonAlon19:16 21-06-2002
question to Alon or FlybyViPr20:41 21-06-2002
Re: question to Alon or FlybyAlon20:53 23-06-2002
Re: question to Alon or FlybyAlon21:15 23-06-2002
Re: question to Alon or FlybyViPr10:04 24-06-2002
Re: question to Alon or FlybyAlon23:22 27-06-2002

> > > > > > i never understood your technique to convert max models to 3ds in such a way that the meshes don't get all misplaced. i just want you to tell me if your technique involves having to reposition the meshes very carefully. in other words is it possible that somebody using your technique could carelessly put the mesh in a position which is better than before but still slightly off from where it should be?

> > > > > > i'd say about half the models i've seen on the Sci Fi 3d site have this mesh problem.

> > > > > It's not "changing the position" at all, you see, what I really did was to use some functions in MAX to help and kinda "collapse" my objects into pure 3D-geometry code...
> > > > > I think The "bug" is because there are many different (and strange?) settings related to object positioning that your program does not support. And you always say the documents you read about the 3DS format were inaccurate (in the least..) ....

> > > > ok thanks, that's good. that makes me happy that modelers can easily avoid the problem, but still it's a pity i can't figure out how to read those weird models properly. it affects way too many models.

> > > > ok another question. i've noticed when some models have cylinders on them, the reflection on the cylinders looks kinda slanted or skewed if you wanna call it that. i know why this happens in my program; it is because they make the cylinder using lots of rectangles divided into triangles with diagonals that all go in the same direction instead of alternating the direction in other words they do:

> > > > like this
> > > > |/|/|/|/|/|/|

> > > > or like this
> > > > |\|\|\|\|\|\|

> > > > instead of like this
> > > > |/|\|/|\|/|\|

> > > > is this something known among artists that you should remember to alternate the diagonals to avoid this normals problem on cylinders? i'm wondering if this problem happens in 3d studio max or other programs as well as my program.


> > > Well....
> > > I don't think that if 2 adj. polygons have the same facing like in a cylinder (or at least a cyl. in Max), the thing you said matters... But then again I almost never do reflections :)

> you don't need real reflections. specular reflection or environment mapping will do. it's a normals problem.

> > > Hmm.. You know what, let me try and I'll get back 2 you.

> > Well it changes a little, but I can't create a pretty-enough scene to fully test it.. :p:

> you should notice it easier the more slanted the diagonal has to be, like if the cylinder is short or there are few facets in the cylinder.

> actually some programs may recognize that the facets belong to a cylinder and will override the regular way of finding normals on them and do the normals specially for cylinders.

> btw are you guys fans of the SCI FI 3D coz it would be easier to talk about some problems i find. anyway if you are, have you noticed some models there seem to have smoothing problems? for example there are ways to stop adjacent triangles smoothing together:

> 1. triangles have smoothing group 0. this means smooth with nothing.

> 2. adjacent triangles have different smoothing group numbers.

> 3. adjacent triangles don't share the same vertices but their own vertices are in the same exact spot making you think the multiple vertices are a single vertice.

> anyway have you noticed number 3 happen at all? i think some models there have that.


> anyway in other news, i've added code now to stop certain calculations when whole models or their individual meshes are not on screen. it should speed things up when you are zoomed in to a model or generally just looking at a part of a model only. it should especially help on environment models where you are usually inside the models. of course those are not needed for voxelization but it's nice to let you view them anyway. maybe it will help you find more bugs for me to fix with my program in general.

> this algorithm should give a speed boost simlar to BSP trees or quadtrees or octrees if i understand correctly, except those usually have more stages in their hierarchy than just 2. it's not gonna be as good as those tree things but i don't like them except i think the quadtree one would work nicely with terrain that is grid based, otherwise if you use it on arbitrary models, i believe it has to chop some triangles into smaller pieces when they straddle zones. that's just too messy. i like portal technology, i think i know how to do it, but models have to be specially designed with it. i'm wondering if you can tell me whether people can build models with portals in 3d studio max.

> i have no idea how the visibility thing in Quake and Half-life works and i think it's better i don't know or i will go crazy. does anyone know if Quake3 still uses it?

Yo ViPr !

"Vertice" is not even a word!

The single for vertices is VERTEX, for example:
1 vertex
2 vertices

So there is no word such as "vertice"

(well at least not in English, there is such a word in Italian and it means zenith. But this isn't an Italian forum!)



Home | Post | Users | Messages