pouët.net

kusma: yeah i dont think it shouldve changed much right? anyway lets see what smash knows.



this is of interest to me as well.

or to quote alucard when addressing the friendly merchant in the library in symphony of the night:

"i'm interested in this"

the lightwave layout of files hasn't changed. there are usually 1 lws and multiple lwo's.

Yes, LWS and multiple LWOs. Altho you can put multiple objects in one LWO (multiple layers).
It's pretty easy to derive all the connectivity information you need in LWO. You dont usually get this info in other formats either.. :)

thanks, i've got my resource types ready then.

lwo2 lacks some 3dsmax modifier stack or some abstract geometry definition, but is my favorite for mesh & spline curves

and here I just simply forget to make this excellent joke, KUSMA, YOU *ARE* 10 YEARS AGO!!

how much time gets wasted reinventing all this shit over and over? :)

how many people have written a config dialog in the past 15 years? thats just how it is.

psonice: don't think of it as reinventing, think of it as an extra learning lesson for the coder. i never really understood the multiple approaches to string manipulation until rewriting all those routines (e.g. printf, cout/cin, strcpy, strcat, ...) myself.

I used LWS for my 99-03 demos and 64k's. A great thing about LWS in the context of 64k's is that you can make a object generator,etc output LWO files and then keyframe them or use magically named objects to control stuff like particle systems. Then when building your 64k you take in the keyframing, the generation information and ignore all those dense meshes.

I can certainly see the point of collada sucking as an "interchange" format since the tools work at such different levels internally.

However, Collada is enormously flexible (yeah probably too much so to be an interchange?) for defining how to define the data, if you just take it in and render "as is".. does it becomes that problematic still? (Don't people writing exporters test their output with some generic viewers?)

As with any system that is enormously flexible, it usually also gets excessively complicated :) I'm not saying that it's impossible to create a codebase/project that pulls it off without completely clouding the usability but Collada sure isn't one of them.

LightWave looks sweet enough for me, for now.

I'm using AssImp to load Collada without having to deal with all the complicated stuff. It's a pretty comfortable abstraction, and it supports other formats as well. Not that I care about the latter point that much.

Good one, I'll have a look (despite it's unfortunate name).

Come on, the name kicks Ass ;)

thank you for a much needed laugh.. nice detailing on that joke sir.

It's a dirty job, but somebody has to do it.

i was wondering about assimp, although it makes me nervous because youd probably have to do a fair bit of work to find out how well it actually supports those formats it claims to support.
and its opensource, which probably means nobody is sufficiently invested in it to *make* it work properly. :)

anyone have any practical experience of it?

It doesn't happen much but Smash, you fall short twofold here. Firstly, it's called AssImp (proper capitalization please) and secondly Kusma seems to have some experience? :)

I'm an Ass.

plek: practical experience in my case means using it for import of a full scene with animated objects, bones, morphs and so on. :) kusma, how far you pushed it?

Okay fair enough. Damnit :) I'm curious as well btw.

smash: I've only been using it for Collada, but that seems to work pretty well for my my use-cases. The AssImp-developers were friendly and helpful to me when I had problems building without Boost. They provided a proper patch within a couple of days after I sent them my work-around patch.

But I can't say I have battle-tested this properly. I haven't used it for any released stuff yet.

Smash: I haven't tested bones or morphs yet. Just static meshes with transform-hierarchies and some simple animations. Worked fine for that, though.

Smash: By the way, you can easily test it yourself by downloading the AssImpViewer-thingie. That way you don't have to spend time figuring out the interfaces and whatnot if it's fundamentally flawed.

i was looking at collada these days, and i was tempted to roll my own parser

don't know if it's a bad habit from my part; but i just like to code it myself when it's pet programming; same goes for containers instead of using stl vectors

though, i'm spending a fairly amount of time doing so; hence building an engine/framework instead of coding a demo (and those who i mostly talk to already know that :p)