pouët.net

http://en.wikipedia.org/wiki/OpenSceneGraph

Hey reading the latest post on OpenGL 3.0, I stumbled upon this library.
Has anyone had GOOD thoughts or experiences on this lib?
Any GOOD demos made with it yet?

I know this also arises the question : how much lib do you want in your demo.
So..lib- and nonlib- fanatics, please promote your personal taste on this subject!
Show me some pouet emotions!

as tom forsyth would put it "scenegraphs suck" ;)

Personally I used scenegraphs a lot, but I never liked OSG, NVSG, etc.
Recently I switched to make my engine non-scenegraph-based but still have the sg in the importer. seems to be the best trade-off for my purposes..

I believe The Flaming Marshmallows used OSG in their latest production.

Personally I think it is a good thing to use libraries like these. There is really no need to reinvent the wheel over and over again. Ironically enough, I have not yet had the will to use OSG, mostly because I am afraid that simple effects require more effort when using a general purpose scene renderer.

Can't find any Flaming Marshmallows on pouët. Care to link?

my personal opinion is, that scenegraphs are too bloated for most projects you come across. for democoding I never had the feeling, that a scenegraph would make things alot better/easier.
concerning OSG: I didn't like it back when I had to use it due to the total lack of documentation and for what I had to use it it was way too high-level. or I didn't find the proper ways to do what I wanted to do due to the already mentioned lack of documentation :)
but things might have become better in the meantime.
anyway, I still think that using OSG in a demo is wasted energy on adapting to it's concept and the way it works.

While looking at http://www.openscenegraph.org/projects/osg/wiki/Screenshots one thing is clear: "VR" = "shitty looking engine with no features above flatshading what so ever"

It just brings up the point one more time: Linux-enthusiasts only want to build the engine, but have no idea (or desire) on how to make it look good. :)

Lord Graga: The link is here, http://pouet.net/prod.php?which=50490

Gloom: A lack of artists does not really equate to "shitty looking engine with no features above flatshading what so ever". OSG has more features than most demo engines, it is just a matter of using them properly.

Thanks for proving my point so explicitly. :)

Gloom: I am confused, what is your point exactly?

Havoc: That is what artists are for?

by the looks of it, apparently not.

hmmmm...thnx guys, after reading some post and doing some research...I think it depends on what your after.

I really can imagine these Graph-thingie tie you down too much...I mean it is well structured..but if you're after a simple effect it really can become too much overhead.

However, if you want to achieve something which requires a LOT of work on several domains (like games)...you may want to consider using some 'standards' at some of these domains. Then the change of succeeding will be greater and more fun.
How much projects will end up nowhere because the idea was too big, and the intention was too 'I want to code all by myself'?

Personally I can say...I don't know really why I should use OpenSceneGraph for my little demoexperiments. But I can imagine people who want a GOOD scenegraph would use a 'standard' like this.

Excellent logic.

(That was meant for Havoc)

bruce: I am confused, what is your point exactly?

Havoc, the point is, the screen shots look ugly because it is obviously coder art and not because OSG lacks important features that would make it impossible to make something that looks good.

Bruce: The screenshots are an overall reflection on the engine. If there are no good screenshots, obviously nobody using the engine has any interest in making something look good - hence my comment about Linux stuff. This all stems from a long discussion in a different thread about why there are no Linux demos, and if there are, they all look like crap. :)

It's like if you have a motion graphic "artist" that spends more time and energy on coding custom PHP-scripts for his homepage than doing actual motion graphics.. which is pretty much the impression a lot of people have of Linux demosceners (and Linux-users in general I guess). A lot of noise, no real demos. :)

can you suggest any visual examples that underline your theory, or is it purely hypothetical?

(that was for bruce btw)

hmm, looks pretty nice but i'm not sure i get it.. a scene graph is some sort of graph data structure that describes a 3d world, no? so in a city, every node can be a house, which has child nodes for walls, which have childs for windows, doors and a chimney, no? or does it completely work different? how is the graph related to the actual meshes? :-)

(the osg site has all kinds of tutorials but i've not managed to find a "for dummies" introtalk to the scene graph concept itself..)

cause if it is indeed like this, then it sounds like a really cool way to generate cool objects/shapes/worlds procedurally without some of the lower-level hassle.

Gloom: The screen shots show what people made using the engine and in no way reflect the full potential of OSG. Sure they look crap, but that does not make it impossible or very hard to create something decent using the engine.

i have the idea that a certain somebody in this thread is currently doing just that!