2009-11 Wednesday 25th, Interview spills a lot of secrets and causes amnesia

[jens]

A short news post about an interview we did yesterday, it gives some insight into our daily life as well as a ton of new information about Amnesia: The Dark Descent.

You'll find the interview over at Linux Gaming News.

[Lee]

I came buckets.

[Klayman]

Nice interview! Lee, why do you have Jens's avatar? Are you twins?

[Lee]

Yeah, why not.

[ROT51]

Good interview. I think Lee prefer Jens. That might explain that avatar. He might be little bit crazy, but thankfully we have at least one funny member in these forums!

[Lee]

I do nights and weekends.

But uh, about the first question of the interview, did you ever actually receive any money for Overture at all?

[jens]

We did, thankfully when we were about to send in the GM we said that "if we don't get that promised payment, you will not get the game" and that at least guaranteed that we got some money. Then also to remember we have sold the game in our own online store and at some other places + it is included with the Collection so in the end it has not been all bad.

[ajvitaly]

So what publishing strategy are you going to use going foward? Will you forgo traditional publishers for your online sales? And does completing most/all of the game self-funded give you more leverage when signing a publishing agreement?

[LargePenguin]

Hi Thomas,

I am interested in this Coherent Hierarchical Culling. Where can I find more information/paper?
is it this: http://www.cg.tuwien.ac.at/research/vr/c...chcull.pdf ?
would be interesting to try it out for a university assignment. In what ways is it better than portal culling?

Do you have your own collision code or do you let newton handle the collision?

[Thomas]

(12-29-2009 09:24 PM)LargePenguin Wrote:  Hi Thomas,

I am interested in this Coherent Hierarchical Culling. Where can I find more information/paper?
is it this: http://www.cg.tuwien.ac.at/research/vr/c...chcull.pdf ?
would be interesting to try it out for a university assignment. In what ways is it better than portal culling?

Do you have your own collision code or do you let newton handle the collision?

Yes, that is it. The algo itself is quite simple and the hardest part is making a good container. For static objects it is fairly simple to do a good kd-tree, but dynamic gets a lot more complex and there is lot of stuff to take into account. Start off with trying to make a static one is my advice

Some pros of the CHC algo are:
- It can be used on just about any sort of environment, but when using portals it only works good in a corridor-room sort of map.
- Even transparent (textures with "holes") textures can add to the culling. So a lot shrubbery will eventually cover to occlude objects.
- No artist set up is needed (lots of this with portals) so it is very simple to work with.
- No long build times like with bsp trees (Except if the object container building is very slow).

Some cons:
- It will never be as fast or as good as something made by a human. With user placed portals there is a lot of tweaking that can be to optimize (remember this also means a lot of hard work)
- When there is too little culled and too occlusion objects checked, it can slow down the rendering instead of speeding it up. There is a lot of tweaking to be done here though and this as a bit to do with how the container is constructed and set up. In other scenes, like viewing an entire city from above where there there are a lot of objects, but little occlusion, some tricks can used.

There is ALOT of stuff to dig into here so I think it would be prefect to do as a university assignment! I spent quite some time with this, but have lots of more things to research and refine.

As for collision, yes, newton is used!