What does the extra power in the new consoles mean for Collision detection?

Hi there ... long time lurker, first time poster. I'm not a dev of any sort, but do enjoy reading all the where's and howto's and the insights that are discussed on here ..

and I was just wondering how much programming resources are needed for a good collision detection system ... be it in a racing game or a battle in Skyrim or a sneak kill in Hitman Absolution ...

Can collision detection be enhanced now that we have more processing power? LIke .. could we have extra points on a model that can be assigned as a hot spot for collisioin which could lead to new aminations for said spot? (if you know what I mean .. apologies for my layman speak)
 
First of all welcome guy!.

Well, Desctruction Derby in the PS1 already had great collision detection and visual damage. GT5 has a pitty visual damage 12 years later. The question is more if devs will get something like collision detection between their development priorities ( the same happen with AI ). For example Skyrim fighting system is not a world different appart from Morrowind, and this for sure is also a design priority issue.

Although, to be clear, the more power, the more vertices you can process and so, the better should be the collision detection, animations, physics, and so on... the most i am looking for is better hit detection/animations in shooters.
 
Last edited by a moderator:
I have a gut feeling better colision will become more important this gen, because of physics based effects, animation and destruction. I could be wrong, but physx style stuff might be one of the next big diferentiators of this gen from the past, and the more detailed your models and their animation are, the mored detailed their colision meshes and physics have to become too so that they don't seem odd.
Driving games for one, have become so close to photorealism this gen, improving car and track destruction might start to provide a higher visual leap than enhancing models and rendering.
That's just my speculation though.
 
I don't think collision detection is going to improve all that much. It's probably still going to be relatively simple bounding boxes with a few games bothering to distinguish between parts of an object and then whole object.

It'd be good and bad if bounding boxes actually conformed to the actual shape and volume of the object, but that would computationally expensive with some drawbacks. The main drawback (especially on console) is that in something like a shooter, a person now has to be much more accurate with shooting.

Then again most console games have aim assist at which point does it matter if the bounding box for collision detection is more accurate? At which point it becomes rather meaningless to increase the accuracy of the bounding box with regards to the object model. For a PC game it might actually matter (where aim assist is non-existant or usually off by default for some console ports).

Regards,
SB
 
I worry that collision/clipping discrepancies may actually get worse with highly displaced tessellated surfaces.
 
i just hope its the end of invisible wall... no more collision with imaginary wall while happily moving around.
 
Sports games will hopefully take a step forward. Maybe NBA 2k14 and NHL 14 or 15 could surprise us :)
 
You make me sad, I was hoping for a full earth sandbox. That'll fit on blu-ray right?

I'm working on a full earth sandbox right now. Fits on 8 DVD's though, and the detail on the ground isn't exactly COD quality, but when you are moving at 2000km/h you don't notice so much :)
 
i just hope its the end of invisible wall... no more collision with imaginary wall while happily moving around.

^This ^ is what started me thinking on this initially ... will we see characters still walk-on-the-spot while he/she/it hits the boundary of the game environment for instance ... (or a table)

Also .. would cloth animation be enhanced with included collision animation against the body for example?
 
Collision detection isn't the big problem, incorporating collision information in animation is the problem ... especially for parts of the scene which aren't really independently animated at all (clothes and hair being the major offenders). Even for human bodies smoothly and realistically incorporating the collision information is far from trivial.
 
Back
Top