Jump to content

  • Log In with Google      Sign In   
  • Create Account


Member Since 03 Nov 2009
Offline Last Active Dec 25 2014 12:06 AM

Posts I've Made

In Topic: SSAO very dark

20 November 2014 - 03:14 AM

Is there a way you can calculate smooth normals per pixel before applying SSAO? Or getting additional info from a vertex shader so you know which vertices are shared and therefore can be used to smooth normals? The problem with smoothing the depth is that completely disconnected objects get smeared together. Another possiblity is having some cutoff for smoothing, e.g. if an edge has too great a depth difference do not smooth it. I'm not sure how easy that would be in a shader - I'd normally avoid too many conditionals.

In Topic: Gameplay For "Split-Reality" Game World?

19 November 2014 - 03:51 AM

A few thoughts:

  • If the dystopian world is the real one, and moving through the utopian world causes you to move through the dystopian world too, how do you not end up dead from hazards that you can't see?
  • If it's more like the Matrix, then the utopian world could represent some version of the real world as it was or could be. The value gained from the utopia could be more informational than concrete. For example, seeing how a machine works in the utopian world would allow you to find the parts in the dystopian junk pile and put them together. Or finding a room in the utopian world would tell you how to find the same blocked off room in the dystopian world.
  • The utopian view could be intended, and you are permitted to survive as long as you act like you're in utopia. You might have to role-play normal life while you see horrific things going on around you.
  • You could maybe mix and match utopia and dystopia, e.g. like you're wearing a visor and various cells of it fail on occasion, so some screen fragments are utopian and others are dystopian. You could move the view around to contrast the two.
  • True reality could be neither option. The utopia could be imposed by the government, and dystopia by the rebels. The trick is seeing that they are both fake.

In Topic: Monte Carlo Tree Search - Fatalistic play

17 November 2014 - 04:39 PM

In general I'd stay that I agree with Alvaro. ;)


Depending upon your game, possibly the trees go very deep and it can't explore enough nodes in the time allotted. Perhaps you need some domain-specific estimate of the win/loss ratio based on the board state to optimise the search for previously unexplored nodes?

In Topic: Taking A Group of AI Followers Indoors

16 November 2014 - 01:26 AM

As noted above, I think it will really be bad if players occupy the same space. I remember an Aliens console game that did this and how weird it looked to see a half dozen colonial marines all merge and demerge from the same spot. That was using 3d with an isometric camera, so I think FPS would look even worse.


I agree that would suck... you could use soft avoidance instead of hard collisions. For example the NPCs could avoid each other in pathfinding using influence maps. But it would still leave open the option of passing through each other temporarily. Another possibility is only allowing greater pathfinding flexibility when an NPC is behind the player or more than a certain distance away.

In Topic: Taking A Group of AI Followers Indoors

15 November 2014 - 07:25 PM

This is perhaps a bit over the top, but there's a nice talk on how Naughty Dog made Ellie as non-annoying as possible in The Last of Us:



Dealing with large numbers of followers... it's a tough problem unless you disable collisions or make the collision volumes tiny. Perhaps in confined spaces have a minimum follow distance. If the player goes too far inside the minimum follow distance, the whole column of followers back up. I think not breaking stealth is quite important, maybe they try to stay in cover if you're sneaking, and the enemies ignore them unless they do anything too blatant.