Thursday, 8 March 2007

Progress! Of a sort...

A bit further on; some experimentation with Blender (one of the most agressively counter-intuitive programs I've ever used) and lots of frustration eventually lead to very slightly more interesting dungeon walls. After a bit of fiddling with the data, they even cast nice shadows.

Given that the major view point is always going to be top-down, I figure the z-pass stencil shadow algorithm is a win, and indeed it seems to work well even on my feeble home PC.

My current problem is the combination of line-of-sight with stencil shadows, and whether an ambient pass makes sense. On the one hand, the shadows accurately represent the area of the dungeon visible from the character's position, and most especially obscure creatures or items lurking out of sight. On the other, the player can see more of the dungeon from his or her elevated viewpoint, and simply not rendering the areas of the dungeon the character cannot see leads to unnatural pitch black shadows, and this makes navigation a little harder by obfuscating useful landmarks or features.

Given that traditional tilewise LOS will need to be implemented for gameplay purposes, I think a combination of hiding monsters based on traditional LOS methods, and using an ambient pass to hint at the structure of the dungeon not directly visible should work well. Now I just need a few monsters and a player avatar to test my suspicions with...

Guess that means I should get back to Blender. Sigh.

No comments: