Fron another thread: Quote:
Quote:
Finally, is it possible in DirectX to make an invisible model cast a stencil shadow and if so could you add this to the todo list, too (assuming it isn't already possible...I'm away from my regular PC, so I can't check).
You already can use, or even already use normally, the LOD of your charcter for the shadow in A6.
Just picked up this sentence, so it can be that this already had been mentioned.
Yep, I'm aware that we can use LOD for stencil shadows (Wintermute has this feature too); however I know that SS don't work when a model is transparent, so I was wondering if they worked if they were completely invisible (my.invisible = on;)? The manual states:
Quote:
Stencil shadows are only cast by completely non-transparent models. The model skin must not contain an alpha channel.
I wonder if there's a way to make a model cast the SS and still be completely 100% visible but not render. It sounds much like the hidden geometry issue doesn't it? If his would work, I'd use a lower poly model for the realtime model that projected shadows and attach a 2D animated character with a lot of frames (so it's smooth and looks like a highly detailed model without having to use a normalshader). I would get the visual results I'm after plus a working stencil shadow all within a 2.5ish world.
------------------------------------------------
New stuff related to the above post:
Is it possible to do something like this? I know that sounds contradictory, but the engine would think the model is visible and non-transparent, but we wouldn't be able to see it because of some flags settings.
I have reasons for wanting to combine a 2D animated character (constructed from a higher poly model or hand drawn) with a lower poly, non-rendered, realtime model. The sprite would appear at the location of the non-rendering model and all animation frames of both versions would match depending on user input. It would be easier to maintain with point and click input, but Intense Pathfinding will need some tweaking to use it (Larry is aware and one day he might resolve the issues according to a past post of his...he has bigger fish to fry right now).
1. One reason is speed and savings on poly count and lack of having to have to use a normalmapping shader. This will allow lower end systems to be able to run the game efficiently since they won't need the shaders technology plus there will be less 3D crap to render out.
2. I believe I've lost my animator. My modeler (Dan) is still onboard but having a run of bad luck, so his turnaround has been off. I'd still use his models for shadows. I have acquired a copy of Poser 7 and what I would do is create a unique character via morphing their base characters (or buying one that I like and doing it) that closely matches the characters that Dan is making for me. I'd use unique textures, of course. The program has animation features including a fantastic walk/run/talk generator (it actually works wonders) and you are allowed to use 2D renders or animated sprites for commercial purposes if you use their 3D models as the base of your art.
Anyway, it all sounds crazy, but it should work if I can get all of of these non-rendering things to work out right. The end result is a unique looking game with much less 3D mesh and geometry to render and higher visible detail to the level art but at the same time still have all the benefits and features of a realtime camera and world.
My User Contributions master list - my initial post links are down but scroll down page to find list to active links
| 
