Good info. Tony also suggested using a material proxy on #hlcoders. I'll hopefully have time to look into it some this weekend. However, your mention of instruction limits is cause for concern. If the alternate vision mode behavior is controlled by a combo variable, wouldn't the compiler optimization of the static branching prevent the normal behavior of the shader from eating up instruction limit?
I'll also look into replacing the materials, but that sounds even harder to implement cleanly. Thanks. On Tue, Feb 22, 2011 at 9:23 PM, Maarten De Meyer <[email protected]>wrote: > Wasnt there a replacematerial method? I remember something like that. For > our alternate vision i ended up making a custom version of all the stock > shaders with a material proxy allowing me to switch between vision modes in > the shader itself. Advantages there are the switch is immediate and you can > change the effects outcome per material, contrary to a postprocess. > Disadvantages are that you have to be able to do it in a shader ofcourse, > and that the stock shaders leave very little room for modification before > you hit the limits of the shader model. > ------------------------------ > From: Zach Brockway > Sent: woensdag 23 februari 2011 0:23 > > To: [email protected] > Subject: [hlcoders] Overriding materials, especially while rendering world > (BSP) geometry > > Hi list, > > I've been investigating the possibility of being able to do > across-the-board material overriding. Motivated by an idea for implementing > an alternate vision mode, it would also entail a few other things like > rendering alternate view setups into render targets; however, all of that > seems to be well-supported and you even get the benefit of examples of how > to do it. The part that I'm really stuck on is the fact that there doesn't > seem to be a way to affect the material used for rendering BSP geometry, > since it's all blackboxed away in the engine. > > Things I've tried: > > 1. Calling ForcedMaterialOverride on modelrender, studiorender, and > g_pStudioRender, before render->DrawWorldLists is called. This had no > effect. > 2. Installing an IBrushRenderer using > render->InstallBrushSurfaceRenderer. My implementation of > RenderBrushModelSurface was never called, so I suspect this only affects > brush-based entities. > > Of course, I still have to worry about how to handle the rendering of > *everything else* if I manage to figure this out, but this particular > problem struck me as the biggest stretch for the engine. > > Thanks in advance for any advice. > > -- > "They've penetrated our code walls. They're stealing the Internet!" > "We'll need to hack all IPs simultaneously." > > _______________________________________________ > To unsubscribe, edit your list preferences, or view the list archives, > please visit: > http://list.valvesoftware.com/mailman/listinfo/hlcoders > > > -- "They've penetrated our code walls. They're stealing the Internet!" "We'll need to hack all IPs simultaneously."
_______________________________________________ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlcoders

