I want a Source 2 port of Half-life with completely dynamic lighting

Well, yeah.

HL1? Also a masterpiece of a video game. And that's why I want a Source 2 port of it with completely dynamic lighting.

Black Mesa is a cool thing but it's a very different game in many ways.
 
Last edited:
And I didn't realize it until tonight but one of the Source updates must've broken HL1: Source somewhere along the way because AI and hit detection issues abound. All the more reason for a Source 2 port.

edit: or maybe it was the Steam Pipe update that split the directories
 
I'm definitely in the camp of people that felt FarCry ultimately stole the thunder of the eventual HL2 and Doom3's launches. While HL2 and Doom3 weren't lacking in polish and technical achievement, I felt like FarCry hit enough hallmark technical check boxes (physics, normal mapping, lighting) while delivering gameplay and environments that felt legitimately novel relative to past games in the genre. HL2's linearly sequenced gameplay and level structure and Doom 3's monster closets felt like a step backwards after that.

Given the Quake engine and hardware in 1996 I'd be hard-pressed to think up a more suitable design target than what became HL1, I would say that FarCry and CryEngine likewise for that 2001-2004 DX8/9 period.
 
Making lighting dynamic on a source engine title such as HL2 would mean re-baking all lightmaps and probes with indirect lighting only, and extracting the light sources and their properties and converting them to the data format source 2 uses at runtime. Possibly, some areas would have to be tweaked for performance, some lights would be left static, and things like shadowmap resolution and draw distance would have to be ajusted by hand.
That is, if valve can find all the original files with lighting properties used to bake the maps in the original game, otherwise you'd have to re-create it all from scratch.
Hardly a trivial task.
 
I wonder if there'll come a day when we'll be able to drag and drop old games into a real time path-tracer or similar global illumination renderer. I'd imagine processing the scene brushes and geometry in a BVH or similar structure is pretty do-able algorithmically. The textures would need to be migrated to a proper physical material system, and they'd look awful anyways so they may as well get redone from scratch. Any billboarding, sprites, and flat shader effects would have to get scrapped and redone by hand also. After all that the light propagation may end up looking perfect, but I suspect the environments would look even more artificial without the graphical vaseline of low res aliasing, bilinear filtering, mip boundaries, fuzzy radiosity, blob shadows, etc.

It would seem to me to be a better use of time to just recreate from scratch the portions of the game that are cool than trying to port everything over and have it look shitty. I'd rather have a fully modernized HL1 tram ride than have the whole game ported to Source2.
 
If inlimbo wants genuine dynamic lighting then you dont need lightmaps

Lightmaps are still usefull for indirect lighting, which can't be trivially done dinamically today. But you could, of course, replace them for ambient probes only, but in that case, there might be areas here and there where the original's lightmap looked better.
As for the dynamic light sources, the problem of finding the old lighting data used for baking the lightmaps in the game (not avaliable on the actual game, but probably in some place at valve's archives) and converting it for the new engine and tweaking it for real-world performance still stands.
 
Yeah I'd want fully dynamic lighting. No baked occlusion anywhere. For Half-Life as it is, it would look fine. And given how low overhead the game is on modern machines, they could probably throw in one or two light bounces in select areas (not that that would be completely trivial to implement)
 
Back
Top