The artifacts and sample rates are starting to become a bit more apparent once you render at this high of a res, so I’ll try doubling the values and optimizing a few things with shadows.
Edit: Noticed the shadows looked horrendous and figured out why the sphere rendered without casting shadows on itself.
View more stuff by clicking here (updated)
Found out that the sphere was completely illuminated because of a typo. Fixed!
Next I approached smoother shadows by decreasing the sample angle by 33% and increasing the sample rate by 2x. The result is 6x smoother shadows
Thanks, and I think I’ll keep that material for now since it adds a little variation in material to the room though I may end up changing it in the future.
Would you be able to summarize the code for that…for…like…noobs. Is it for me? no no no, its just for my cousin’s wife’s cousin. [whispers to self] dang it, should’ve just said my brother, now now one believe me
Honestly though, Im not that great in lua, but Im pretty experience with unity’s c#, and it doesn’t use concepts much more different. How were you able to get the raycast to recognize shadows, if it was even raycasting. I mean, this would be awesome for stuff like in-game photo booths for your character, as well as security cameras. You has mentioned bloom, and other effects, how though?
EDIT: Turns out a lot of people do make guns from plain-sawn wood. Probably more out of necessity though, since it’s pretty difficult to get non-plain-sawn lumber these days unless you have a friend with a sawmill or have a sawmill…
I wrote a Lua parser in Lua that converts Lua code to some table format. To check whether it actually parsed everything and didn’t leave stuff behind, I made an unparser. Then I parsed the parser, unparsed the result and ran the result to see if it worked the same. It did, so I’m happy.
Took 78 seconds to parse to my AST format and 0.01s to unparse from AST to Lua code. Ran it in PowerShell using Lua 5.1.4 binaries, no idea how slow it would be in studio or so. I did write the parser in a style that made it easy for me to debug, so now I’m gonna write an efficient version of it and look how that’ll go. unparser is less useful and fast enough that I’ll skip on making that performant
EDIT: I don’t remember making it that a == b^3 and c becomes (a == (b^3)) and c and not ((a==b)^3) and c, probably should do that
Localized string methods (e.g. str:find(...) → find(str,...) with find defined at top) and switched from self:Method(…) to Method(…) (localizing ‘self’ and creating separate parser functions for each parser), yet speed increase is almost unnoticeable. Gonna try some stuff, mostly reducing calls… guess it’s doomed to be slow.
I find it interesting the amount of developers building with Shadow Mapping instead of the CPU based voxel system (aka production). Doesn’t look Roblox-y to me, which is probably good as they look BA, but these builds really depend on Shadow Mapping shipping instead of the GPU based voxel system.
