ok so ive made some progress and even started using editable images to get things done. doing so has managed to make it so much faster fps wise (0fps>~35fps during rendering)
its slower on the render time (id say it takes about 5 minutes to render 1 image on the 12 samples + 3 bounces now now a 512x512 img like the one below)
but now this can go up to the full 1024x1024 image that roblox supports without crashing and dropping to id say around 15-20 fps instead of 0 on a rtx 2060 (i havent yet checked that)
if its the cpu that matters then its actually quite decent for a pretty bog standard Ryzen 5 2600 which is what ive got atm
heres a 512x512 image from my most recent render which took probably abt 50 minutes on 128 samples and 3 bounces. this is starting to look extremely good on high res
i also got this 720p, 32 sample+3 bounce render which i think took around 15-25 minutes to render last night just before i went to bed
both images consistantly kept above 30 fps which is a massive improvement
and yes, the line artefacts were fixed, it was just due to some index with nil errors which i fixed by just zeroing the rendered value for a pixel if that happens. lines were caused by the line scanning being used to not blow up my pc
its not quite ready due to the obvious greying of light not reflected off a mirror, and i will need a proper ID to upload it since i am using editable images after all, so ill try do that within the coming months.
I do have an id, however it just doesnt work with persona due to a format difference compared to traditional national ids which sucks cause i rlly do wanna get this out before implementing depth of field and hopefully even real time rendering similar to Radius which is another path tracing engine in roblox which is what fully inspired me to use editable images due to how quick it is
and yes the 1024 resolution means this can now support 62 million rays on the currently used settings in the most recent (public) version instead of 980k which nearly crashes your client
also this is actually a path tracer
ray tracing is doing 1 bounce toward a light source or off a reflective object and seeing if it should be in shadow or reflecting smtn
path tracing is allowing a ray to bounce multiple times in random directions depending on material along with its smoothness/glossiness, etc, and having it bounce until it reaches a light source or reaches a maximum threshold, which is what im doing