I wonder when the added subdivisions stop helping (as in like is there a mathematical way to judge how many polygons are needed) or is it manually judged. It's a bit interesting because I remember phasmophobia had one issue for a while where one of its beds was represented by apparently 121k polygons, but eventually got fixed and reduced to 3k (without much obvious difference except for much better performance).
weszhuang
Is there some sort of acceleration to the rendering of loop subdivided meshes that approximate the resultant image without explicitly calculating the new subdivided vertices?
Mehvix
@weszhuang given the recursive nature of subdivision (level n depends on n−1) I don't believe we can skip calculating intermediary values
aravmisra
These are really good questions- to (attempt) to answer the first question, I believe a lot of game engines have algorithmic ways to determine subdivision for their meshes based on the distance a character/camera (player) is from the object? For example I recall seeing a video where as the character moves away from a mesh tree, the needed # of triangles decreases (proportionate to distance), and of course the reverse is true. I'm not sure if these algorithms are proprietary or if there is an industry-accepted one that everyone uses though! Would love anyone to step in and provide context.
I wonder when the added subdivisions stop helping (as in like is there a mathematical way to judge how many polygons are needed) or is it manually judged. It's a bit interesting because I remember phasmophobia had one issue for a while where one of its beds was represented by apparently 121k polygons, but eventually got fixed and reduced to 3k (without much obvious difference except for much better performance).
Is there some sort of acceleration to the rendering of loop subdivided meshes that approximate the resultant image without explicitly calculating the new subdivided vertices?
@weszhuang given the recursive nature of subdivision (level n depends on n−1) I don't believe we can skip calculating intermediary values
These are really good questions- to (attempt) to answer the first question, I believe a lot of game engines have algorithmic ways to determine subdivision for their meshes based on the distance a character/camera (player) is from the object? For example I recall seeing a video where as the character moves away from a mesh tree, the needed # of triangles decreases (proportionate to distance), and of course the reverse is true. I'm not sure if these algorithms are proprietary or if there is an industry-accepted one that everyone uses though! Would love anyone to step in and provide context.