This is a pretty clever way of referring to data structures to be memory-efficient (which is often a difficult problem in computer systems), as it maps neatly to efficient systems usage. It's also neat that an initial transformation can be applied through the incoming edge to customize repeated instances of the same item.
jifengy
So this is almost like referencing a class whenever it repeats in order to create a more readable overarching graph?
pw123
Are there more memory efficient ways to represent repeated objects in a scene so that you don't have to hold every transformation A on incoming edges?
dchen1
I would say this is already pretty memory efficient... You have to store the transformation info somewhere, and it's only a 4x4 matrix I think. If not stored with edges, I guess you could store them in some array somewhere? But in terms of pure memory usage I think this is the best you can get..... I don't know if there's some way to compress the matrices somehow and be able to restore them back into the original transformations, but that's the only way I can think of.
triangle
To clarify, we still get to define our own transformations for each of the pointer nodes right? So that we are able to move the object copies in different ways?
This is a pretty clever way of referring to data structures to be memory-efficient (which is often a difficult problem in computer systems), as it maps neatly to efficient systems usage. It's also neat that an initial transformation can be applied through the incoming edge to customize repeated instances of the same item.
So this is almost like referencing a class whenever it repeats in order to create a more readable overarching graph?
Are there more memory efficient ways to represent repeated objects in a scene so that you don't have to hold every transformation A on incoming edges?
I would say this is already pretty memory efficient... You have to store the transformation info somewhere, and it's only a 4x4 matrix I think. If not stored with edges, I guess you could store them in some array somewhere? But in terms of pure memory usage I think this is the best you can get..... I don't know if there's some way to compress the matrices somehow and be able to restore them back into the original transformations, but that's the only way I can think of.
To clarify, we still get to define our own transformations for each of the pointer nodes right? So that we are able to move the object copies in different ways?