Hi. Can anyone suggest strategies or best practices for speeding up Composite operations other than reducing the size of the images?
I am using a Composite TOP with Screen method on rollover of a button. The images in question are quite large. When I rollover there is almost a 1 second delay before the Composite effect has effect.
In AfterEffects there is the idea of ‘Pre-composing’. How would you do the same thing in TD?
Is it possible to reproduce a Composite Screen effect in GLSL and achieve a speed increase or is the Composite TOP already optimized?
ok will try
The 6 buttons on the left on rollover cause a Composite screen effect to be applied to 6 different images (over the 'Baser_Color_Map)
There is more than a half second delay when i mouse over the buttons for the ‘highlited’ mouseover. This is because the Composite effect is being applied to large images.,
If I instead bypass the Composite TOP there is no delay.
Will try ‘precomposing’ all 6 images and feed into 6 Cache TOPs.
Generally, tops are already optimized glsl code, I have a feeling that you might be running out of gpu mem? usually memory bandwidth ends up being the bottleneck with processing large textures. maybe check task manager to see if gpu mem is quite high. Does this 1 sec lag happen when you have 4 images? or 2?
Def can go the route of caching using the cache top. also switch top usually can accomplish the same thing in simpler situations, as it doesn’t consume any memory itself, it just essentially “references” to one of it’s inputs.
However, if you are going past the dedicated memory your gpu has, windows has to extend the memory using slower cpu based memory, and that is technically why sometimes patches with large texture use slows down after a certain number of textures.
