Tweet
hi.. ive just finished rendering a scene with a load of animated character proxies. was fine while they were instances, nice and fast to scrub the timeline, - i actually copied the proxy to my ssd which also helped speed things up.
however there are a few hundred of them, and my next step was to make them all unique, and use a script to randomise the frame offset. this works great to make the crowd animation look natural, and unlike pointcache, you dont lose the offsets when you remap the paths at your renderfarm.
buuut. im waiting about 2 mins for the viewport to update on each frame change, as its loading the mesh for each one seperately. its also taking a significant chunk of the rendertime just to load the proxy data.
are there any efficiency improvements that could be done here? an option to load the entire proxy before rendering / caching it to ram so viewports can be scrubbed? the whole animated proxy is only 200 meg, so it would fit quite happily in my ram, and on my ssd should take less than a second to load in its entirety. instead its being accessed on my disk once for each offset copy, and taking a couple of minutes to do its thing.
another option that would be helpful, would be if you chose "display as box" it simply ignored the animation? then you could have a bounding box, and scrub away on the timeline? or maybe a very lightweight animated version that can be loaded in milliseconds?
basically its a bit dog slow at the moment and i feel sure it could be faster/more efficient.
however there are a few hundred of them, and my next step was to make them all unique, and use a script to randomise the frame offset. this works great to make the crowd animation look natural, and unlike pointcache, you dont lose the offsets when you remap the paths at your renderfarm.
buuut. im waiting about 2 mins for the viewport to update on each frame change, as its loading the mesh for each one seperately. its also taking a significant chunk of the rendertime just to load the proxy data.
are there any efficiency improvements that could be done here? an option to load the entire proxy before rendering / caching it to ram so viewports can be scrubbed? the whole animated proxy is only 200 meg, so it would fit quite happily in my ram, and on my ssd should take less than a second to load in its entirety. instead its being accessed on my disk once for each offset copy, and taking a couple of minutes to do its thing.
another option that would be helpful, would be if you chose "display as box" it simply ignored the animation? then you could have a bounding box, and scrub away on the timeline? or maybe a very lightweight animated version that can be loaded in milliseconds?
basically its a bit dog slow at the moment and i feel sure it could be faster/more efficient.
Comment