ROUBAL wrote:Maybe I have missed something, and probably I am going to say something stupid, but I have done some fast calculus and I don't understand the interest of Cloud computing from the distant client point of view :
imo - like almost everything - it just depends on; if you are a "pro" user, it might be much cheaper to invest in a fast internet connection to shorten upload times, than to wait 1-n hours per image. in case of corrections based on clients' requirements (usually told today evening and to happen until tomorrow morning
) this will save even more than just money (sleepless nights, reputation, ...).
if you take archviz for example, there are usually not that lots of big texture maps in use. also geometries compress extremely well - a 350mb .obj. only needs 85mb if compressed.
all in all there will be imo enough demand to let the farm gpus burn 24/7 right from the start, if only a fraction of the current octane userbase pushes the service.
... and if you i.e. use live-db materials in a scene, their textures might even be excluded from the upload. another possible path is to extend the live-db to hold a per-user library of materials online (with local caching), means even if it is empty at the beginning of a project, you propagate it with mats while you work on a scene, and therefore upload your stuff in chunks already along the way to a last & final rendering. it might be possible to cache texture maps, so that the same evermotion wood.jpg
isn't needed to upload for every project anew for 1000 people that are using it