Speed Optimization of Shapediver model

I have a few questions regarding speed optimization for my models.

This closet consists of some ‘boxes’ (and some boxes with holes in them) which are spilt up in order to texture them as plywood.

Is it correct that Shapediver will send each face to the browser, so it does not matter if the faces are merged? Resulting in a wooping 462 requests which takes so much time to send that it slows everything down a great amount.

The models itself is quite fast to compute but its the sending which is the problem.
I have seen this blog post:
Desktop vs Cloud Computation Time Explained! and I have changed the graphics a bit to reflect reality for my model. :smiley:

Maybe you can supply me with some tricks to make this definition perform a bit faster?

Uhmm, dosent really make any sense I think. I don’t think that this is a computation time problem which are what is mostly addressed in this blog post. The meshes generated in my file are not complex at all…

The model is fairly simple in terms of geometry and it should run faster. Here are a couple of tips:

  • avoid boolean operations such as Solid Difference or similar and ideally create meshes straight away
  • merging objects can make a difference but you probably need to keep objects separate for texturing
  • flatten the ShapeDiver Display geometry input if you can
  • check for duplicate objects in your definition
  • simplify meshes as much as possible for rendering
  • if you need output for manufacturing, turn it off with a Toggle and Stream Filter and enable it only when the final configuration is ready
  • texture size should be around 1MB at most, obviously the smaller the better

Opening the definition on your computer should be fast, if it takes more than a second, start optimising. Following tutorials give you ideas what you can improve.



You missed to answer my question I am affraid. :smiley:

Is it correct that Shapediver will send each face to the browser, so it does not matter if the faces are merged? Resulting in a wooping 462 requests which takes so much time to send that it slows everything down a great amount.

Those optimizations that you mention seems to have more of an effect to computation time (which is not really a problem now as it is around 1 second locally)? Or am I wrong?

You can get that many of requests for large models but it should be lot lower number in your case. The issue is probably the data tree you feed into the ShapeDiver Display component, lot of branches would explain the number of requests. Try to flatten the data and also check the texture sizes.

That was actually a usefull answer… And it helped.
From 348 to 179 requests… That was nice! Its still not as fast as I would expect and like, but its okay…