Wanted to know if there is a reason why something like Karl Henkel both his “3D Object to Layer” and “Layer to 3D Object” nodes are not included in the stock nodes ?
Using it quite often to bypass some limitations (Use Copy 3D Object with Material with layers), having scene based compositions not having to select different Z-axis to blend nicely etc.
Once “Allow custom node paths” will be implemented it won’t be really a problem anymore, but until then, it’s sometimes complicated to share compositions with custom nodes.
We’d like to provide a way to convert between 3D objects and layers easily. For example, this would be useful to overlay text instructions on a 3D scene. I’ve opened this feature request for voting.
The Parabox nodes are a starting point. But, to avoid unexpected/buggy behavior in some cases, we’ll need to go further. The Parabox nodes take advantage of the fact that you can easily turn a layer into a 3D object and vice versa, superficially. We’ll need to do a deeper conversion. Layers and 3D objects have different assumptions. (Layer shaders assume they’re always going to be rendered in a certain order, on a quad, facing the camera, without depth testing. Scene objects can be rendered in any order, using any mesh, with whatever transformation, with depth testing.) We’ll need to convert from one set of assumptions to the other.
having scene based compositions not having to select different Z-axis to blend nicely
I’m not a coder or so I don’t understand what’s under the hood. Part of QC’s strength if I’m right was that you could combine Sprites and Billboards correct ?
So could you theoretically combine layers and 3D objects without them having to be converted ? For example a text layer on a scene where the text would not be affected by the camera movements ?
Is that covered by Improve rendering of transparent objects (OIT, Depth Peeling)?
I guess so. Although sometimes a lazy solutions like this quick conversion from Paradox does the job very well. No depth testing for 3D converted to layers and they look great. Again I don’t really understand the code behind but if “no Depth Testing” is GPU lighter than some complex testing like that feature request, I guess it should be an option for the cases where it works.
But yeah agreed stock nodes should behave in a buggy way. Or perhaps a version 1.0 with some limitations described.
I have added 2 modules in this node pack that include basic Layer ↔ Scene Object conversion, like the Parabox nodes did (that don’t work in Vuo 2 anymore).