Rene Damm

@ndickson

Thanks for chiming in.

I've attached the HDA with the network and the script this is happening with (though given it needs a specific project setup, it's probably hard to replicate the problem with it).

While I still don't know why exactly it is happening and whether it may be a bug or whether it's simply an invalid node setup, I've found what's causing it and how to work around it. And you're right, it's almost certainly not an internal limitation.

The problem happens if the Mantra ROPs I'm using have no output file set and the Render COPs are picking up the render directly from the ROP (regardless of whether it has an output file set or not!). If instead I use File nodes in the compositing network and pick up renders from disk instead of directly from the Mantra ROP, then things work fine and everything updates as expected. However, if either the Mantra ROP isn't rendering to disk or is directly connected to the compositing network, then the output will not update after the first render.

The latter behavior seems especially strange to me. Also, it's obvious that Mantra *does* re-render in that case (which it does *not* do if there's no output file set on the ROP node). It's just that the Render COP will not re-cook after the first render and thus eternally stay on the first render from the Mantra node.

//Edit: just to mention it, the network setup that doesn't update correctly during script execution updates perfectly fine while working in the GUI. Especially all the compositing nodes will update immediately in response to any change in inputs.

Does Houdini Indie allow only a single cook of the graph for every single execution of a script in order to not allow people to work around the rendering size limitation on animations?

I have a script that renders (or, well, tries to) multiple views of the same object (no animation; no changes to or dependency on $F). When running it, however, all output files contain the same single view.

Messing around a bit I found that only the first view (any arbitrary one) set by the script will render as expected. Subsequent setParms() calls will have no effect on the renders – none of the nodes in the network will re-cook.

So, that makes me wonder, is that a limitation specifically built into Houdini Indie? Or am I just missing some important call during script execution that will trigger re-cooking? (I tried explicit .cook(True) calls but they have no effect)

Thanks

(BTW Houdini is flippin awesome. Enjoying this software so much)