You will certainly use them a lot as you will have surely many streams of data (a bit like if you had in one single object multiple parallel operator stacks that you can blend/merge/dispose/etc…
My take is to try to do things at object level due to easiness with for example transformations, material assignment, scene optimisation and LOD. For example, every component of a wheel of a car I separate and make objects and have a hierarchy, this allows me to do very quick low resolution objects out of big ones. Transformations are much faster and ultimately I can do clever camera based hiding and what not. Also given I use bundles a lot having objects is very convenient as I can do text searches that bring the objects to the bundles so it is a major win after a bit of a slow prep time of course. So I would say my best friend is “object merge” operator rather than merge. ;-) hope it helps jb > On 10 Mar 2015, at 17:47, Jason S <jasonsta...@gmail.com> wrote: > > (see addendums in bold) > > On 03/10/15 13:32, Jason S wrote: >> On 03/10/15 12:15, Christopher Crouzet wrote: >>> This is a core concept when you have to deal with such graphs—it is so >>> essential that the `Merge` node is probably one of the most used nodes in >>> Houdini. >> I can understand why, whether for optimization, [or] manageability purposes. >>> Groups in Houdini share roughly the same purpose than clusters from >>> Softimage. >>> They are a core concept in Houdini as every node understand them. What you >>> can do with clusters, you can do with groups, and much more out of the box. >> I can imagine, as core [or as basic of a concept] as in Soft I would assume. >> [or so it would seem] > > And thanks for the, I think important clarification. > >