What's a good work flow for animation? Let's say I'd like to make an animation that about 5 min total, and is comprised of 10-15 smaller clips. Do you generally make each clip a separate hip file? Or one hip file with a camera for each clip?

You should work on a per shot basis . Then make an edit with your rendered images.

rob

There are some nice tools that allow you to pull nodes to and from scene files between shots.

There is the hscript command “mread”:
http://www.sidefx.com/docs/houdini10.0/commands/mread [sidefx.com]
This command from the textport allows you to pull a camera from one of your files like this:
mread -m cam* my_hipfile_s01.sh03.hip
would grab all objects with the prefix “cam” from the hip file called my_hipfole_s01.sh03.hip in to the current scene.

This means you better get your naming conventions down for the nodes in your scene.
e.g.:
All cameras have the prefix “cam_”
All lights have the prefix “lgt_”
All lights that are key to have the prefix “lgt_key_”
All lights that are fill to have the prefix “lgt_fill_”
etc…

And never ever deviate from those conventions that you set for yourself. As you start to automate scripts to manage large projects, you live and die by prefixes and suffixes.
As an aside you can use smart bundles along with the prefix and suffixes to have bundles automatically populated with nodes to aid lighting and rendering.

To write out part of a hip file you can use the textport command “mwrite”:
http://www.sidefx.com/docs/houdini10.0/commands/mwrite [sidefx.com]
You can use mwrite to write out the entire scene to disk.
More importantly you can also use mwrite just to write out the animation channels for your scene. These channels can then be read in with mread -o -M … .

It makes a lot of sense to use Digital Assets to wrap up some parts. For example, if you have a few scenes that share the same lighting setup, it makes sense to wrap up those lights in to an HDA and write that to a local directory called “otls”. Same goes for props and set elements.

It is also a good idea to see what are your hero objects, foreground objects, mid-ground objects and background objects that are shared between your shot files in a scene. Again wrap these various components up in to HDA's and manage things that way. Make a change to the HDA and all the changes will be inherited by all the shot files reading those assets.

If you have a one-off shot change to an asset, simply copy/paste the .otl file containing the HDA in a file browser and rename the otl file to a version shot suffix like:
cp midground_trees.otl midground_trees_sh07.otl
and then in Houdini, using the Operator Type Library manager from the Windows menu to point the current definition to the shot specific otl file on disk. Because the HDA's will have the same name, you will be now pulling that asset's definition from the new .otl file to do your one-off shot edits.

—-

If you want to do a couple camera cuts in the current scene file, you can use the Switcher Object. The Switcher Object allows you to wire in two or more cameras and then switch between them by animating the input parameter. The switcher object shows up in the viewport as a regular camera so you can look through it and see your camera cuts. You can also render from a Switcher Object from a mantra ROP output driver.
In fact you could do your entire scene this way but for 5 minutes, that would be difficult to manage but hey, I have seen one single hip file do a continuous 12 minute shot bopping between a dozen cameras or so with a Switcher Object in a feature shot.


Larger facilities will have a rigid process in place where each shot is in a separate directory for a scene. For personal projects, you can get away with putting your entire scene with the various shot hip files in a single directory and work like that. If things get crazy, you can create separate sub-directories for individual shots later on.

Thanks for the detailed reply Jeff!!

Probably you can use “takes” to handle your animations and cameras within shots.