JS LOAD RENDER PASSES

Everybody seems to love Multilayer EXRs, especially when it comes to ACEs/OCIO Color workflows for CGI Compositing. Well …

I DON’T!

Here is why I prefer rendering separate files instead, and why you should probably too:

• Performance. Any Compositing tool works much more fluently and efficiently if it does not have to extract all information for all AOVs/passes out of one huge bulky file.

• File Sizes. With different AOVs/passes come different needs. Not all AOVs need to be full 32 bit color depth. When rendering layers to separate files, you can choose the right setting and right file type accordingly. This will save you much disc space on the long run and also use your RAM more efficiently.

• Flexibility. If some frames of one of the AOVs/passes are corrupted, or if one single AOV has had a wrong setting for whatever reason, you can easily re-render just that one AOV, instead off having to re-render everything.

The only Downside is: Organizing all these files manually. And this is where JS-Load Render Passes comes in. You can simply have the best of both worlds. And actually even a bit more.

• You render all AOVs/passes of given compositing layer as separate sequences in the same directory.

• You choose the right settings and desired file type for each AOV/pass.

• You just need to follow a simple naming convention and you are all set. All you need to do is to browse the beauty rendering and off you go.

• See how it works in detail below.

At Studio Rakete, where I work from time to time, we use a similar approach. But I also worked on a different project lately where I had to learn the hard way that Multilayer EXRs containing ALL passes can be a pain, because of the bad performance. I ended up redoing most parts of the rendering, just to get a responsive setup.
And that’s why I’d like to share this.

The JS-LRP Properties work from top to bottom.

First you decide if you need or want any of the optional settings like inverting the AO or if you render with blender, if you like to standardize some of it’s unusual RGB values or axis.

Then you hit ‘Load Beauty’. A file browser will pop up and you are asked to browse the beauty rendering.

Please make sure there are ONLY the files in this directory belonging to this given render sequence.

JS-LRP will examine this folder and load all the sequences in new readers, create channels and layers for you and pipe them all together in one stream.

You can see how the Properties and name update.

After Loading the ‘Unload Passes’ button becomes select-able. This is where you can unload the sequences. This will also Reset the Gizmo and revert to it’s initial state.

Also the option to expose AOVs for further compositing becomes select-able.

‘Further Information’ will bring you to this page.

Naming your Files correctly is essential to make JS-LRP work. Here are some rules you have to follow:

• Directory: All AOVs/passes belonging to this compositing layer need to be in the same Folder. But there mustn’t be other files in there. Keep these folders tidy.

• Naming: All AOVs/passes should follow the same naming convention. Let’s say you want to include the project name, sequence number and shot number in your filenames, a possible name could be:
  “Projectname_sq010_sh0010_PassName_####.exr”
  All other AOVs need to follow the same structure.

• Frame Range. All AOVs/passes should have the same frame range and the same quantity of digits.

• The end of the name: This is most essential. It needs to end with the following pattern:
  “…_PassName_####.fileextension”
  The underscores “_” are essential, as well as the fact that there is NO underscore between the frame count and the file extension.

• Substrings. If you want JS-LRP to handle the passes listed below with some common sense, please use the listed substrings in your pass names.
  The check for substrings is not case sensitive, so for example “…_ao_####.exr” will work.
  And “…_Occlusion-AO-Pass_####.exr” will do the same! eg.

• Pass names. You can name them to your liking, just don’t use underscores “_” within your pass names!

Here you can see what has been created under the hood inside the group. In the picture you just see the top- and the bottom part of the group, because it would otherwise cover way too much screen space. But I hope you get the idea.

At the top is the main reader, the Beauty Reader and to the left are the responsible secondary readers that hold the AOVs/passes.
Depending on the naming convention it looks for the substrings mentioned above and then handles the layers and channels by best practices.
But if you need to adjust anything, feel free to open the group and fix stuff.

But just DON’T rename ‘BeautyRead’ or ‘PassesOutput’. Otherwise the Gizmo can’t operate anymore.


See below how all converted AOVs for this particular render layer look like:

Here you can see, what happens when you click the Add AOV Comp button.
The Idea here is to work with subtractive compositing.
This means even though you start with your beauty rendering, you still have the possibility to expose and tweak individual AOVs separately.

You need to have a node in this stream selected first, so JS-LRP knows where to build the little comp.
Another Popup will appear and you can chose from AOVs.
Depending on your naming convention, not all AOVs will be included in this list.That is on purpose to not clutter up the list unnece-ssarily. In a subtractive compositing workflow it makes only sense to expose AOVs, that actually contribute to the Beauty Image.
Other Masking- or Helper-AOVs like AO, Vector, Depth, Position, Crtypto or Normal do not really make much sense in this place.