PSYOP Pipeline


Got a link from @tokejepsen about this video and thought I’d pass it forwards.

The presentation itself is Shotgun-centric, but the pipeline components don’t necessarily depend on it and are equally suited to pipelines without it, or any database for that matter.

Many ways to publish

Before I begin I feel it necessary to point out that this is not intended as any form of criticism against Tony or Psyop, pipeline development is an art-form in its own right and there is no right nor wrong. But I wanted to draw some parallels between how their development differ from ours in an effort to learn and improve upon each others experiences.

Apart from involving a few aspects of a pipeline outside the scope of Pyblish, there was one thing I found motivating and very interesting in terms of publishing, which was the amount of ways there was to publish.

  1. Plain file-saving, with comments and versioning
  2. Export of specific formats like geometry and animation, with validations (called “sanitizing”)
  3. Export of playblasts
  4. Post-processing of published data
  5. Point-caching
  6. Generic publishing of files

Each of which with a dedicated tool and workflow.

In Pyblish-speak, publishing is synonymous with sharing. Which means that whenever there is any form of content sharing happening between artists within an organisation, they are publishing that content either formally or informally. Publishing can range from conventionally putting things in the same place each time to formatting of file names to including metadata in some form or another.

Each of the 6 different ways to publish all revolve around sharing and are equally susceptible to the four steps of Pyblish, SVEC. With these ways now visual and in mind, we can start having a conversation about how each could fit a common workflow with Pyblish!


Publishing a file

Publishing a format

Publishing a playblast

General purpose publishing

Publishing a point-cache

Publishing a file


This topic is now listed. It will be displayed in topic lists.


You also need to bare in mind that Tony has heavily customized the Shotgun Toolkit, and I’m not entire sure how much it resembles the “mainstream” Toolkit. For example when I used Toolkit last (1+ year ago) we mostly relied on the “Publish” tool for outputting formats like caches, playblasts etc.;
Think we also versioned up the work file, but can’t remember. Maybe @mkolar can:)


Ah, I assumed these were all of Psyop’s own design, as I haven’t seen any of them in the Shotgun videos, except for some resemblance to their “Multi-Publish”.


We were always publishing the workfile as a primary publish and all it’s outputs (quicktimes, alembics, cameras etc) as secondary, in toolkit speak, also versioning up the workfiles of course. We had easy option of validating, but didn’t utilize it much at the time, mostly to time constraints if I remember well. (to put in perspective on top of setting up pipeline, me and Toke have done lighting and rendering on about 1000 shots plus Toke did all the rigs, so busy is probably an understatement) I think it worked quite well, however it was lacking the ‘desktop’ feel that psyop gave it.