Pipeline R&D

DESCRIPTION

Building a complete in-house pipeline from scratch with just a single developer was unrealistic, despite being challenging and very interesting.
We thus decided to look for existing solutions that could meet the studio's needs. A wide variety of existing tools and solutions were explored, to identify the best fit.

For production management, we assessed industry-standard as well as lesser known solutions, including ShotGrid (formerly Shotgun, now Flow), ftrack, Kitsu, TACTIC, and Aquarium.

In parallel, we explored a range of open-source pipeline and asset management tools such as OpenPype (now AYON), Prism, Plex, Damas, and Kabaret.

Ultimately, we chose a combination of OpenPype and ftrack, which formed the foundation of our pipeline.
This setup has since undergone extensive customization to better align with the evolving workflow requirements.

ROLE

Production Managers

To identify the best production management tool for our needs, I conducted an in-depth analysis of existing solutions.
Each was evaluated based on maturity, features, customizability, support options, and community engagement.

The main systems considered were Flow/ShotGrid, ftrack, Kitsu, TACTIC, and Aquarium.

Being open-source and locally deployable, Kitsu and TACTIC were the first candidates I tested extensively through local installations.

Kitsu impressed us with its clean interface and focused feature set, but ultimately it lacked several capabilities required by the studio (though it has since improved considerably).

TACTIC, on the other hand, felt overly complex and more suited to large enterprises than VFX studios.
A specialized VFX version existed but was outdated and buggy. I made several code adjustments to improve compatibility with newer environments, but persistent issues and the lack of support for the open-source version led us to discard it.

I later came across Aquarium, which showed strong potential but was still in early development and not production-ready.

ftrack was already in use at the studio, so we already knew what it was capable of.

ShotGrid couldn't be directly tested due to licensing constraints, but internal knowledge from the team provided sufficient insights and made testing unnecessary.

Throughout the evaluation process, I regularly presented findings to studio leadership, the production team and supervisors. I also organized basic training sessions to let team members test and give feedback on each solution.

In the end, we chose ftrack, which offered the best balance of features, cost, and integration potential.
ShotGrid was a strong contender, but its higher pricing and the lack of integration into OpenPype (see below) made us choose ftrack.

Pipeline

In parallel to production management research, I also explored open-source pipeline and asset management tools.

The main ones were OpenPype (now AYON), Prism, Plex, Damas and Kabaret.

Prism and Plex were well-designed and user-friendly but seemed tailored more toward smaller teams or individuals.

OpenPype stood out for its robust feature set, scalability, and customizability. Its integration with ftrack was also a big positive.
Even more, as it is based on Pyblish, we were already familiar with its usage and some of our work could be reused.

Damas was more of an experimental project with an innovative UI, but it lacked essential features for production use.

Kabaret was very powerful and flexible, but it was more a development framework on which tools have to be built than a ready to use solution.
I built a few prototype tools with it to test its capabilities, but it required significant development effort to build a usable foundation.
It would be ideal as an added layer to an existing pipeline.

Ultimately, we built our pipeline around OpenPype, tightly integrated with ftrack, and adapted it extensively to meet our specific production needs.

---