This article is about a web app I started years ago and recently restarted. It's called VFX Talent.

In the mid 2000s I lost the chance to work on Zack Snyder's 300. The VFX Supervisor called, but I was holding out for another gig so I asked him to wait. The other job didn't materialize, and when I called back, I was too late for the 300 work too.

I also saw problems from the employer side. At MPC (a large VFX studio), I was involved in the process of crewing up for new productions. Most artists are contracted only for the production duration. We developed a Ones document, a spreadsheet listing positions and the number of vacancies for each, and tracked each position as it was filled.

Artists were hard to find. Most came through word of mouth or from previous studio projects. HR was always swamped with hundreds of email threads and piles of CVs.

VFX hiring is fundamentally different from hiring for permanent positions. Productions have to assemble large crews quickly, then disband when production wraps. Sometimes, the production company itself will build-and-strike, forming for one project before dissolving.

Productions chase tax credits around the globe, and artists have to be nomadic. The life of many VFX artists is a game of musical chairs.

I wanted to create a platform that fixed the broken VFX job market. VFX specific job boards exist, and of course there's LinkedIn, but none address the unique challenges of finding and building entire short-contract VFX crews.

I fleshed out a plan, and after a few unsuccessful attempts at outsourcing, I decided to build it myself. I learned Ruby on Rails and made great progress, but ultimately I focused on the wrong aspects and didn't finish. Here's an old mock up.

Artists never pay, and employers only pay for successful hires. The system handles all notifications and communications that stem from the auction flow, such as "Sorry, I accepted a competing bid".

See the bidding flow in the mockup for more details.

In February this year I found myself with more time, so I decided to restart the project. Soon after, conincidentally Technicolor collapsed, ending MPC, The Mill, and other subsidiaries. Thousands of VFX artists lost their jobs and found themselves wandering around LinkedIn.

I'm trying not to get too distracted by graphic design at this stage. I quickly picked a Russian-styled font and made it red to give a revolutionary, take the power back, feel. But the focus now is on getting the auction logic right.

A user can have both roles. Someone might be hired as a supervisor at a studio, then need to hire their own team. Users choose their role for the session when signing in, not when registering.

Representatives have the employer role and can represent multiple organizations. They might be an agent serving several studios, or work for a studio with separate organizations for different locations or productions.

Representatives can then create projects for those organizations.

I battled for a while with the UX for forms like Create and Edit. It felt like too much navigation to have a new page for each form, especially as more resources are introduced. Users would get lost pretty quickly. So now there are just a few main pages rendered, and when something needs to be created, edited, or deleted, it's done with a modal dialog, or a single click. Either way, the user stays grounded in the world they are working in.

Index pages display resources as cards. I'm experimenting with an expand icon for overflow content instead of scrollbars. The jury is out. Try it below:

The project page is where I expect an employer to spend the most time. This is the equivalent of the Ones document mentioned earlier.

The Select an Artist field in the create bid form below is currently a simple dropdown. However, the employer will need a better UX to find the right artists. He'll need to match project requirements with artists' skills, location, availability, and so on.

No decision yet, but it felt like a good time to flesh out the artist profile a little so I can discover what that experience should be like while building it.

Every bid includes a message thread for negotiation and other communication. The system also records status changes in the thread. The Notes field in the bid form starts the conversation.

An artist's profile must provide enough information for employers to assess them. I figured skills and credits would be a good start.

People have diverse skillsets, but they do not always want to do the things they are good at. They may also have, say, 5 years experience at something 20 years ago, and wouldn't know where to start now. I wanted a way to express this level or weight metric without it being explicitly about aptitude, years, or passion. It's not a simple novice-to-expert scale. It's more about the artist saying, "I wan't you to notice these things more than these other things."

I created a skills cloud. The configuration panel avoids labeled levels like novice or expert. Instead, text size and weight indicate the relative importance.

Skills matter most in context anyway. If an artist shows an example of great crowd simulation work in their gallery, knowing whether they used Maya or Houdini helps employers to decide whether they are the right crowd sim artist for the job.

I'm not sold on the skills cloud yet, but I like how it discourages putting all the sliders to max.

Credits are a list of projects that the artist has worked on. Here's a quote from Kevin Todd Haug's MANIFXTO blog.

There's also a batch import tool. This lets artists quickly import their entire credits history from IMDB. Since IMDB charges $150k annually for API access, and blocks scraping, the best workaround is to have artists visit their IMDB page, select all, and copy-paste from there. The credits are extracted and formatted automatically.