Quasar post-mortem: Part 5

Below is a summary of the articles in this post-mortem. Although having full context is recommended, the subject matter explored herein covers many disciplines and can be consumed non-linearly. Feel free to jump around to whichever area interests you most:

Project Conclusion

Congratulations! You made it to the end! 

Believe it or not, there are many aspects of development which didn’t make the cut in this post-mortem. Anthropometrics, ML/CV implementations like facial hairstyle detection, stylization procedures, additional wireframes, FACS channel sculpting, predictive speech detection -- the list goes on and on. But overall, I think this is a good summation of the project. It is my hope that this series of articles can help product designers, technical directors, artists, and others avoid our pitfalls and maybe find some inspiration along the way. 

Shortcomings

Ultimately some portions of the Quasar product vision didn't make it in to the final application:

• We never deployed the hair-styling tool to detect touch-patterns, interpreted as common hairstyles -- e.g. Drawing a small circle on the back of the head to instantiate a hair tie, and then flowing lines from that circle to instantiate a ponytail [which matches the regional template’s most likely hair type] at that position on the scalp

• We never deployed the revolving Regional Template System which continuously incorporated new users’ face scans and regenerated the templates with their data

• We never deployed the facial hairstyle detection CV model -- opting instead to just ask the user to choose a facial hairstyle

• We never deployed the FBA Interpreter in full form -- opting instead to just transmit blendshape values rather than encode/decode them from an FBA data stream -- if we had deployed to more platforms than iOS and VR this would have been a higher priority

Quasar was the greatest learning experience of my career -- which meant there were innumerable challenges and failures along the way. I often felt like I wasn’t well-equipped to solve a particular problem area, but with humility, patience, and curiosity, things generally worked out.

As promised, I am providing samples of generated avatars in binary glTF form!

You can download the zip archive linked below, unzip the GLBs to your desktop, and then drag and drop them in a viewer like Don McCurdy’s glTF Viewer. After you drop them in the viewer, expand the “Morph Targets” dropdown menu on the right side of the viewer and try adjusting some of the blendshape values to make some funny faces!

• Note: ARKit expects morph target #21 (close lips) to always be used in conjunction with Morph target #23 (jaw open) -- it will look very odd if you use #21 without #23.

• Note: The lower teeth and tongue are removed for these GLBs as they are driven separately from the facial articulation.

The developer who made all of this possible:

Michael Bitsakis! Michael designed and developed all of the procedural operations, Houdini graphs, and authored the entire iOS application himself. If you have the opportunity to work with him, Michael is a powerhouse of an AR/VR developer and a great person all in one package. Hire him and you won't regret it!

• Michael's ArtStation

• Michael's LinkedIn

• Michael's Twitter

• Michael's incredible self-authored rendering and procedural generation application : InstantLight

Personal Contributions:

I really (truly) don’t like to toot my own horn, but it is a requirement for most potential future roles to describe my contributions to the project. Unless noted otherwise, I didn’t have help with any of the following buckets of work.

Gratefully, I was granted the time and freedom to develop all of these areas:

• Product Design

  • I was given an expressed goal to represent >95% of the global population accurately but other than that I was given free reign to establish and develop the vision of Quasar of my own accord

  • Aspirational mockups

  • Human centered design research

  • Defining user requirements

  • Working with a very talented technical artist and engineers to develop prototypes

• UX/UI

  • Authored all wireframes, user flows and design artifacts, copy, feedback and metrics strategy for avoiding bias, etc.

  • Authored all production UI assets

• Art Direction

  • Establishing/iterating avatar art style

  • Authored all 3D art assets -- regional templates, facial animation system blendshapes, materials, textures, stylization procedures, etc.

• Technical Direction / R&D

  • Theorizing/designing most features and their high level systems

  • Troubleshooting technical problems with the team, managing backlogs

  • Authoring most pipeline design diagrams, documentation, presentations, research, etc.

• Project Management

  • Tracking work, assigning work, inflection points, etc.

  • Attracting/recruiting talent

  • Engaging/managing contractors

Thanks for reading!