A collaborative design environment for transmedia workflows.

the vision

Leon Silverman, head of Disney’s Digital Studio, remarked that every production is like a snowflake – each one is unique. Media production workflows are complex and in flux. The choices filmmakers must make to create their project seem infinite and daunting – Which camera? What workflow is best, given the size and scope? Which processes work best in post-production? Studios can afford to fully test multiple cameras and workflows, but DIY and Indie filmmakers cannot. Funded by Google, the Snowflake project was launched to replace this testing process with data-driven color and emotion analyses, and a database of tried and tested workflows.

Media Production

Machine-Learning

Collaboration

the approach

Data-driven tools have been developed to support workflows and decision-making. Most of these tools are trained with large studio production examples, and their AI also caters to large studios. Snowflake aims to support independent artists and student media-makers, who have fundamentally different needs. Optimizing box office revenue or script conformity to formula writing is not the interest of indie, art, and student media-makers – instead, they take risks to explore new forms of storytelling and keep evolving the medium. As a data-driven solution for creative media-making, Snowflake is predominantly a search and analysis tool, rather than a prediction or optimization tool. Decision-making is not handed to AI, but stays in the hands of the creative human; AI supports creators to make informed and inspired decisions. Snowflake uses a hybrid approach, combining machine-learning and AI for search, image analysis, and scene comparison through a rich and evolving database with creative methods and examples.

current efforts

The Snowflake team has been preparing Stage 2 of the project through Spring and Summer of 2022. Stage 2 is a focused effort that will result in two main deliverables: a white paper and an immersive design fiction video. Pioneered in USC’s Mobile & Environmental Media Lab, immersive design fiction is a method of discovery and communication among stakeholders in large-scale projects, using VR to convert design opportunities into embodied experiences, translating user needs for developers.

snowflake-sketch-animation-6

snowflake-sketch-animation-2

snowflake-sketch-animation-3

snowflake-sketch-animation-4

snowflake-sketch-animation-5

snowflake-sketch-animation-1

background

In the 20th Century, there were essentially two workflows for film and television production. Material was shot using one of a limited number of cameras on 35mm negative film (in lower budgets, that might be 16mm film, but the workflow essentially remained the same) or some form of videotape. Film was then processed in a lab where a positive image copy – the “print” – was struck from the negative for editing purposes. When editing was completed, either the negative or original videotape was used to create a matching copy of the edited master, where it was then color corrected, married with a soundtrack, after which multiple copies were created from this master for distribution – either to movie theaters in 35mm projecting at 24 or 25 frames per second, or to television network operations centers for broadcast using videotape. Life was fairly straightforward; those days are gone forever.

There are now a large number of options for digital camera equipment, each with a variety of frame rates, raster sizes, and codecs (compression/decompression schemes). The finished products are distributed in an increasing variety of formats. By one estimate, approximately 400 different versions of a film need to be created to adequately supply typical international distribution needs. Each of these scenarios creates images with different looks and feels, needs different considerations in capturing and finishing sound, and requires very different workflow paths as they go from planning, shooting, editing, finishing, and finally into distribution and exhibition.

The problem for makers of cinema today is that very diversity – what we call “snowflake production.” It is a daunting task for both professional and independent filmmakers to determine the best artistic and financial choices to fulfill their vision and budget. Similarly, educational institutions, faced with the rapid expansion of workflow possibilities demanded by their students, are finding it nearly impossible to keep up in their instruction. At SCA, we create approximately 1,500 projects each year, and it is extremely difficult to give proper guidance to the students and faculty who make those projects – whether they be traditionally time-based media, interactive/gaming projects, multi-channel installations, or fine art. We think Snowflake is a potential answer.