In the coming decade, machine learning and artificial intelligence will transform all aspects of film production.
There is a tremendous opportunity today to develop a community of artists and technicians who are skilled in new ways of creating powerful images...
‍
...alongside their machines.
The following is a description of a new kind of company that leverages machine learning to serve film and game production.
‍
We are currently developing two specific processes that meet existing needs in these industries.

We have sufficient funding now for two years of research and development.
‍
We are currently seeking to expand our engineering talent and to connect with industry partners who will love what we are building.
Incorporating machine learning into film post production promises not only to reduce costs but also to improve quality while freeing filmmakers and designers from cumbersome limitations.
But this is not magic.
This company will train neural networks to do labor intensive tasks but it will also train human beings
to build, select and apply those networks in a creative process.
We are building
Machine Learning Assisted Compositing
Most film production requires replacement of background elements and other types of moving image editing. Traditionally this can require cumbersome techniques such as a green screen or painstaking human labor known as rotoscoping.
‍
We propose a novel method that can dissect film footage into foreground and background elements and predict occluded regions using automatically generated training data and transformer based neural networks.
We are building
Neural Network Based Performance Capture
Both film and game production require systems that can capture an actors performance and transfer it to a computer generated character. Many of the existing methods for doing this create encumbrances that increase production costs as well as interfere with actors.
We propose a novel method that can recognize and recreate individual performances based on the conditions that existed at the time of filming using a combination of synthetic training data and generative adversarial networks.
We believe the resulting captured performance will not only be easier to record but also potentially a better representation of the human nuance that can allow these characters to cross the “uncanny valley” in emotion and realism.
We Are Hiring
‍
Currently seeking talented engineers and artists for an in person research and development team.
Please send resumes and a description of how you fit this specific effort to manifoldvalley@gmail.com
CG Artists/3d Programmers
CG artist with strong programming skills for an in-person team development effort and to continue with us into our involvement in production.

A strong candidate has knowledge of Blender, Houdini, Unreal Engine and/or similar platforms. Experience with script based manipulation of mesh data structures and textures.

Experience with scanned asset acquisition and manipulation, character rigging, animation, lighting, rendering.
Machine Learning Engineers
A strong candidate has knowledge of deep learning, generative networks, adversarial networks as well as a background in graphics programming, image classification, neural network model architectures and training techniques as well as frameworks/languages such as Tensorflow, Python, CUDA, C++.
Photogrammetry Hardware Developers
Seeking talented hardware and software engineer.

The initial project for this job is to select technologies for and build a networked multi-camera photogrammetry capture system, but we hope to find a candidate that can expand into other programming roles and image capture roles.
‍
Skills required C++ or similar, hardware interfacing, networking.

An excellent candidate has knowledge, background and personal interest in photography, cinematography, computer graphics and visual effects.
Who Are We
Ian Bloom
Founder/Lead Developer
Ian has a unique combination of skills. A computer programmer since childhood, his family home was also an engineering firm. He has continued to develop those skills but has worked as a cinematographer and serial entrepreneur in the film industry for most of his life.
Over the course of his career Ian filmed over a dozen feature films and countless commercials. He also started a small software company which pioneered 4K raw workflows and started a post production facility in New York. His work combines a deep respect for cinema with a knack for technology.
For the past several years Ian has been focused on applying the science of machine learning to film production. His most recent software project is a GPU based rendering engine capable of producing labelled images for neural network training.
Tyler Jackson
Business Development Advisor
A veteran film producer based in Los Angeles California, Tyler is frequently on buyers side of visual effects production. With over 11 feature film producing credits including "Dragged Across Concrete", "What They Had", "Rudderless" and countless other roles in film production and post-production, Tyler is an excellent source for getting our processes to market. Formerly a Canadian and currently a great dad.
Contact
‍