FaceSwap Artist - all levels - London Studio

Position SummaryFaceSwap is a new, exciting, and growing discipline within the visual effects industry, using Machine Learning (AI) to create digital faces and likenesses. We are currently looking for people with a keen eye for, and interest in, photo realism, as well as those comfortable with the technical side of visual effects. The FaceSwap Artist will  be a pioneer and trailblazer in a new area of visual effects technology. While a background in compositing is advantageous, artists with visual effects experience in any discipline are encouraged to apply.What You’ll DoUse a variety of deep learning based technologies to produce photorealistic images, including open source tools, and proprietary in house softwareCreate templates to facilitate model evaluationMonitor the training progress, and assess improvements and quality; making necessary adjustments to the configuration of trainingPrepare datasets to train models, which includes gathering reference material from various sources, curating and managing imagesUsing trained models to perform faceswaps in a shot context, producing slap comps for review in dailies.What We’re Looking ForExperience in visual effects or CG industryA curious mind that is interested in learning new things, and enjoys experimentation and testingFlexible and adaptable to changing workflows, detail oriented and meticulousWorking knowledge of Nuke to prepare precomps and slap compsSome experience with CG compositingComfortable working with or learning command-line based tools Interested in, or have basic familiarity with, deep neural networksFamiliarity with ComfyUI and other GenAI tools Nice to haveExperience working on projects with realistic looking CGI humans or other face replacement approachesAny experience using or training deep neural networks models, such as Nuke Copycat, DeepFaceLab or other face swapping toolsExperience with Keen ToolsPython scripting skills to automate and streamline high volume of repeated day-to-day tasksReference: 10160278