In this tutorial, Josh guides viewers through creating high-quality digital doubles using AI technology. By following his detailed workflow, users will learn how to record themselves with optimal lighting, camera angles, and techniques to capture natural movements. The process involves creating multiple avatar variations with a consistent naming system, allowing for seamless video production and editing. After completing the tutorial, viewers will be able to generate professional, versatile digital avatars that can be used across different video projects with ease and consistency.
Following are the key things you will be able to do after you watch this demo:
Create multiple avatar variations with a consistent naming system
Record high-quality source footage for AI digital double training
Select optimal recording environments (green screen or natural settings)
Capture multiple camera angles for flexible video production
Apply three-point lighting techniques for professional video quality
Use camera settings to record in 4K resolution
Develop a systematic approach to avatar creation and management
Experiment with different avatar styles and gestures
Optimize video recording for AI digital double learning
Implement a multi-camera editing workflow for seamless avatar transitions
Building Digital Doubles from Scratch 0:08
Josh Lomelino explains the importance of following earlier steps, especially around lighting and image quality, to avoid costly post-production fixes.
He emphasizes the need for a two-minute video of oneself speaking directly to the camera, suggesting the use of a wireless mouse for discreet recording.
Josh prefers recording against a green screen for flexibility in background changes, but acknowledges the natural setting option.
He recommends experimenting with different avatars, using a consistent numbering system for organization, and provides examples of naming conventions for avatar variations.
Creating and Managing Avatars 3:19
Josh discusses the importance of capturing as many versions as possible for each outfit in one session to ensure consistency in hair, lighting, and clothing.
He explains his approach to recording multiple shots or angles simultaneously using different camera angles and a multi-cam edit in video editing software.
The three essential angles he always records are a close-up, a medium shot, and a three-quarter side view.
Josh mentions the challenges some AI tools pose with the three-quarter view but recommends capturing it for added realism and variety.
Recording and Equipment Considerations 4:43
Josh advises using a Logitech 4K webcam for better image quality, though a 1080p camera can also yield decent results.
He shares his experience with different recording devices, including a phone's rear-facing camera in 4K, a webcam, and a DSLR, and emphasizes the need for experimentation.
Josh recommends using the built-in Windows or Mac camera app for recording at the highest resolution possible, with instructions on adjusting settings to force 4K recording.
He advises recording a clip without the green screen, looking straight into the camera, and speaking casually to ensure the digital double learns natural behavior.
Batch Creating Avatars 6:07
Josh introduces a workflow in his video editing software for batch creating avatars, which speeds up the process.
He mentions the importance of recording a clip that is at least two minutes long to avoid issues with awkward movements being mimicked by the avatar.
Josh explains his setup for recording, including using an adjustable camera arm mounted to his desk for flexibility.
He concludes the demo by stating that he will cover more in the next video, indicating the end of the current session.