Proprietary AI layered into industry LLMs to provide new solutions
Execution of In-Scene Media uses 4 components
Execution by Rembrand is completed via 4 Components, commencing at the video production phase and incorporating advanced AI technology for a refined and controlled Generative Fusion™ within videos.
1. In-App Experience
Rembrand's proprietary in-app technology drives the overall process, offering unparalleled control and an enhanced Creator experience. This technology facilitates the efficient transfer of video content to computational servers for automated AI-driven placement. The results are returned as lightweight files, easily integrated into the editing project for the final video render.
2. Autopilot for Contextual Augmentation
The Autopilot feature processes videos uploaded by Creators, intelligently identifying the contextually appropriate segments and spatial areas in the video for optimal virtual augmentation (patent pending). This is achieved using advanced 2D and 3D visual perception models that are capable of video and object segmentation, understanding physical environments, and combining audio-visual cues for context awareness. This deep contextual insight is crucial for aligning Creator content with virtual products, pinpointing ideal moments for product animation to maximize brand value.
The image below shows multiple views of a creator’s set as the AI works to determine camera movements, scene geometry, materials, and lighting conditions.
3. Controllable Generative Fusion™
The challenge of placing objects in a visually appealing and photo-realistic manner is addressed by Rembrand’s Generative Fusion™ technology. It accurately matches camera movements, scene geometry, materials, and lighting conditions to ensure products appear as they would in real life. This technology advances current generative models, offering tailored adaptations to each Creator’s setup and control mechanisms to steer clear of unwanted AI generations, ensuring that AI training is not leaked into generated video content.
4. Quality Assurance
The final stage involves an automated verification component. It utilizes both generative and perception models to evaluate the quality of the augmented content and to identify any shortcomings in the previous stages. The outcome of this component is used for further optimization of the generated video. In short, Rembrand takes an existing video (leaving the artistry of the video up to the Creator) and inserts an existing product (leaving the message of the product up to the Brand), but leverages complex AI mechanisms to eliminate the annoying and time-consuming process of manually merging the Brand's product into the Creator’s video.
