VIVACE - Visualization, AI Generation, and Capturing of Assets, Characters, and Environments
Design tech-minimal and artist preferred digital content creation workflows, enhancing realism and 3D asset procurement efficiency, democratizing innovative digital storytelling.
Realistic, efficient, and scalable production of 3D environments
The VIVACE project aims to transform digital content creation by developing artist-preferred, tech-minimal workflows that enable realistic, efficient, and scalable production of 3D environments and virtual characters. The research addresses key bottlenecks in current media pipelines, namely, the lack of realism, inefficiency in asset generation, fragmented workflows, and limited tool adoption, especially within small-scale and budget-constrained markets like Flanders.
The project’s core objective is to acquire knowledge that bridges cutting-edge 3D capture techniques and generative AI with intuitive, creator-friendly tools. This includes innovations in photorealistic 3D reconstruction, dynamic relighting, and controllable avatar generation using Gaussian Splatting and multimodal AI inputs. VIVACE will also explore novel methods for partial regeneration, style-consistent asset collections, and real-time scene composition.
At the product level, VIVACE will deliver integrated toolkits for virtual production (VP) and virtual influencer (VI) creation, enabling rapid, high-quality output. At the process level, it will streamline workflows by reducing manual effort, setup time, and iteration cycles, making complex 3D tasks accessible to non-technical creatives. At the service level, it will empower media studios to produce immersive content with greater flexibility and lower cost, unlocking new storytelling formats and market opportunities.
By validating these innovations through real-world use cases and performance benchmarks, VIVACE will position Flemish media creators to compete globally, while fostering sustainable growth in local creative industries.
The consortium
The project industrial partners are:
- Datameister
- EMG Belgium
- VRT
- Uncanny The academic partners are:
- IDLab-MEDIA
- Howest - Digital Arts & Entertainment
For this project, we received imec.ICON funding from VLAIO and imec, for the 2025 call for proposals. The project will run for two years, from Feb. 2026 until Jan. 2028.
