SOP: Content & Video Production Workflow
Status: DRAFT (V1.0)
Last Updated: 2025-12-22
Owner: Nqobizitha Mlilo
🎬 Core Video Stack
Our video production is built on a high-fidelity pipeline for editing, motion graphics, and audio.
- DaVinci Resolve: Primary nonlinear editor (NLE) for assembly and color grading.
- Fusion Integration: Moving primary Motion Graphics and VFX work into the Fusion page for node-based control.
- Cavalry: Used for real-time 2D procedural motion graphics.
- Blender: Used for 3D elements or hybrid 2D/3D visual effects within video content.
🎵 Audio & Voice Pipeline
Audio is treated as a core production element equal to visuals.
- Ableton Live: Used for music production, soundtrack scoring, and final audio narration processing.
- AI Voice Synthesis:
- ElevenLabs: Primary tool for high-end AI narration and character voices.
- Google Cloud TTS: Secondary tool for scalable voice content requirements.
🔄 Production Lifecycle
- Scripting & Audio First: naration is generated (ElevenLabs) or recorded, then timed in Ableton to create the "heartbeat" of the video.
- Visual Asset Assembly: Bringing in Moho/Blender animations or live-action footage.
- Editing (Resolve): Assembling the visual story on the timeline.
- Motion Graphics (Fusion/Cavalry): Layering in typography, trackers, and VFX overlays.
- Color & Export: Final grading in Resolve to ensure the "Nafuna Look" is consistent across platforms.
🤖 AI Video Production Pipeline
For Music Videos (Internal & Client), we use a cutting-edge AI generation pipeline that prioritizes consistency and cost-efficiency.
1. Pre-Production: Weavy Storyboarding
- Platform: Weavy (Primary storyboard orchestrator).
- Models:
- Flux: High-fidelity base images.
- SeedDream / Nano Babana: Specialized image editing and style-consistent character models.
- Goal: Create a locked visual sequence before moving to motion.
2. Production: AI Video Generation
- Platform: RunningHub (Used for cost-effective credit management).
- Models:
- Wan 2.2: The "Minimum Standard" for standard movement and texture.
- Wan 2.5: Used for complex shots and higher fidelity temporal consistency.
- Wan Infinite Talk: Specialized for Lip-Sync and performance-driven shots (as used in Koji - Ndiye One).
3. Post-Production: High-End Refinement
- Edit: Assembled in DaVinci Resolve.
- Upscaling: Refined passes through Topaz or internal Resolve Superscale if required.
- Overlay: Final lighting and texture overlays added in Fusion to blend AI generations with live-action or clean NLE plates.
📈 Quality Standards
- Temporal Stability: No jittering or "AI hallucinations" in final delivery.
- Audio Clarity: No video leaves without an Ableton-processed audio pass.
- The "Nafuna Look": High-contrast, vibrant color grading established in Resolve.