KreativeCore
AI-powered visual search that turns hours of footage scrubbing into seconds of natural language queries. "Find person smiling" - done.
Coming to Adobe ExchangeSearch with natural language: "beach sunset", "person running", "product close-up". Vision understands context, not just keywords.
Automatically cluster faces into "Person 1", "Person 2". Rename once, and Vision updates all clips featuring that person.
Visual landmark recognition identifies locations. Tag once, search forever. "Show me all Bosphorus Bridge shots".
Automatically categorizes shots: Wide, Medium, Close-Up, Steadycam, Handheld. Perfect for finding coverage.
Search results auto-generate a new sequence with In/Out points set. One click from search to stringout.
All AI runs on your machine. No cloud uploads, no credits, no internet required. Your NDA-protected footage stays private.
Select clip, sequence, or entire project. Vision analyzes with CLIP AI and InsightFace models - running locally on your GPU.
Type what you're looking for: "happy crowd", "interior office", "sunset timelapse". Vision searches semantically, not just tags.
Click any result to preview in Source Monitor. Double-click to load with In/Out points. Refine face/location tags as you work.
Hit "Create Sequence" (The Lifesaver). Vision generates a stringout with all search results, perfectly timed. Edit from there.
We leverage the same models used by OpenAI, Google, and Meta - optimized for local deployment. No compromises on accuracy, no cloud dependency.