Easy AI VFX You Can Add to Any Video

Building the Signature Intro Effect

Matt Wolfe opens by explaining his signature intro effects, the most-requested technique from viewers. He starts in DaVinci Resolve by capturing two frames: an empty room shot before he sits down, and a second frame right before he starts talking. He uploads both frames to Runway's keyframe animation feature and prompts the AI with something like 'the man burst through the back wall and then sits down in the chair.' Seed Dance 2.0 is his preferred model, though Kling and VO3.1 also work well for this technique.

Polishing Transitions and Creative Variations

After generating the AI video, Matt brings it back into DaVinci Resolve and uses the 'smooth cut' transition to blend the AI-generated clip with his actual footage. He shares creative variations he has made, including a claw machine that picks him up and deposits him in his chair, and a werewolf morph where he used ChatGPT to generate a wolf image as his starting frame. The animal-to-human morph remains difficult for current AI models, requiring creative workarounds.

Location Transitions and Teleportation Effects

When Matt knows he will be filming in a new location, he records a short intro in his studio mentioning the scenery change, then uses the same keyframe technique to transition between locations. He shares an unreleased NAB demo where he teleports from San Diego to Las Vegas, and another where a tossed microphone spins in slow motion before landing in his hand at the destination. Wearing the same shirt in both locations helps sell the illusion.

Background Effects with Google's Omni

Google's Gemini with the Omni model excels at adding elements to existing footage without altering the main subject. Matt demonstrates a Yeti walking through the background while he talks, Godzilla stomping through a Google IO crowd scene, explosions behind him, smoke rising from his head, and even weather transformations that add rain and umbrellas to a sunny scene while preserving the original location.

Generating B-Roll with Claude Code

For custom B-roll like animated website highlights, Matt uses Claude Code (with Opus 4) or Fable to generate code that takes screenshots, scrolls, and highlights specific text. The result is an MP4 video he can drop into his edits. Fable produces smoother animations, but both tools create usable B-roll without hunting through stock video sites. He also generates cheesy corporate B-roll by prompting video models with standard stock footage descriptions.

Logo Reveals, Text Animations, and Lower Thirds

By installing the 'Remotion best practices' skill in Codex or Claude Code, Matt generates logo reveals and text animations through code. He demonstrates particle explosions reforming into his Future Tools logo, paint splatters morphing into his avatar, and thought bubbles appearing over his head. For lower thirds, VO3.1 handles text more accurately than Seed Dance, though it may still take several attempts to spell everything correctly.

Motion Graphics for Explainer Videos

Remotion in Codex produces solid motion graphics for explainers: mock text message exchanges, visual explanations of SVG images, stock charts showing Nvidia's growth, and animated diagrams. NotebookLM's video overview feature also generates usable animated B-roll from source documents, including charts and illustrations that work well for documentary-style content.

Character Animation and Matt's Philosophy on AI VFX

Matt recreates a text exchange between Sam Altman and Mira Murati by recording himself reading both sides of the conversation, then using Runway's character script-to-video feature with uploaded images. He edits the two outputs together to create a back-and-forth dialogue. He emphasizes using AI for about 5% of his content to add entertainment value while keeping his videos 95% human, purposefully making the AI elements obvious so viewers never feel deceived.