MCP guides
Generate Veo 3.1 video in Claude: a practical setup
Veo 3.1 is one of the strongest video models around, and the usual way to reach it means a separate Google account, a separate billing setup, and a separate tab open next to your work. StoryStudio removes that detour. You connect it to Claude once, and from then on you ask Claude for a shot and Veo 3.1 runs it, without you ever leaving the conversation.
This guide walks through the setup and a first real shot.
Why Veo 3.1, and when to pick it
Veo 3.1 Fast produces eight second shots at 720p, in landscape or vertical, and it can generate the audio along with the picture. That last part matters more than it sounds. A clip that arrives with ambient sound or a bit of dialogue already feels finished, where a silent clip still needs a pass in an editor.
So Veo is the model you reach for when you want a clean cinematic look with sound on the first try. If your shot needs to land at very high resolution instead, Seedance 2.0 climbs to 4K and is the better call there. Inside StoryStudio you are not committed to either one. You pick per shot and switch when the next shot asks for something different.
Connect StoryStudio to Claude
Open Claude's connector settings and add the StoryStudio MCP server with its URL. That is the whole setup. There is no Google API key to create, no per-model credential to store, and no video app to install beside Claude. The connection stays put, so the next session you just start asking.
If you have used StoryStudio for images already, the same connection covers video. Nothing extra to wire up.
Generate your first Veo shot
Talk to Claude the way you would brief a colleague. Something like: "Generate an eight second cinematic shot of a coffee being poured in slow motion, warm light, with ambient cafe sound." Claude reads the brief, routes it to Veo 3.1 through StoryStudio, and the clip comes back in the chat.
You can be looser than that too. Ask for "a moody establishing shot of a city at night" and let Claude fill the detail, then refine in plain language once you see the result. Because everything happens in one thread, iterating is just the next message, not a new export from another tool.
Image to video, with sound
Veo 3.1 takes a starting image, which is where a lot of real work happens. Drop a product photo or a frame you already like into the chat and ask StoryStudio to animate it into an eight second clip. StoryStudio hosts that reference image itself, for free, so there is no third party file host in the loop and no extra upload step.
Pair this with the audio generation and you go from a single still to a moving shot with sound in one request. For a launch clip or a quick social cut, that is often all you need.
Keep a scene consistent
A single shot is easy. A sequence is where most AI video falls apart, because the character drifts and shot two no longer matches shot one. StoryStudio keeps a story bible with your cast and your world, so the same person stays the same person across every Veo shot you generate. That turns a pile of clips into something that reads as one scene.
If you want the broader picture of how this works across models and editing, the companion guide on how to generate AI video inside Claude covers the full flow from brief to export.
FAQ
How do I generate Veo 3.1 video in Claude? Connect the StoryStudio MCP server to Claude, then describe the shot in plain language. StoryStudio routes the job to Veo 3.1 and returns the clip in the chat, with no separate Google key to manage.
Does Veo 3.1 generate sound with the video? Yes. Veo 3.1 can produce audio together with the picture, so a clip can arrive with ambient sound or dialogue already in place.
Can I turn a photo into a Veo video? Yes. Drop an image into the chat and ask StoryStudio to animate it. It hosts the reference for free and uses it as the first frame for an eight second shot.
Want to generate Veo 3.1 video from your agent? Get started with StoryStudio.