Getting started with video generation in VIGPT

VIGPT can create a short video from text. Unlike a text answer, which comes back immediately, video generation takes time and the result is ready a few moments later.

This feature is part of the Pro and Max plans. Below is the whole path, from start — from switching to the right mode to getting the finished clip.

Prerequisite: a Pro or Max plan

Video generation isn't available on the free plan or on Standard. For it you need Pro or Max. If you're on a plan without the feature, when you try to generate a video you'll get a message saying you need to upgrade to Pro or Max to continue.

Each plan has a monthly credit. Simple messages use a little, and heavy work — including video generation — uses more. Max has a larger monthly credit than Pro, so if you generate videos regularly, that difference shows.

Prices and each plan's credit may change. For current numbers, see the pricing page.

Switching to video mode

In the message bar, choose video generation from the modes. The message bar turns into a video editor: right where you write the prompt, options for aspect ratio and duration now appear beside it.

This is a separate mode, not an ordinary conversation. While you're in it, each send is a video-generation request rather than a text answer. To go back to a normal conversation, just clear the selected mode.

Writing the prompt

The prompt is the description of what you want the video to show. The more clearly you describe the scene, the closer the result is to what you had in mind. Say the subject, the motion, and the mood:

• "A coffee cup on a wooden table, steam rising gently, soft morning light."
• "Waves breaking on a sandy beach, close-up, golden sunset light."
• "A piece of paper spinning in the wind on a street, the camera following it slowly."

A specific sentence with one clear motion usually works better than a long, detailed list. If the result isn't what you wanted, sharpen the description a little and generate again.

Choosing the aspect ratio and duration

Before you send, you set two things:

• Aspect ratio. Three options are available: 16:9 for landscape video, 9:16 for portrait (suited to mobile and stories), and 1:1 for square.
• Duration. Videos are short, on the order of a few seconds. The duration options vary by model and plan, but on many models something around 4 to 8 seconds is available.

Both options have a sensible default, so if you're not sure, you can leave them and experiment later.

Starting from an image (optional)

Instead of starting from scratch, you can attach a reference image so the video is built from it — a still photo you want to bring into motion, say. The text description still helps and sets the direction of the motion.

The thing to know is that not all models are the same: some accept a reference image as an optional guide, and some only build video from an image and won't work without one. If a model needs an image and you haven't attached one, you'll get a message saying a reference image is required.

The result arrives asynchronously

After you send, the video generation doesn't finish immediately. The job goes into a queue and is processed in the background; its status moves from "queued" to "processing" and finally to "ready."

You don't have to wait and stare at the screen. When the video is ready, the clip appears in that same conversation. You can go do something else and come back.

If you stay on Auto

You don't need to pick a video model yourself. When you stay on Auto, the model choice is Auto's job once you switch to video mode, and you just provide the prompt and the options.

If you've manually pinned a model that can't generate video, instead of rejecting the request, VIGPT builds it with a suitable model and tells you it did so; your pinned model doesn't change for later conversations.

Finer controls — like quality, camera motion, and each model's dedicated settings (Media Studio) — are part of the Max plan. You don't need any of them to start.

In short

Switch to video mode, describe the scene in one clear sentence, pick the aspect ratio and duration, and attach a reference image if you want. Then send and go do something else — the video is built in the background and the finished clip appears in the same conversation.