Generate Video

Use Choices when you want to explore a new look from scratch and control appearance, clothing, scene, and style step by step. Use a reference image when you already have a specific AI girlfriend look and want the image or video to stay closer to that face, body, and overall vibe.

Pick one clear action, one clear camera idea, and one clear mood instead of combining too many competing ideas. Short prompts with a focused subject, setting, and motion usually produce cleaner results than overly packed prompts. Starting from a strong reference image also helps videos stay more consistent.

A practical workflow is to first create a few images until you get the exact character design and atmosphere you want. Then switch to video generation and use the best image as the reference so the motion starts from a look you already like.

The biggest impact usually comes from the reference image, visual style, main action, focus, and clothing choices. Keeping those settings aligned, for example realistic styling with believable actions and a simple environment, generally leads to more photorealistic output.

Images require the model to create one frame, while videos require many frames that also need to stay visually consistent over time. Longer clips naturally need more processing, so they take more time than still-image generations.

Yes. Image and video generation both use credits, but image generation is lighter and usually costs less than video generation. On this page, the current credit requirement is shown directly on the generate button before you start.