Getting started in Most AI: balance, payments and history

An image edit, a video from a photograph and a narration are different workflows. Begin with the intended result and the source files you have, then open a compatible model. Check its actual name in the studio; a family name on a landing page can cover several versions.

For your first test, avoid combining several difficult requirements. A short scene, one image revision or a small speech excerpt gives you a result that can be evaluated against the brief.

Use the payment and balance information currently shown in your account. Media cost can depend on model version, resolution, duration or text length. A starting price in a catalogue is not the price of every possible configuration.

Text-model usage may use separate message-credit conditions. Do not infer that a displayed media balance gives unlimited access to every text model, or that a free attempt is available for every account. Read the selected form before submitting.

Store the accepted result alongside the original sources and the brief. Note the model, version and settings that mattered. A downloaded file without its inputs may be difficult to revise later.

For referenced work, a prompt alone does not describe every input. Keep permitted source files and their roles clear. Review the full clip, image or recording before using it in a project.

• Use descriptive file names rather than relying on thumbnails.
• Keep the source and accepted output in separate folders.
• Record the reason a result was rejected so that the next attempt has a specific goal.

If the face changes, reduce the motion or improve the source. If narration is too long, shorten the script before increasing speaking speed. If a product label is wrong, plan an exact finishing edit.

Change one requirement at a time when comparing attempts. The purpose of a first test is to establish a useful workflow, not to prove that every output will work without review.