24K Magic AI Photo Guide – Choose Your Best Portrait

Use one sharp photograph with one adult face clearly visible. A head-and-shoulders portrait in soft, even light is a practical starting point for 24K Magic AI. Choose a photo you own or have permission to use for an AI-generated performance. The current upload accepts JPG, PNG, and WebP files up to 3 MB, with at least 340 pixels on the shorter edge. Meeting those limits lets the file pass the basic upload checks.

How should the face be framed?

Keep the full face inside the image, with enough space around the forehead, chin, and sides of the head. Both eyes, the nose, and the mouth should be easy to see. A forward-facing photograph or gentle turn gives more facial information than an extreme profile. Avoid cropping through the forehead or chin when you have a clearer version available. Tight crops can remove information needed to interpret the outline of the face.

A full-body photograph is optional for this prepared scene. The generation constructs the moving body and can invent clothing, hands, and shoes that a portrait never showed. A distant full-body photograph with a tiny face may supply less useful facial detail than a closer portrait. Choose the picture where the adult’s face is easiest to inspect rather than selecting the widest possible view of the person.

What lighting should I choose?

Soft daylight from a window or a bright shaded area can illuminate the face evenly. Aim for visible detail on both sides of the face, with skin tones that look plausible in the original image. Strong backlighting can turn a face into a silhouette. A small overhead light can produce dark eye sockets, while direct flash may create glare or hard highlights. Inspect those areas before deciding which photograph to use.

If you have several recent photographs, compare them at a comfortable zoom level. The most dramatic photograph is not always the most informative input. Favor a frame with clear eyes and natural texture. Adjusting exposure slightly may reveal existing detail, but extreme brightening cannot recover information that was never captured. Avoid heavily processed versions when a clean original is available.

Should I remove filters or accessories?

Choose a photograph without heavy beauty filters, face reshaping, or stylized overlays. Those effects can alter the features you want the video to retain. Sunglasses, a mask, a hand over the mouth, or hair covering an eye can hide important information. Transparent prescription glasses may still introduce reflections, so compare an alternative portrait if glare obscures the eyes. Keep the appearance as clear and readable as the available photographs allow.

Expressions also matter. A relaxed face or ordinary smile gives a useful view of the mouth and cheeks. An exaggerated expression may be carried into the generated performance in an unexpected way. Motion blur around the eyes or jaw makes the input harder to interpret. A photo that already looks distorted when viewed as a still image is a poor basis for judging whether the generated clip preserved the person.

How do I check file size and dimensions?

Look at the image information in your phone’s photo app or your computer’s file properties. Confirm the file format, its storage size, and the width and height in pixels. A 1200 by 1600 pixel portrait has a shorter edge of 1200 pixels and meets the dimension requirement. An image measuring 240 by 320 pixels has a shorter edge of 240 pixels and is too small for the current upload limit.

If the file is larger than 3 MB, export a smaller JPG, PNG, or WebP copy using an image tool you trust. Keep the face clear after compression and check the resulting file size again. Renaming the extension does not convert the image format. If your phone stores a photograph in another format, export or convert it into one of the supported formats before selecting it at the Casting Desk.

Can I use a photograph of someone else?

Use another adult’s portrait only when they have agreed to AI video generation and understand the intended setting and audience. Tell them that their face will appear in a fictional dance and that clothing, movement, and other details can be reconstructed. Permission for a photograph to be posted online may cover a different purpose from creating a synthetic performance. Resolve that question with the person before uploading.

Choose private adults and avoid minors or public figures. Group pictures can leave the intended subject ambiguous, so select a photograph with just the person you are authorized to use. Keep private documents, addresses, and other unnecessary personal information out of the frame. The privacy policy explains the browser draft, generation processing, private delivery, and support request options.

What should I do at the Casting Desk?

Sign in, select your prepared portrait, and inspect the displayed image before continuing. Check that you chose the intended file and that the face remains visible after the preview is fitted into the upload area. Read the current clip settings and credit quote. The default is a silent, ten-second 720p video for ten credits. A ten-second 480p request costs five credits. Choose the resolution and frame format before checking the live quote.

Confirm the photo permission checkbox yourself. If your balance is insufficient, review the credit purchase offer or compare the available packs. After payment, return and confirm the generation when the portrait is ready. A local photo draft can be restored in the same browser for twenty-four hours, but you should retain the original file independently so a cleared browser or different device does not interrupt your preparation.

How should I evaluate the finished video?

Compare the moving face with the input across the whole clip. Check the opening, turns, distant views, and final pose. Also inspect hands, clothing, and background details that the portrait did not define. AI output can vary even when a file meets every upload requirement. Improving an unclear input is a practical step, while repeated generation still uses additional credits and cannot promise an exact match.

Download a result you want to keep within its fourteen-day delivery window. If the task reports a confirmed technical failure, its reserved credits return to the account. For unresolved processing or a specific problem in a finished file, use account support with the task identifier. Read what the service creates or how free access and credits work, then return to your portrait.