AI Image to Video

Upload the photo at the top of this page, type how it should move, choose 480p or 720p and press Make the video. The model (GROK Video 1.5) uses your photo as the first frame and generates six seconds of motion from your description. The clip opens in a chat where you can ask for changes, a longer version or a different movement.

Yes. A clear, front-facing photo works best. Describe small, natural movements: a smile, a blink, a slight turn of the head, hair in a breeze. The instruction sent with your photo asks the model to keep your face, clothes and background as they are, but AI video is not exact: check the result before you share it, and run it again if something looks off.

New accounts get free welcome credits with no card, and the amount depends on your country. A 6-second clip costs 24 credits at 480p or 48 at 720p, so the welcome credits may cover your first clip. There are no daily free credits. After that the smallest pack is $9 for 400 credits, about eight 720p clips or sixteen at 480p. If a generation fails, the credits are returned.

Yes, and it is one of the most common uses. For a faded or damaged print, run it through the photo restore tool first so the model has a clean face to work with, then upload the restored image here and type something like: she gently smiles and blinks, with a slight movement of her head, keep the vintage look.

You can ask for the mouth to move, but this page does not lip-sync to a script or an audio file, and any sound in the clip is generated by the model, not your voice. For a talking-head video with your own words, a dedicated lip-sync tool is the better choice.

This page makes 6-second clips, the shortest and cheapest length. In the Agent chat the same model goes up to 30 seconds, priced per second, and other models offer their own lengths and looks. The chat shows the price before you send.

An MP4 at 480p or 720p from this page; 1080p is available in the chat. The frame follows your photo: landscape, portrait or square. You can download the file from the chat and share it anywhere.

Describe the movement, not the photo. One or two motions in plain words work best: she smiles and turns her head slightly; the camera slowly orbits the shoe; clouds drift and the water ripples. Add what must stay still if it matters. Long, complicated instructions give the model more chances to get something wrong.

If the generation fails, the credits are returned automatically. If it finishes but the motion is not what you wanted, that run is still charged: try again with a simpler or more specific motion, or ask the Agent in the chat for a variation. Two or three runs are normal for a clip you want to keep.

Video downloads carry no watermark. You own what you create and may use it for personal or business projects under our Terms of Service, provided you have the right to use the photo and respect the rights of the people in it. Check the result before publishing: AI can still distort hands, text or logos.

The clip starts from your photo as its first frame, and the instruction sent with it tells the model to keep the person's face, clothing and background while adding the movement, so the likeness is usually close. Results vary, most of all with blurry, damaged or very small photos, and small differences can appear during the motion. For the best match, restore and sharpen the photo first, keep the movement small, and compare the clip with the original before you share it.