How to make an I Love Smith AI video

1. Pick the right photo

The scene is a tight, close frame, so the face is most of what the viewer sees. One recent, sharp photo per person, looking toward the camera, works better than a stylised one. Daylight or soft indoor light is easier for the model than a hard flash or a lamp directly behind you.

Keep the head and shoulders in frame and avoid anything that hides the face: sunglasses, a mask, a hat pulled low or a heavy beauty filter. Group photos, screenshots and pictures where you are small in the frame give the model less to work with. If you crop, crop close rather than wide.

If the only photo you have is a group shot, crop to one face rather than uploading the whole frame. A still taken from a video is usually softer than a photo, so use one only if nothing else is available.

2. Upload both photos

Sign in with Google, then drop the photo into the generator or choose it from your phone. JPG, PNG and WebP up to 10 MB are supported. If the picture is on your phone, the QR code sends it to the desktop without signing in on the phone.

There is one format: 18 seconds, 16:9, 100 credits. You do not choose a style or a duration, and there is no vertical version. The table below is the whole price list for this video.

The photo stays in your browser until you sign in or start a render. If you send it from your phone by QR code, it passes through our server on its way to your computer and is deleted once it arrives.

There is no style picker and no duration choice, which is deliberate: the scene has one format, and changing it would make the result less recognisable. If you want a different cut, you are looking for a different tool.

3. Confirm consent and generate

Before you render, tick the consent box. It confirms the photo is of you, or of an adult who has agreed, and that the person is not a minor, celebrity or public figure. This is a hard rule; photos that break it are removed.

Start the render and leave the page. The job runs in the background and the finished video appears in My videos. There is no need to keep the tab open, and no fixed time to plan around; the queue decides.

If a render fails, the credits go back to your balance automatically. You can retry with the same photo or a different one.

When it arrives, watch it once before you post. Check the face, the framing and the timing of the look; if something is clearly broken, contact support with the video ID instead of re-rendering blindly.

4. Download and share

When the video is ready, download the MP4. It is 16:9 and already cut to 18 seconds, so it can go to TikTok, Reels or Shorts without another edit. The file carries an AI label; keep it and say the video is AI-generated when you post.

Photos are deleted 7 days after upload and videos are kept for 30 days, so download anything you want to keep. Sharing creates a public link that anyone can open; the link stops working when the video expires or is removed.

If someone in a video asks you to take it down, or you see a version that shows someone without consent, use the report form. Reports are reviewed within 48 hours.

The 16:9 frame suits a normal player and most feeds; if a platform crops it, check the preview before you publish. The clip is short enough to loop cleanly, which is what a trend post usually needs.

Keep the original download. Re-uploading a compressed copy can soften the face, and the AI label in the file metadata may not survive a re-encode.

If you post the clip, keep the caption honest: say it was made with AI and, if you like, link back to the tool. The label lives in the file, but the caption is what people actually read.