Image to Video AI

You give the model a still image and a short prompt. The image is used as the first frame, so the subject, the colours and the composition come from your photo. The model then invents the motion that follows: a camera move, drifting steam, moving water, a turning head.

That is why photo-first clips tend to look more consistent than text-only ones. With text to video the model decides what the subject looks like. With image to video you have already decided.

The model can only move what it can see clearly, so the source image matters more than the prompt.

• One clear subject with room around it in the frame, so a camera move has somewhere to go.
• Sharp and well lit. Blur and heavy noise in the photo tend to carry into the clip.
• Things that already suggest motion: water, steam, hair, fire, clouds, rain.
• Avoid photos that rely on readable text or logos; small lettering often warps as things move.
• Only use photos you have the right to use, and only people who have agreed to it.

Skip describing what is already in the photo. Name the movement and the camera instead, as the prompts on this page do: the balloon rises slowly as mist drifts across the hills, or slow aerial push forward over the cove. One move per clip works better than a list of actions.

If the first result drifts, shorten the prompt, ask for a slower move, or run the same photo on a different model. A cheap model is a good way to test the idea before you pay for a longer or sharper render.

Image to video uses paid credits, like all video on Lovino. The 3 free daily credits cover image tools only. Credits come from a plan (from $7.99 a month). Anyone can buy a credit pack, with or without a plan. A starter pack is $2.99 for 15 credits.

Prices start at 2 credits per clip and rise with length, resolution and model. The clips above used Wan 2.2 Fast, which costs 2 credits at 480p and 3 at 720p.