How do I make a video of my OC with AI?
If you already have art of your OC, use an image-to-video tool that takes your picture as the first frame or as a reference, such as Midjourney's video[1], Kling 4.0[3] or Seedance through Dreamina[7], and describe the motion in text. If you only have the OC in your head, describe it in detail, generate a reference image first, and animate that.
Route 1: you have a drawing or render of your OC
- Prepare one clean reference. A full-body or half-body image, front or three-quarter view, plain background, the outfit you want in the video. Crop it to the shape you will post (9:16 for TikTok and Shorts).
- Pick a tool that takes images. Midjourney's video is image-to-video: your image becomes the first frame and the prompt is optional[1]. Kling 4.0 takes start and end frames plus up to 10 keyframes[3]. Seedance 2.0 in Dreamina takes up to 9 photos as references[7].
- Describe motion, not appearance. The picture already carries the look. Write what happens: "she turns to the camera, flicks her hair, smirks, the wind lifts her cape".
- Keep it short. One action per 5 seconds. Long action lists make the model rush and drift off-model.
- Check the face at full size before you post. Small details such as eye colour, markings and accessories are the first to change.
For what kinds of drawings animate well, and what happens to line art, see can AI animate a drawing.
Route 2: the OC exists only as a description
Write the character down the way a model reads it: age range, build, face, hair (length, colour, parting), eyes, outfit piece by piece, one signature accessory, then three words of personality. Generate reference images from that, pick one, and treat it as the master picture for every video. Can AI design an original character covers this step, including how to get several options to choose from.
A woman in her twenties: short choppy silver hair with one braid behind the left ear, warm brown skin, amber eyes, a thin scar through the right eyebrow, wiry build. Oversized mustard rain jacket over a black turtleneck, cargo trousers, scuffed red boots, a brass compass on a cord around her neck. Restless, sarcastic, loyal.
Physical details first, then outfit piece by piece, then a signature object and three words of personality. Avoid naming franchises or famous people: in our renders, video models blocked characters that read as either.
Keeping your OC on-model
- Use the same master picture and the same written description in every generation. Do not let the model redesign the outfit; write it out each time.
- Give the model more than one view if the tool allows it. Kling says its Omni Reference takes up to 15 reference assets, including multiple views of a character[4].
- Generate the first frame as an image, then animate it. In our own videos, a clip that starts from a still keeps the face far better than one that only has a text description.
- Design an original face. Characters that look like a famous actor or a well-known franchise character get blocked by video models' copyright filters, which in our renders happened after the whole render had run.
The full method, with what broke for us, is in keeping an AI character consistent.
How long, and what it costs
One generation is short. Seedance 2.0 outputs up to 15 seconds[9], Kling 4.0 from 3 to 30 seconds[5], and Midjourney's video starts at 5 seconds and extends to a 21-second maximum[2]. For longer pieces, render in parts: take the last frame of one clip and use it as the first frame of the next. Free options exist: Dreamina gives free daily credits[8], Vidu says all users get free credits[10], and Kling offers a free tier where exports generally carry a watermark[6].
Where Cast fits
- Your OC is a description, not a drawing: type it in a sentence, Cast draws four looks to pick from (15 credits), and the one you pick becomes the reference for every video.
- Short vertical clips of that OC from one line, in styles such as Influencer, Cinematic, Sakuga Anime, Glam Editorial, Retro VHS, Cozy Vlog, Dark Fantasy and 3D Toon.
- Putting your OC in a scene with up to two other characters.
Where it does not:
- Animating your existing drawing. Cast cannot take an uploaded image; your OC is redrawn from the text description, so it will not match your art exactly.
- Clips longer than 15 seconds, or a full animated episode.
- OCs based on an existing franchise character; in our renders the video model blocked most of those.
People also ask
- Can AI animate my OC from my own drawing?
- Yes, with an image-to-video tool. Midjourney's video uses your image as the first frame[1], and Seedance 2.0 in Dreamina accepts up to 9 photos as references[7].
- What if I don't have a reference picture?
- Describe the OC in detail and generate one first with an image model, then animate it. Picking from several generated looks is faster than trying to get the exact face in one go.
- Why does my OC look different in every clip?
- Usually because the clips start from text rather than the same image, or the outfit is not written into each prompt. See why does my AI character's face change.
Sources
Claims about other companies come from their own pages, read on the date shown. Products and prices change; check the link for the current version.
- Midjourney docs: video (read )
- Midjourney docs: video (length) (read )
- Kling AI: AI video generator (read )
- Kling AI: AI video generator (Omni Reference) (read )
- Kling AI: AI video generator (durations) (read )
- Kling AI: AI video generator (free tier) (read )
- Dreamina: Seedance 2.0 (read )
- Dreamina: home page (read )
- ByteDance Seed: Seedance 2.0 launch (read )
- Vidu: home page (read )
Related answers
- Can AI animate a drawing of my character?
- Can AI design an original character for me?
- How do I keep an AI character consistent across videos?
- How do I make an AI character video?
- Can I turn a picture into a video with AI for free?
- Can I make AI videos of fan characters from anime or movies?
Part of the Cast answers library (Your own characters, OCs and fan characters).
Updated