Can AI design an original character for me?
Yes. Describe the character in one detailed paragraph (face, hair, outfit, one signature detail, personality), have an image model draw several versions, and pick the one that feels right; that image then becomes the reference for every picture or video of the character. You shape the design by what you write and what you choose, but under US guidance a prompt alone does not make you the author of the image[1].
How to describe a character to AI
Image models draw what you specify and average out what you do not. A vague prompt ("a cool fantasy girl") gets a generic face. Write it in this order:
- Who and age range: "a woman in her twenties", "an old fisherman", "a small ginger cat".
- Face: skin, eye colour and shape, brows, one distinctive feature such as a scar, freckles or a gap tooth.
- Hair: length, colour, texture and parting.
- Outfit, piece by piece: top, outer layer, bottoms, shoes, one accessory. Name materials and colours.
- A signature detail: the thing people will recognise first, such as a compass necklace, a single red glove or a cracked visor.
- Personality in three words, which shapes the expression and pose: "proud, sarcastic, secretly kind".
- The style: photoreal, anime, 3D animated feature, painted fantasy. Pick one and keep it.
A small ginger cat with one torn ear and bright green eyes, wearing a tiny red captain's coat with brass buttons and a battered tricorn hat. A silver earring in the torn ear. Swaggering, greedy, brave when it counts. Premium 3D animated-feature style, plush fur, warm bounce light, full body, plain background.
Two lessons from making our own characters. Ask for a full-body reference on a plain background first; scenes come later. And watch the wording: on Midjourney, a prompt with an exact teenage age was blocked where "a woman in her twenties" passed, and very long prompts packed with pose and body detail failed.
Getting several options
Generate a batch and choose. Most image models return several images per job; every Midjourney job we run returns four. Do not try to get the perfect face in one prompt. Pick the closest version, then change one thing at a time in the description. When you have the one, freeze it: save that image as the master reference and save the exact description next to it.
Design an original face. In our video work, a character whose face resembled a famous actor got blocked by the video model's copyright filter again and again, and redesigning the face was the only fix. The same goes for characters that echo a famous franchise; see fan character AI videos.
Then animate it
The master image is what makes a character usable in video. Image-to-video and reference-to-video models start from it, which keeps the face stable across clips. How to make a video of my OC covers the animation step, and keeping an AI character consistent covers what we learned keeping three characters on-model across dozens of videos.
Who owns an AI-designed character?
- The US Copyright Office concluded that prompts alone do not give enough human control to make you the author of the output[1], and that purely AI-generated material is not protected[2].
- Your own contributions can be protected: human selection, arrangement or modification of AI outputs, and anything you draw or write yourself[3]. Whether that is enough is decided case by case[4].
- The tool's terms decide what you may do with the images. Midjourney says subscribers may use outputs commercially, but companies with over $1,000,000 in yearly gross revenue must be on Pro or Mega[5].
- In practice, the more you redraw, edit and build the character's story yourself, the stronger your claim to it.
Where Cast fits
- You want a character designed and animated in one place: describe it in a sentence, Cast draws four Midjourney looks to choose from (15 credits), and the one you tap becomes the reference for every video.
- You want to pick a style up front from 8 options, including Sakuga Anime, 3D Toon and Dark Fantasy.
- You want to see the character move quickly: one line of text gives a 9:16 clip of 5, 10 and 15 seconds.
Where it does not:
- Iterating on a character from your own sketch; Cast cannot take an uploaded drawing.
- Turnaround sheets, multiple poses or print-resolution art; Cast draws four looks for video use.
- Characters based on real people or existing franchises.
People also ask
- How do I describe a character to AI?
- Age range, face, hair, outfit piece by piece, one signature detail, three personality words and a style, in one paragraph. Specific beats long: name colours and materials rather than adjectives.
- Can I get several options to choose from?
- Yes. Generate a batch, pick the closest, and change one detail at a time. Each Midjourney job we run returns four images.
- Can I then animate the character?
- Yes: use the chosen image as the first frame or reference in an image-to-video model. See how to make an AI character video.
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.
- US Copyright Office: Copyright and AI, Part 2 (read )
- US Copyright Office: Copyright and AI, Part 2 (purely AI material) (read )
- US Copyright Office: Copyright and AI, Part 2 (human contributions) (read )
- US Copyright Office: Copyright and AI, Part 2 (case by case) (read )
- Midjourney docs: comparing plans (read )
Related answers
- How do I make a video of my OC with AI?
- Can AI-generated videos be copyrighted?
- How do I keep an AI character consistent across videos?
- How do I create an AI influencer for free?
- Can I make AI videos of fan characters from anime or movies?
- How do I make an AI character video?
Part of the Cast answers library (Your own characters, OCs and fan characters).
Updated