Short answer: Attach your photo and say "the subject from the reference photo", then describe the anime look instead of just naming it: "clean ink outlines of varying weight, flat cel-painted colour fills, simple two-tone shadows with hard edges". List what must stay recognisable (hairstyle, eye shape and colour, skin tone, the outline of the outfit) and what must go ("no photographic texture"). Add a setting, the light, "no text", and the aspect ratio.
"Make me anime" works, sort of. You get a generic face with big eyes that could be anyone, often still half a photograph. The difference between that and a frame that looks drawn by hand is three things: what you keep of the person, how precisely you describe the drawing style, and what you rule out.
How do you keep it looking like you?
Name the features the model should carry over, and let it redraw everything else. The gallery's cel prompt puts it like this:
A clear, front-lit photo of one person helps more than any wording: one face, no sunglasses, nothing covering the hair. Gemini's documentation describes exactly this kind of edit: "Provide an image and use text prompts to add, remove, or modify elements, change the style, or adjust the color grading."
How do you describe an anime style without naming a studio?
Describe what the style looks like. A studio name gives the model a vague memory of many films; a description gives it something to draw:
| Instead of | Describe |
|---|---|
| A famous studio's look | "hand-painted watercolour backgrounds, warm pastel palette, hand-drawn line art, nostalgic summer light" |
| "90s anime" | "cel-shaded, VHS colour palette, grainy film scan, hand-painted background" |
| "Manga" | "black-and-white panel, screentone shading, speed lines, bold ink line art" |
It also keeps your picture your own: every gallery prompt ends with a line asking the model not to copy any existing film or character.
Photo-to-anime prompts to copy
All of these are in Prompt Gallery, written for a reference photo; attach yours and paste the prompt. The pictures show what each made.
Hand-painted 1990s anime cel (Midjourney, Flux, Nano Banana):
A single frame from a hand-painted 1990s animated film, featuring the subject from the reference photo as the character.
Translate the subject into animation rather than copying the photograph: keep the recognisable elements (hairstyle and its parting, eye shape and colour, skin tone, the silhouette of their outfit, any distinguishing marks) and redraw them as an animated character with clean ink outlines of varying weight, flat cel-painted colour fills, and simple two-tone shadow shapes with hard edges. No photographic texture, no digital gradients on the character, no airbrushed shading.
Setting: a quiet train platform in early summer, at late afternoon, low golden sun. The background is a separate painted layer: soft gouache and watercolour, richer in detail and texture than the character, with visible brushwork in the clouds and foliage and gentle atmospheric haze toward the horizon. The character reads as drawn ON the painted background, exactly as traditional cel animation looks.
Composition: a wide, calm establishing frame. The character sits or stands off-centre, small in the frame, with a lot of sky or architecture around them. Warm light rakes in from one side and blooms very slightly.
Mood: nostalgic and unhurried. Include the small imperfections of the medium: a faint cel-edge line, gentle film grain, slightly imperfect registration.
Widescreen 16:9, film still, hand-drawn animation, not 3D, not photorealistic.
Keep it original: do not depict any celebrity or real public figure other than the subject of the reference photo, and do not copy any existing painting, artwork, film or game character, mascot, logo or trademarked design.Modern anime (written for Midjourney Niji 6; the closing --ar 2:3 --niji 6 is Midjourney's, so in ChatGPT or Gemini write "Aspect ratio 2:3" instead):
Anime-style illustration of the subject from the reference photo, a dynamic action pose, a cherry-blossom courtyard, cel shading, vibrant colours, crisp clean line art, expressive eyes. Keep it original: do not depict any celebrity or real public figure other than the subject of the reference photo, and do not copy any existing painting, artwork, film or game character, mascot, logo or trademarked design. --ar 2:3 --niji 6Countryside anime painting:
Hand-painted countryside anime illustration of the subject from the reference photo in a windswept green field, soft watercolour backgrounds, warm pastel palette, hand-drawn line art, nostalgic summer light. Keep it original: do not depict any celebrity or real public figure other than the subject of the reference photo, and do not copy any existing painting, artwork, film or game character, mascot, logo or trademarked design. --ar 16:9 --niji 6Chibi:
Cute chibi version of the subject from the reference photo, oversized head and tiny body, big sparkling eyes, soft pastel colours, gentle cel shading, kawaii sticker style, clean background. Keep it original: do not depict any celebrity or real public figure other than the subject of the reference photo, and do not copy any existing painting, artwork, film or game character, mascot, logo or trademarked design. --ar 1:1 --niji 6The gallery has three more anime styles with results: manga ink, retro 1990s anime and Korean webtoon.
Why did my anime picture come with Japanese or Korean text?
Because the model copies what it knows about the medium. Ask for a "90s anime screencap" and it may add a subtitle line; ask for a "webtoon" and it may add speech bubbles. The gallery's short retro and webtoon prompts did exactly that:
If you want the picture clean, say so: "no text, no subtitles, no speech bubbles, no sound effects, no watermark". The long cel prompt above leaves no room for text, and its result has none.
Can you do this with someone else's photo?
Only with their permission. A stylised picture is still recognisably that person, and posting it is posting their likeness. For a group, a couple or a pet, attach one photo and say who is who: "the two people from the reference photo, the woman on the left and the man on the right".
Where can you check this?
Where can you find more photo-to-art prompts?
Prompt Gallery has all nine anime prompts with their results, next to comic, painting, 3D, toy and handmade styles, 142 style prompts in all. Open one and use Customize to add your details and pick one of five aspect ratios before you copy it.
