GPT Image 2 · hand-drawn key visual

GPT Image 2 Hand-Drawn Animation Key Visual Generator

A prompt list ends in somebody else's picture. This one ends in a frame of your own film — pencil, cel paint and paper grain.

Hand-drawn animated frame of Venti sitting on a fountain in the Mondstadt square with a lyre

Hand-drawn animated frame of Nobara Kugisaki in a Shibuya back alley lit green by a vending machine

Hand-drawn animated frame of Yoshikawa Yuki on a school stairwell at dusk, an orange sky in the tall window

Hand-drawn animated frame of Ryomen Sukuna on a rubble throne before a black torii shrine and a red moon

Hand-drawn animated frame of Zhongli on a stone terrace above Liyue Harbour at dusk, paper lanterns rising over dark water — **Genshin Impact · Liyue at dusk** — An old general on a stone terrace. The whole film in one frame, and the man still recognisably himself.

Hand-drawn animated frame of Raiden Shogun on a storm-lit balcony, a violet lightning bolt splitting the sky over Inazuma — **Genshin Impact · the storm balcony** — Violet on violet inside five colours. The lightning is one brush stroke, not a glow filter.

Hand-drawn animated frame of Nahida cupping a glowing seed under the giant luminous tree of Sumeru at dawn — **Genshin Impact · Sumeru canopy** — A small figure against a huge painted tree — the paper tooth is still visible at full size.

Hand-drawn animated frame of Satoru Gojo at the rainy Shibuya crossing at night, neon signs blurred behind him — **Jujutsu Kaisen · Shibuya rain** — Rain over neon, drawn rather than filtered. His hair and collar come back from the reference.

Hand-drawn animated frame of Yuji Itadori on a Shibuya rooftop at dusk looking back over his shoulder — **Jujutsu Kaisen · the rooftop** — Warm rim, cold shadow, one last orange band. A character beat with no props at all.

Hand-drawn animated frame of Hori Kyouko under a clear umbrella at the school gate in summer rain, wet street reflecting an orange sky — **Horimiya · the school gate** — Summer rain on a clear umbrella. Everyday light, painted by hand.

Hand-drawn animated frame of Miyamura Izumi leaning on a desk in an empty classroom at night, moonlight through the window — **Horimiya · empty classroom** — Moonlight in a quiet room. The grain is the point, not the resolution.

Hand-drawn animated frame of Homelander on a skyscraper roof at dusk, an American flag and a helicopter behind him — **The Boys · the roof** — A live-action series redrawn as hand-drawn animation — the jaw and the flag still read.

Hand-drawn animated frame of Starlight pushing open a stage door in a backstage alley at night — **The Boys · stage door** — Bulb light and vent steam. The costume keeps its gold at thumbnail size.

A frame is the first test of a world.

Neta Studio is an AI world builder — a cast, a style and a set of places that stay put while you make things. GPT Image 2 paints the frame; the world is what makes the next frame look like the same film.

One cast, every frame — the world keeps the faces, hair and costumes, so frame twenty still wears the coat from frame one.
A style you can paste — one written brief holds paper, palette and ink line across scenes — all nine frames above came out of a single paragraph.
Frames become the world — every still lands as a work: storyboard the sequence, or open the world and play it.

No skill in the catalogue draws a hand-drawn key visual yet, so nothing here hides behind a preset. The button sends a written brief: style, scene and the on-model constraint.

Hand-drawn key visual questions

Do I get a still or an animation?

A still. One 3:2 frame per ask, drawn as a single production cel. Motion is the next step: lock the look first, then animate the shot you actually want.

How do I keep my character on-model?

Attach a character sheet or name a character whose cover already lives in the world. The brief tells the model to keep face, hair and outfit, and the nine frames above were each drawn from one reference portrait.

Why does it look like paper and pencil?

Because the brief asks for it: a visible underdrawing, flat cel paint and a watercolour background with paper tooth. Delete those clauses if you want a cleaner digital finish — the rest of the prompt still works.

What size is the frame?

1536×1024, 3:2 — a deck slide, a thumbnail, or a poster after an upscale. The reference sheet for each character stays in the world either way.

Can I build a whole sequence in one style?

Yes. Reuse the style sentence and change only the scene. That is how the Genshin, Jujutsu Kaisen, Horimiya and The Boys frames above hold one look across four different worlds.

Can I redraw a live-action show this way?

That is what the two The Boys frames are: a live-action series redrawn as hand-drawn animation. Whether you may publish it is a rights question, not a drawing question.

Can it put the title on the frame?

Ask for the frame without type. Generated lettering still wobbles, so the drawing leaves headroom and you set the title in your editor.

What if the first frame misses?

Change one thing and regenerate — the weather, the pose, the landmark. Style and scene sit in separate sentences in the brief so a bad frame rarely means a bad look.

One shot in, one painted cel out.

GPT Image 2, one 3:2 frame, your cast drawn on paper.

Make My Frame