Videos Images
Sign in
Witch in Japanese Washitsu Portal

Witch in Japanese Washitsu Portal

views342
likes1
comments1
shares0
bookmarks0
quotes0
Translated
Subject: {argument name="主題" default="和室と漫画の境界に立つ魔女の少女"} / manga fantasy portrait

Person & Expression: {argument name="人物" default="20代女性、長いダークブラウンの髪、大きな赤いリボン"}, translucent skin, natural makeup with a natural flush, slightly open mouth in a faint surprised expression, gentle gaze directed at the camera.

Costume & Pose: {argument name="服装" default="深いネイビーのワンピース、黒のショートブーツ、背中に小さなブラウンのバッグ"}, with a black cat peeking out from the bag, holding a broom with a red jewel decoration in one hand, and stepping forward with the other foot, the front foot half-submerged into a giant torn hole in a manga page.

Background & Light: An elegant Japanese room with tatami mats, soft natural light streaming through the shoji screens, wooden pillars and floor, a hanging scroll and a single-flower vase, a green garden visible outside, a quiet and warm Japanese space.

Composition & Camera: Vertical 9:16, full-body portrait, slightly low angle, placing a large opened manga book prominently in the foreground, centering the person, with a composition that shows the boundary between manga and reality, 35mm lens, shallow depth of field, dynamic cinematic framing.

Texture & Style: High detail, 4K quality, realistic AI portrait, natural skin texture, delicate hair flow, cinematic lighting, a fantasy expression fusing live-action and manga. The manga page content consists of black-and-white manga panels featuring a woman with the same red ribbon and a black cat flying over Tokyo on a broomstick, including aerial-view scenes with Tokyo Tower in the background, close-up smiles, and flying scenes. Emphasize the torn holes and paper scraps of the page, clearly expressing the sensation of stepping into the manga world from reality.

Negative: Low quality, unnatural hands, distorted face, extra fingers, distorted body, unnatural footing, disjointed manga panels, excessive processing, character corruption, logos, watermarks, excessive exposure.

Machine-translated — the original is what produced this result.

✨ Generate Original post Open in Hitboard

How to use this prompt

Copy the prompt and paste it into GPT Image 2. Keep the composition, lighting and style terms as written — they are what produced this result. Swap the subject or scene details to make it yours, keep the aspect ratio, and attach the reference images the prompt mentions if any. Regenerate two or three times; the strongest version is usually not the first.