To generate an image, pass only the following instructions to GPT image2. 3:4 vertical composition Create a scene resembling a cinematic anime scene. Do not copy the composition, pose, or screen layout of the reference image; reference only the character design, clothing, face, color, and worldview. Depict an extremely dynamic and memorable moment that aligns with the reference character's personality, emotions, profession, fighting style, and worldview. Create an image that looks like a captured moment of action, emotion, or movement, rather than a simple staged shot. Utilize strong perspective, wide-angle lens composition, depth of field, foreground occlusion, the effect of wind blowing through hair and clothing, focus shifts, bokeh, atmospheric perspective, and cinematic lighting. Emphasize horizontal visual guidance and the creation of expansive space. Make it look like a "single frame from a video work," including a background. The background information density should be high and well-organized, while suppressing visual noise. Fragmentation is prohibited. Other characters, besides the main character, should be kept to a minimum. Static poses, bilateral symmetry, sky backgrounds, poster-style compositions, and social game-style character art are prohibited. Japanese TV anime style Anime screenshots, 2020s anime series, clear cel shadows, flat anime color scheme, exquisite anime lines, non-painted rendering, clean digital anime finished product, highly recognizable silhouettes, cinematic anime depth of field.
Подсвеченные слова — поля шаблона. Кликните по любому и впишите своё: бренд, цвет, персонажа. Остальной текст отвечает за свет и композицию. «В студию» откроет генерацию с этим промптом, «Скопировать» положит его в буфер — уже с вашими правками.