@Ragazul — Open Skill 创作者

@Ragazul

  • dolphin-surfer

    Use when the user wants to create a video of a person surfing on the back of a fast-swimming dolphin (from an uploaded photo of the person, or a described person), where the dolphin acts like a surfboard gliding above the waterline on a sunny calm sea — including when they want to change the person, the format, the background audio, or redo the video.

  • dachshund-skater

    Use when the user uploads a photo of a person and wants a video of that person riding on a fast-running dachshund that acts like a skateboard — standing on the dog's back while it sprints underneath and behaves like a board through tricks (flip & grab tricks like kickflip, heelflip, 360 flip, indy grab; grind & manual tricks like nose-slide, 50-50 grind, boardslide, manual/wheelie, nose manual; ollie). The dachshund is the main focus of the video; the person from the uploaded photo only needs to be briefly visible in between, otherwise only their legs are visible. By default the video is rendered in 9:16 portrait unless the user requests another format. Also use when they want to change the setting (beach promenade, halfpipe, street, skatepark, subway station, desert road, snowy city), add more tricks, or make a longer version.

  • dialogue-lip-sync

    Use when generating any video with speaking characters (dialogue scenes, film/short clips, MVs, interviews, talking-head shots) where each character must speak the correct line at the right moment with accurate, natural lip sync — apply dialogue-accurate, lip-synced performance direction to prompts and verify/regenerate wrong, delayed, or out-of-sync speech.

  • realistic-anime-clothing-hair

    Use when generating any realistic/cinematic image or video of anime-inspired characters where their signature hairstyle and outfit must be preserved but rendered believably — realistic hair (strand detail, volume, natural movement) and real wearable garments (fabric folds, seams, weight, material response) with consistent identity — verify/regenerate stiff, flat, cel-shaded, or costume-y output.

  • anime-facial-expressions

    Use when generating any anime or animated-style image or video with characters (anime film shots, MVs, cutscenes, portraits) where each character's face must clearly and believably express the emotion of the scene — apply anime-style expression direction (expressive eyes, sharp brows, clear mouth shapes, blush, sweat drops, etc.) and verify/regenerate stiff, blank, or mismatched faces.

  • realistic-cinematic-look

    Use when generating any realistic or cinematic image or video (film shots, short videos, MVs, trailers, scenes, portraits) where the result should look believable, physically grounded, and filmed with a real camera — apply realistic-cinematic style direction to every prompt and verify/regenerate artificial, plastic, or visually incoherent output.

  • mirror-reflection-consistency

    Use when generating any image or video (film shots, short videos, MVs, trailers, scenes, portraits) where a mirror or other reflective surface (glass, shiny floors, polished surfaces, chrome, water) is present or visible — apply physically correct, spatially consistent mirror-reflection direction to every prompt and verify/regenerate missing, lagging, wrong-pose, broken, or mismatched reflections.

  • weapon-combat-director

    Use when the user wants melee-weapon, traditional martial-arts weapon, or uploaded custom weapon combat in generated images or videos where each fighter's exact weapon, fighting system, grip/handedness, range logic, weapon interaction and effects must visually drive HOW they fight — also when only advancing or revising one step of such a combat sequence (new setup, new clip, restyle, verification)

  • realistic-superhuman-combat

    Use when the user wants to create a realistic, cinematic superhuman battle — still keyframes and/or a full multi-shot action sequence (photoreal combat choreography, superpowers, shockwaves, impacts) built from uploaded character/environment/prop assets as mandatory source material, with no dialogue.

  • live-action-anime-movie-look

    Use when generating any image or video in the style of a high-end live-action anime movie adaptation — real people inspired by anime designs, realistic skin/hair/clothing, grounded stylized elements and powers, blockbuster cinematography, striking realistic lighting, integrated cinematic VFX — verify/regenerate flat anime rendering, plastic CGI, or inconsistent character design.

  • dance-only-no-lipsync

    Use when the user wants music video scenes or clips where visible people/characters only dance, move, celebrate, and react to the music — never sing, rap, speak, or lip-sync — including when an uploaded song has vocals/lyrics and the visuals must stay dance-only with no mouth movement synced to the vocals.

  • safe-model-routing

    Use when the user uploads or references an image, or asks to generate, edit, animate, or process an image, and a visual content check must run before any generation or editing model is selected — classify nudity/adult content categories, enforce age safety, route to an officially permitted model channel, preserve uploaded-image fidelity, and avoid wasteful retry loops.

  • music-video-source-sync

    Intro: Music Video Source Sync and Performance Control Create the music video using only the user’s uploaded source files as the main foundation. Use the uploaded audio as the official soundtrack and match all visuals, pacing, camera movement, lip sync, character performance, mood, rhythm, and genre to it. Do not replace the audio, ignore uploaded files, invent unrelated music, or switch to a different genre.

  • use-all-uploaded-source-files

    Use when creating a video from uploaded source files (images, videos, audio, reference files, character/style references, visual assets) where every uploaded file must be analyzed and used as mandatory creative input — uploaded audio is the official soundtrack defining genre, pacing, and mood; uploaded images are identity/style references; never silently ignore, replace, reinterpret, or override uploaded material — verify each file's purpose in the final result.

  • character-expression-director

    Use when generating any character-driven image or video (film shots, short videos, MVs, portraits, dialogue or emotional scenes) and the character's facial expression must match the emotion and situation — apply cinematic, believable expression direction to every prompt that shows a face, and verify/regenerate stiff or mismatched faces.

  • combat-style-director_3

    Use when the user wants combat/fight scenes in generated images or videos where each fighter's chosen martial art must visually drive HOW they fight

  • non-verbal-character-sounds

    Use when generating any video with characters where spoken dialogue is not desired — characters should make only non-verbal vocal sounds (breathing, gasps, grunts, groans, cries, screams, etc.) matched to emotion and action — direct prompts to avoid spoken dialogue and verify there is no readable speech, with mouth movement matching the sound.