DeCat · @DeCat2025
Black Cat Independent Mirror Horror
Three timed beats separate a real cat from an independently moving mirror cat before a glass-breaking horror payoff.
出典を確認したプロンプトと元の投稿
クリエイターが公開したプロンプトを、完成した MiniMax H3 動画とあわせて確認できます。シーンの演出、動き、カメラワーク、連続性、音声をどう組み合わせているか学べます。
すべての事例に元の出典へのリンクがあります。公開プロンプトをそのままコピーすることも、構成を自分のシーンに合わせてアレンジすることもできます。
確認済みの事例 706 件
カメラ技法、絵コンテ、アニメーション、参照素材の制御、音楽、会話、モーショングラフィックスから事例を探せます。プロンプト公開済みの事例と、ワークフローのみの事例を区別して表示しています。
706 件の事例が見つかりました · 211〜240 件目を表示
DeCat · @DeCat2025
Three timed beats separate a real cat from an independently moving mirror cat before a glass-breaking horror payoff.
Kevin Le Guillou · @kevinleguillou
An H3 Max anime story test uses a new character sheet from Anima and GPT Image, with Anima backgrounds.
Rising Sun Interactive · @RisingSunInt_
A 3D remake of the 1955 cartoon Cellbound combines GPT-6 with H3 video generation on a local PC.
すーさん · @su_nagomi
A nine-panel reference drives a timed cafe recruitment ad with five fixed characters, Japanese text, dialogue and music.
WebStudio YAS · @Studio_YAS_
Episode three of a five-part sci-fi series reports locally generated video and audio on an RTX 3060.
おのふみ| 3D×動画生成×個人開発 · @onofumi_AI
Six timed shots turn homework frustration into triangle eyes, steam, stress marks and a return to a normal anime face.
ささゆう|MiniMax H3・AI撮影スタジオKonte開発 · @forticodesign
A creator reports stronger reproduction when specifying the 3D underlay data in an H3 generation.
keys 🧪 · @u1tra_instinct
A Mac Silicon studio combines Qwen Image 2.1 with an H3 MLX implementation and draft or high-quality options.
ささゆう|MiniMax H3・AI撮影スタジオKonte開発 · @forticodesign
A twenty-commercial exercise reuses cast and props through a structured production flow directed with Claude.
鈴木憂一 | Highdrama · @yu_ichi_suzuki
A 5090 and three Spark-compatible systems pipeline seven-second voiced clips, using Irodori audio as the H3 guide.
ぽかりんAI/PokarinAI · @pokarin_grok
A creator reuses a public seven-scene Seedance action prompt for a local H3 schoolgirl-versus-helicopters test.
T · @tkvtk
An idol MV uses a finished Lyria song, three-to-four-second audio guides, fixed character images and ffmpeg assembly.
AiDeNier(アイデニール) · @de_nier_ai
An eight-second H3 output demonstrates an 8GB VRAM run while the author explicitly notes melting artifacts.
🐣NYAP🐣 · @nyappokopoko
A custom local Qwen27B system develops a cat reconciliation brief into an H3 film, with dialogue quality concerns.
新手老兵🇭🇰的AI世界 · @Langby2
A Seedance, WAN and H3 comparison uses separately adapted prompts and reports H3 foot and travel-direction errors.
ヒロ(HG)|実務で使えるAI活用 · @hsrk_g_hsrk
A fan short combines canonical character sheets and MCP character context with H3 video, Irodori TTS and Remotion.
TOONIA AI Short Films · @tooniaanimation
A rain-soaked slow-motion shot dissolves an umbrella-bearing figure into rising blue glass shards and digital pixels.
DeCat · @DeCat2025
Shonen-style power buildup and eye close-ups lead to one tiny glass tap, then an instant return to innocent calm.
Tom𝕎 · @Tomw852
A local RTX 5070 short combines DaSiWa Hybrid V3 with Turbo, Realism People, Combat and Motion Continuity LoRAs.
LinacoWSL · @breath_in_code
Two camera-instruction tests explore an orbit with and without a reference image; the MV workflow also produces singing.
レオ(七深麗音)|AI×SNSコンテンツ戦略 · @singa_44
An H3 Max Turbo air-combat fan video pairs action with teammate conversation and explicitly labels itself AI imagery.
LAPTOP LIFESTYLE BLOGGER · @MonetizationDon
A basketball player’s exhaustion, drink interaction and renewed energy form a creator-style product story.
遠野の世界 · @tonoseka
A low-quality local R2V test uses reference markers and After Effects tracking to attach a replacement face.
その名前はすでに使用されています · @tekitou115330
A widely shared local H3 anime clip shows a yellow-ribbon schoolgirl against a rooftop setting.
あるてぃすと@AIイラスト · @arute0081
A shark test illustrates unwanted movement onto the sand and the difficulty of controlling a video trajectory.
Nobu-Kobayashi : Generative AI Technology · @nyaa_toraneko
First and last line-art frames guide inbetweens while text controls action timing; some melting still needs hand repair.
Nobu-Kobayashi : Generative AI Technology · @nyaa_toraneko
An FF/LF turn test fails to reproduce an unusual eye-only anticipation beat, highlighting a fine acting limitation.
OscarAI · @Artedeingenio
A fast anime-opening montage demonstrates multiple shots from one image, with the detailed tutorial reserved for subscribers.
Nobu-Kobayashi : Generative AI Technology · @nyaa_toraneko
A local H3 creator reports a connected 43-second movie, with an attached clip of about 43.4 seconds.
とすくん · @tokyo_Valentine
A serious heroine fails to trigger an automatic door until an ordinary customer arrives, in three timed silent shots.
プロンプト集の使い方
各事例を再利用できる構成として捉えましょう。動きや連続性を制御する部分を残し、元のシーン固有の創作要素を置き換えます。
カテゴリを閲覧するか、被写体、カメラ技法、クリエイター、モデルで検索します。作りたい動画に動きや構成が近い事例を選びましょう。
テキストが公開されている場合は、完全なプロンプトか一部の抜粋かを確認してコピーします。未公開の事例は、出典とアレンジのヒントを参考にしてください。
被写体や舞台を自分のものに変更し、必要な参照素材を用意して、生成前にすべての制約を確認しましょう。
確認済みの事例をコピーし、創作要素をアレンジして、MiniMax H3 で自分の動画を生成しましょう。