LTX-2.3の動画生成プロンプトはチャッピーさん(ChatGPT)に作ってもらおう(女性アイドル編)

前回の記事をまだ読んでない方は、こちらをどうぞ。

動画の主役は、前回と同じです。
ChatGPTで描いてもらった、AIアイドル「アイちゃん(仮)」です。

ある日のチャッピーさん(ChatGPT)との会話。

俺「LTX-2.3って、SFっぽい効果音とかエフェクトがイマイチなんよなぁ」

チャ「そないなことあらへんで。ワシがカッコええエフェクト出るように、プロンプト作ったるわ」

俺「えぇ〜？ホンマかいな？なんか信じられへんなぁ」

チャ「任しとき。ビームもバチバチ、未来感もギュンギュンやで」

A cute Japanese female idol is performing on a brightly lit concert stage, singing and dancing energetically in front of an excited audience. The camera captures her from the waist up, then slightly pushes in as she performs.

First, she smiles sweetly and forms a heart shape with both hands in front of her chest. She looks directly at the camera with a charming idol expression.

Then she says in a sweet, cute voice, “アナタのハートに、ズッ、キュン！”

Immediately after that, she quickly changes her pose and holds both hands forward in the shape of finger guns, aiming toward the camera. At the exact moment of the finger-gun pose, a flashy cute sci-fi effect appears: a bright pink heart-shaped energy beam shoots out from her fingertips with sparkling particles, glowing light trails, and a small magical burst.

At the same time, a sharp gunshot-like pop sound rings out, mixed with a cute magical shooting sound effect. The stage lights flash briefly in sync with the shot. The whole scene feels energetic, adorable, and dramatic, like an idol performance with cute sci-fi special effects.

俺「ほんまにカッコよくなるん？」

俺「ホンマヤン」

チャ「こんなんもできるでー！」

と、チャッピーさんからの提案で、追加で3パターン作ってくれました。

・かわいい系

A cute Japanese female idol is performing on a brightly lit concert stage in front of a cheerful audience. She sings and dances with a sweet smile, full of charming idol energy. The camera frames her from the waist up and gently moves closer as she performs.

First, she looks directly at the camera and forms a heart shape with both hands in front of her chest, smiling in an adorable and affectionate way.

Then she says in a sweet, cute voice, “アナタのハートに、ズッ、キュン！”

Immediately after that, she quickly changes her pose and points both hands forward in the shape of finger guns toward the camera. At the exact moment of the finger-gun pose, a bright pink heart-shaped energy beam shoots out from her fingertips. Sparkling glitter, heart-shaped particles, soft glowing light trails, and a cute magical flash appear around her hands.

At the same time, a sharp pop-like gunshot sound rings out, mixed with a cute magical shooting sound effect. The stage lights flash in pink and white in sync with the shot. The whole scene feels adorable, energetic, sparkly, and full of idol charm.

・かっこいい系

A stylish Japanese female idol is performing on a dramatic concert stage with powerful lighting and a confident presence. She sings and dances sharply with strong charisma, captivating the audience. The camera captures her from the waist up with a slightly dynamic forward push.

First, she raises both hands and forms a heart shape in front of her chest while staring confidently at the camera, giving a cool but charming expression.

Then she says in a sweet but impactful voice, “アナタのハートに、ズッ、キュン！”

Immediately after speaking, she snaps both hands forward into a finger-gun pose aimed directly at the camera. At that moment, a vivid pink energy blast fires from her fingertips, surrounded by glowing sparks, sharp light streaks, particle bursts, and a stylish sci-fi muzzle-flash-like effect. A visible beam of energy shoots straight toward the viewer.

At the same time, a crisp gunshot-like sound rings out with a layered sci-fi shooting effect. The stage lights flash dramatically, and the atmosphere becomes intense and powerful. The whole scene feels cool, flashy, dramatic, and stylish, like a futuristic live performance.

・ライブMV系

A Japanese female idol is performing in a live concert music video on a large stage with colorful spotlights, LED screens, and an excited audience waving glow sticks. She sings and dances with lively energy and expressive movements. The camera feels cinematic, starting from a medium shot and slightly pushing in for emphasis.

First, she smiles brightly and forms a heart shape with both hands in front of her chest while facing the camera. The crowd reacts with excitement.

Then she says in a sweet, cute voice, “アナタのハートに、ズッ、キュン！”

Right after the line, she quickly changes into a playful finger-gun pose, aiming both hands toward the camera. At the exact moment of the gesture, a bright pink heart-shaped beam shoots from her fingertips, with sparkling particles, glowing trails, glittering effects, and a burst of magical light. The effect is flashy, cute, and visually clear, like a highlight moment in a concert music video.

At the same time, a gunshot-like pop sound rings out, mixed with a cute electronic sci-fi sound effect. The stage lights strobe in sync, the LED background flashes, and the audience cheers. The whole scene feels like a memorable climax in an idol live music video, energetic, emotional, and visually exciting.

チャッピーさん、すごすぎ。

LTX-2.3の動画生成は、ComfyUIで、このワークフローを使用しました。
ノイズシードはすべて「489741023421064」です。

スタートフレーム用の入力画像は、この記事の冒頭にも添付した、この画像です。

チャッピーさんが、LTX-2.3の動画生成プロンプトを作る場合の「7ステップの型」を教えてくれました。

7ステップの型

1. 主人公
・誰が出るか
　例: A cute Japanese female idol

2. 場所
・どこで起きるか
　例: on a brightly lit concert stage

3. 基本動作
・何をしているか
　例: singing and dancing

4. 時系列
・まず→次に→その直後→同時に
　例: First / Then / Immediately after that / At the same time

5. 見た目の演出
・エフェクト、光、動き
　例: heart-shaped energy beam, sparkling particles

6. 音
・セリフ、効果音、歓声
　例: she says..., a gunshot-like pop sound rings out

7. 全体の雰囲気
・かわいい、かっこいい、MVっぽい
　例: adorable, dramatic, cinematic

どんな動画を作りたいか、動画のイメージを、この「7ステップの型」に当てはめます。
番号の順番通りでなくても、7つの要素が入っていれば良いです。

例は英語で書いてありますが、動画のイメージは日本語で書き、ChatGPTに
「ComfyUIのLTX-2.3で動画生成するプロンプトを英語で作って」
とまず書いて、7ステップの型をチャットに投げれば、英語のプロンプトを作ってくれます。

セリフは、この記事のプロンプトの実例のように、ひらがなとカタカナで書き直した方が、ローマ字で書くよりも流暢にしゃべります。

スタートフレーム用の画像は、あらかじめ用意しておいてください。

あとは、ChatGPTで作ってもらったプロンプトで動画を生成します。
その結果を見て、エフェクトをもっと大きく、音をかわいらしく、など、続きのチャットで指示を出し、
　動画生成 → 改善版プロンプトを作ってもらう → 動画生成 → ・・・
という感じで何度かやり取りをして、動画を完成させます。

前回の記事では、LTX-2.3って無料だし、こんなもんかと思ってましたが、まだまだポテンシャルを引き出せてなかったですね。

続編をお楽しみに！