Upgrade to Pro — share decks privately, control downloads, hide ads and more …

オモシロAIごった煮会 に参加させていただく

Avatar for ayane ayane
June 17, 2025

オモシロAIごった煮会 に参加させていただく

画像生成が楽しい!って話です(語彙力)

Avatar for ayane

ayane

June 17, 2025
Tweet

More Decks by ayane

Other Decks in Technology

Transcript

  1. 󰢨 自己紹介 ✿ 旅とテクノロジーが大好きです ✿ 『綾祢は今日も旅をする。』 という雑記ブログを書いてます ✿ 最近のマイブームはAIを 生活の中で気軽に実験すること

    ✿ ChatGPTとDifyがメイン Cursor、GenSpark、Canva、Gemini、NotebookLM、Devin、etc.. ツールを問わずにさわってます X: @blueb Website: https://ayatabi.net/ もとは自分の写真です。笑
  2. 「オモシロ AIごった煮勉強会」というイベントの人たちが関心を示そうなイラストを生成したいです。 プロンプトを考えてください。 ## Webサイト https://omoshiroai.connpass.com/ ### 概要 テキストからの画像生成 AI(StableDiffusion,

    Novel AI等)、ChatGPTなどの自然言語処理 AI、LiDARなどの位置計測技術など、近年目覚ましい進歩 を遂げています 技術の変化のスピードが速すぎて全てウォッチするのは不可能なので、勉強会という形で仲間を募りたいと思います 本勉強会ではこれらを特にジャンルを問わずにトピックを扱うことで、技術の応用の使い道や相互での組み合わせの可能性を探ります データ分析の専門の方のご参加はもちろんですが、それ以外にも色々な分野の方にもお気軽に参加していただけると嬉しいなと思っています (普段と違 う分野の方が集まった方がアイディアとしては良いものが出そうな予感 ) みなさんが「面白い・・!」と思うトピックでしたらなんでも OKです まずは遊んで使ってみた、くらいのノリでも OKですので、お気軽にご参加ください アイディア的な部分、情報収集文脈 (どのような応用がありうるか、 twitter事例や最新動向 )をメインにしつつ、 30%くらいは技術的な背景を深めるなど、 皆さんと楽しみつつこの分野の知見が自然と貯まる形を目指します ### スケジュール 毎月第3土曜日 10:00-11:00 ### オンライン開催 会の方針としまして (2023/2/21時点) 基本的にはネタを元にディスカッション・ブレストをしたいので、双方向でのコミュニケーションを重視したいなと考えています 1回の勉強会ではオンラインを中心に最大 20〜30名程度にて、各自可能な範囲での画面 ON、また簡単な自己紹介等をお願いすることになると思いま す もちろん公序良俗に反するトピックは扱えませんので、その辺りはどうぞご配慮ください ・AIの面白い可能性を発掘していきましょう・・!! 画像生成 https://chatgpt.com/?model=gpt-4-1 https://claude.ai/new
  3. イラスト生成用プロンプト案(日本語 →英語併記) 1. 「AIのごった煮パーティー」 日本語: 色とりどりのデジタルガジェットやロボット、パソコン、スマートフォン、センサーなどが大鍋で煮込まれている。 周りで楽しそうに議論・ブレストする多様な年代・職業の人々(男女混在)。 背景にはカラフルなサイバー空間や AIアイコン、数式やコード片が浮かぶ。 「遊び」「発見」「未来感」「ごった煮」の雰囲気を強調。

    明るくポップなタッチ、コミュニティ感のある和やかな表情。 英語: A vibrant digital stew: a big pot filled with colorful gadgets, robots, computers, smartphones, and sensors, all simmering together. Surrounding the pot, people of various ages and professions (male and female) are happily brainstorming and discussing. In the background, a colorful cyber space with floating AI icons, mathematical formulas, and snippets of code. The scene should emphasize a sense of playfulness, discovery, and a "melting pot of the future." Bright, pop-art style, friendly community atmosphere. 2. 「ごった煮 AI研究室」 日本語: 研究室のホワイトボードに多様な AI技術のスケッチ(画像生成、自然言語処理、 LiDAR地図など)がびっしり。 その前でカジュアルな服装の男女がアイデアを語り合い、時に驚いた顔、時に笑いあっている。 デスク上にはノート PC、タブレット、マグカップ、電子基板などがごちゃ混ぜで置かれている。 カラフルで温かみのあるイラスト。 英語: A bustling lab scene: a whiteboard covered in sketches of various AI technologies (image generation, NLP, LiDAR maps, etc.). In front of it, casually dressed men and women discuss ideas, sometimes surprised, sometimes laughing together. On the desk, a mix of laptops, tablets, coffee mugs, and circuit boards. Colorful and warm illustration style. 3. 「AI大集合のブレスト会」 日本語: テーブルの周りに、ロボットや人間、 AIアイコンキャラクター( StableDiffusion、ChatGPT、LiDARなどを象徴する)が一緒に座り、アイデアの光(電球アイコン)が浮かぶ。 みんなが楽しそうに話し、時折手書きの図やメモを見せ合う。 全体的に「お祭り」「ポップカルチャー」「未来感」「交流」の雰囲気を。 英語: Around a table, robots, humans, and AI icon characters (representing StableDiffusion, ChatGPT, LiDAR, etc.) sit together, with floating lightbulb ideas above their heads. Everyone is happily sharing ideas, showing each other sketches and notes. The overall vibe is festive, pop-culture, futuristic, and interactive. 一行プロンプト例( Stable Diffusion系向け) A lively, colorful illustration of a diverse group of people and robots brainstorming around a digital stew pot full of gadgets and AI icons, with a playful, friendly community vibe, pop-art style 備考 どのプロンプトも「バラエティ豊かな参加者・ AI技術」「遊び心」「未来的な雰囲気」を重視しています。 出力画像が「イベント告知」「 Webバナー」などに使えるよう、明るく親しみやすいタッチを推奨します。 もし特定の AIキャラクターやガジェットの描写イメージがあれば、追加で指示可能です。 ご希望のスタイルや要素(例:和風テイスト、サイバーパンク、手書き風、など)があればさらに絞り込んだバージョンもお作りできますので、ご要望があればお知らせください! GPT-4o
  4. まず、その Webサイトの詳細を確認してから、イベントに最適なイラスト生成プロンプトを考えてみますね。 https://omoshiroai.connpass.com/の取得に失敗しました Webサイトへのアクセスはできませんでしたが、提供していただいた情 報 から「オモシロ AIごった煮勉強会」の特徴を把握できました。以下のようなイラスト生成プロンプトを提案します: 基本コンセプト向けプロンプト 1. カジュアル・親しみやすさ重視版

    A friendly, colorful illustration showing diverse people collaborating around floating AI symbols (brain icons, neural networks, chat bubbles, camera icons, robotic elements). The scene has a warm, welcoming atmosphere with speech bubbles containing lightbulb ideas. Include elements like: ChatGPT chat interface, image generation previews, LiDAR scanning visualization, data charts. Art style: modern flat design with soft gradients, approachable and non-intimidating. Color palette: warm blues, friendly greens, and energetic oranges. 2. 勉強会・コミュニティ重視版 A cozy online meeting scene with multiple video call windows showing diverse participants, surrounded by floating AI elements: generated artwork, chatbot conversations, 3D point clouds from LiDAR, neural network diagrams. Include calendar showing "第3土曜日 " and time "10:00-11:00". Add elements of discovery and learning: magnifying glasses, lightbulbs, question marks transforming into exclamation marks. Art style: warm, inviting illustration with hand-drawn feel and soft colors. 3. 技術とクリエイティビティの融合版 A dynamic illustration depicting the convergence of various AI technologies. Show interconnected nodes representing different AI fields: text-to-image generation (paintbrush with digital pixels), natural language processing (conversation bubbles with code), computer vision (eye with data streams), and sensor technology (radar waves). Include diverse people of different backgrounds collaborating, with floating holographic displays showing AI outputs. Style: modern isometric design with vibrant colors and glowing effects. 4. 「ごった煮」コンセプト強調版 A playful "mixed pot" or "hotpot" metaphor illustration where various AI technologies are ingredients in a bubbling cauldron of innovation. Show different AI elements as ingredients: neural network noodles, data visualization vegetables, chatbot speech bubbles, generated image fragments, sensor readings as spices. Steam from the pot forms creative ideas and solutions. Diverse hands reaching in to add ingredients, representing community collaboration. Style: whimsical and colorful with a touch of humor. 使用時のポイント 日本語を含める場合は英語プロンプトの後に "with Japanese text elements" を追加 より具体的な色合いや雰囲気を指定したい場合は、カラーパレットや参考アートスタイルを明記 イベントロゴやテキストを後から追加する前提で、余白のあるレイアウトを意識 どのプロンプトが一番イメージに近いでしょうか?さらに詳細を調整することも可能です! Claude Sonnet4
  5. 結論: ChatGPTでもClaudeでも、 プロンプトが作れれば OK。 → 画像生成は ImageFXが “今のところ一番丸く解決 ”してくれます! 納得いかない場合は、

    Adobe Fireflyのような 他ツールも組み合わせましょう。 まずは遊んでみましょう! AIは使ってナンボ! まとめ