# nano-banana-pro

> # Manga-Autonoma (powered by nano-banana-2)　Autonomous AI 4-panel manga production system using Gemini API. / Gemini APIを活用した、完全自律型AI4コママンガ制作システム。

- **URL**: https://www.freshcrate.ai/projects/nano-banana-pro
- **Author**: FURUYAN1234
- **Category**: AI Agents
- **Latest version**: `v4.6.4` (2026-06-03)
- **License**: Unknown
- **Source**: https://github.com/FURUYAN1234/nano-banana-pro
- **Homepage**: https://furuyan1234.github.io/nano-banana-pro/
- **Language**: Python
- **GitHub**: 3 stars, 1 forks
- **Registry**: github
- **Tags**: `ai-manga`, `autonomous-agent`, `autonomous-ai`, `gemini-api`, `generative-ai`, `imagen3`, `llm`, `manga-generation`, `python`

## Description

# Manga-Autonoma (powered by nano-banana-2)　Autonomous AI 4-panel manga production system using Gemini API. / Gemini APIを活用した、完全自律型AI4コママンガ制作システム。

## Recent releases

| Version | Date | Urgency | Changes |
| --- | --- | --- | --- |
| `v4.6.4` | 2026-06-03 | High | ## What's New / 更新内容 - Fixed a dialogue extraction bug where character names containing specific particles were incorrectly filtered out. / セリフ抽出時の助詞（「と」「の」等）を含む特定文字長キャラクターのセリフ欠落バグを修正。 - Improved exact cast matching logic. / 完全一致判定で、名前の後に続く発言内容が正しく描画されるように改善。 |
| `v4.4.3` | 2026-05-28 | High | ## What's New / 更新内容  ### v4.4.3 (2026-05-28) - **[Fix & Model Sanitization]** OpenAI画像生成APIクライアント（`openai.js`）内に残存していた「gpt-image-2」という架空モデルの定義を完全に排除し、実在モデルである「dall-e-3」に正常化。また不要なデバッグ用モデルリストスクリプト等を削除。 - **[Fix & Model Sanitization]** Completely eradicated fictitious model ID "gpt-image-2" inside openai.js and standardized on "dall-e-3". Cleaned up obsolete diagnostic/test scripts. |
| `v4.0.3` | 2026-05-21 | High | ## What's New / 更新内容  ### v4.0.3 (2026-05-21)  - ** [Fix & UX] ** 吹き出しへの五感ラベル混入バグおよびアクターZONE・EMOTIONスタイル誤切替の修正 / Fix sensory label hallucination in speech bubbles and incorrect actor ZONE/EMOTION style transitions |
| `v3.54-alpha` | 2026-05-15 | High | ## What's New / 更新内容 - Completed 360° background integration pipeline. Injected 4 cropped perspective images as multimodal references in Gemini mode for highly accurate per-panel scenery generation. / 360°背景統合パイプライン（Phase 3）の完了と安定化。Geminiモードで4枚のクロップ画像をマルチモーダル参照として注入し、コマ別の方角に合わせた精密な背景描画を実現。 - Fixed UI layout issue where text and image overlays overlapped in the camera work preview panel. / カメラワーク確認UIにおいて、テキスト（「各コマで使用される背景の方角」）と画像オーバーレイが重なって表示されるレイアウト崩れを修正。 |
| `v3.12-alpha` | 2026-05-09 | High | ### v3.12-alpha (2026-05-09) - **[UI/UX]** UI/ログの完全日本語化を実施。思考トレース（Thinking Trace）が利用できない標準モデル使用時のメッセージを「通常処理が完了しました。」と簡潔化し、不要な注釈を排除しました。 / Localized UI logs to Japanese completely. Simplified the fallback message for standard models without Thinking Trace to "通常処理が完了しました。" (Normal processing completed), removing confusing annotations. |
| `v3.02.0-alpha` | 2026-05-03 | High | ## What's New / 変更点  ### BugFix - **[BugFix]** Fixed a bug in ChatGPT mode (GPT-image-2) where a black border was drawn around the entire page. The phrase "OUTER FRAME" in the CANVAS LOCK section was interpreted by DALL-E as a physical border drawing instruction; replaced with "image boundaries" and added explicit "NO PAGE BORDERS" rule to the ANTI-NOISE section as a double-block countermeasure. - **[BugFix]** ChatGPTモード（GPT-image-2）で生成されるプロンプトにおいて、ページ全体に黒い外枠（border）が描画される問題を修正。CANVAS LOCKセクションの |
| `v2.77.0-alpha` | 2026-04-29 | High | # v2.77.0-alpha — Enhanced GPT-image-2 Anti-Noise Protocol / GPT-image-2ノイズ対策プロトコル大幅強化  ## What's New / 変更点  ### ✨ Feature / 新機能  - **Enhanced GPT-image-2 Anti-Noise Protocol**: Integrated comprehensive noise countermeasures based on community research to significantly improve image quality when using ChatGPT Images 2.0 mode.   - **Positive Safe Rendering Block**: Injected safe rendering keywords (`smooth cel shading`, `clean color surfaces`, `low texture density`, etc.) to actively guide the mo |
| `v2.66.0-alpha` | 2026-04-22 | High | ### 📝 Update Documentation for ChatGPT & Enhancements  #### English * **[Docs]** Updated `README.md` to formally document the new **ChatGPT Images 2.0 Mode** and **Scenario Enhancement** features. * Adjusted terminology to reflect compatibility with DALL-E 3, enforcing A4 aspect ratios and vertical text rules via prompt injection.  #### 日本語 * **[Docs]** `README.md` を更新し、新たに搭載された **ChatGPT画像生成サポート** および **シナリオ演出強化** 機能に関する公式ドキュメントを追加しました。 * DALL-E 3生成に最適化されたA4縦長指定や、縦書きの視線誘導フォーマットに対応した設計であることを技術詳 |
| `v2.60.0-alpha` | 2026-04-21 | High | # Nano Banana Pro v2.60.0-alpha  🚀 **Dynamic Camera Recovery & Scenario Normalization Update**  ## 🔧 Fixes & Optimizations *   **Camera Lens Mapping [Critical Fix]**: Fixed an issue where Extreme Camera Angles (v2.55) were being stripped of their specific lens distortion weights (fish-eye, telephoto, dutch angle, etc.) during generation. AI-selected camera names are now explicitly mapped to heavy weighted lens tags, restoring the hyper-dynamic camera distortion. *   **Scenario Enhancements Nor |
| `v2.59.0-alpha` | 2026-04-21 | High | ## Nano Banana Pro v2.59.0-alpha  ### EN - **Optimization**: Completely overhauled the Gemini API model fallback lists in preparation for the deprecation of the official Imagen series models (June 24, 2026). Promoted `gemini-3.1-flash-lite-preview` and `gemini-2.5-flash-image` as primary free-tier fallbacks, and correctly updated modal payloads for native Gemini image generation.  - **System**: Implemented the multi-agent `Model Audit Protocol` across the standard `AGENTS.md` file, standardizing |

## Citation

- HTML: https://www.freshcrate.ai/projects/nano-banana-pro
- Markdown: https://www.freshcrate.ai/projects/nano-banana-pro.md
- Dependencies JSON: https://www.freshcrate.ai/api/projects/nano-banana-pro/deps

_Generated by freshcrate.ai. Indexes github releases for AI-agent ecosystem packages._