diff options
| author | Paul Buetow <paul@buetow.org> | 2026-04-20 19:07:56 +0300 |
|---|---|---|
| committer | Paul Buetow <paul@buetow.org> | 2026-04-20 19:07:56 +0300 |
| commit | d773e26b19dcea2f4a1917ebb6edaa4f5721693f (patch) | |
| tree | cc38b381c0d367a84ad802c16bf7fcb0c67bc8ad /README.md | |
| parent | 7bb500b8cd567e45634fc01c97e8dcd5a1a9db22 (diff) | |
Remove story feature for task 05
Diffstat (limited to 'README.md')
| -rw-r--r-- | README.md | 106 |
1 files changed, 0 insertions, 106 deletions
@@ -35,23 +35,6 @@ It has mainly been vibe coded using Claude Code CLI. - **Config-driven selection**: Set `image.provider` to `openai` or `nanobanana` - Scene generation creates memorable contexts for each word - Batch processing of multiple words -- **Vocabulary story generation** (`--story`): - - Generates a ~250-word Bulgarian story that naturally uses every word in a batch file - - All human characters are adults; story genre/setting driven by `--story-theme` - - Produces **12 pages** per comic: cover + 5 story pages (2×2 panel grid) + 5 gallery pages (close-up character art) + back cover - - All output saved under `comics/<title-slug>/` with files named `<slug>_*.png` - - **Rendering mode** chosen randomly 50/50 each run; force with `--ultra-realistic` or `--no-ultra-realistic`: - - *Ultra-realistic*: photography-language style prompt (DSLR, cinematic stills, hyper-realistic) + mandatory photorealistic rendering requirement — produces near-photographic panels - - *Standard comic*: comic art style pool (90% "ultra realistic comic strip", 10% manga / watercolor / noir / pop-art etc.) - - Art style override: `--story-style` replaces the random pick for both modes - - **Iterative character consistency**: each page is generated with the cover + previous page as pixel references so characters stay visually consistent - - **Character bible**: Gemini generates a detailed visual guide (age, clothing, colours) used in every prompt - - **Cinematic narration** via Gemini TTS (opt-in with `--narrate`; off by default to save quota) - - Output: `<slug>_narration.mp3` — Bulgarian phonology, dramatic pacing, random voice from a curated pool - - Override voice with `--narrator-voice`; intro teaser + main story chunks (~100 words each) + epilogue - - Falls back to `<slug>_tts_todo.txt` if narration fails - - **Vocabulary learning file** (`<slug>_comic_vocabulary.txt`) — word list with translations + full story text - - **Theme file** (`<slug>_theme.txt`) — records the `--story-theme` used for easy reproduction - Anki-compatible export - Random voice variants and speech speed @@ -178,95 +161,6 @@ Key features: totalrecall --archive # Archives cards to ~/.local/state/totalrecall/archive/cards-TIMESTAMP ``` -6. Generate a vocabulary story + comic book from a batch file: - ```bash - totalrecall --story words.txt - ``` - - Outputs to `comics/<title-slug>/`: - - `<slug>_story.txt` — ~250-word Bulgarian story using every word naturally - - `<slug>_cover.png` — traditional comic book front cover with Bulgarian title - - `<slug>_page_1.png` … `<slug>_page_5.png` — five 2×2-panel story pages (16:9) - - `<slug>_gallery_1.png` … `<slug>_gallery_5.png` — five close-up character gallery pages - - `<slug>_back.png` — back cover with blurb - - `<slug>.pdf` — all pages assembled into a single PDF - - `<slug>_narration.mp3` — cinematic Gemini TTS narration (only when `--narrate` is passed) - - `<slug>_tts_todo.txt` — written if `--narrate` was given but narration failed - - `<slug>_comic_vocabulary.txt` — vocabulary words + full story text for learning - - `<slug>_theme.txt` — records `--story-theme` for easy reproduction - - Customise the story: - ```bash - # Set a specific theme/setting - totalrecall --story words.txt --story-theme "a Wonder Woman inspired heroine in a futuristic city" - - # Override art style - totalrecall --story words.txt --story-style "Japanese manga with clean linework and speed lines" - totalrecall --story words.txt --story-style "retro 1960s pop art in the style of Roy Lichtenstein" - - # Force ultra-realistic photorealistic mode - totalrecall --story words.txt --ultra-realistic - - # Force standard comic style (default is random 50/50 between ultra-realistic and standard) - totalrecall --story words.txt --no-ultra-realistic - - # Enable cinematic narration (default: off — use --narrate to opt in) - totalrecall --story words.txt --narrate - - # Choose narrator voice (default: random from pool; only relevant with --narrate) - totalrecall --story words.txt --narrate --narrator-voice Charon # deep, authoritative - totalrecall --story words.txt --narrate --narrator-voice Fenrir # strong, resonant - totalrecall --story words.txt --narrate --narrator-voice Enceladus # breathy, intimate - totalrecall --story words.txt --narrate --narrator-voice Algieba # smooth, warm - totalrecall --story words.txt --narrate --narrator-voice Aoede # breezy, expressive - totalrecall --story words.txt --narrate --narrator-voice Schedar # steady, grounded - - # Repair a partial run (skip existing pages, regenerate missing ones) - totalrecall --story words.txt --story-slug my-comic-slug - ``` - -#### Optional Gallery Videos (Veo) - -After `--story` comic generation completes, the CLI prompts you to generate short MP4 videos from the gallery close-up images using **Google Veo**. - -**How it works:** - -1. Once the comic pages, PDF, and narration are written to disk the CLI automatically searches `comics/<slug>/` for gallery PNG files. -2. It lists the found images and asks: - ``` - Found gallery pages: - comics/my-story/my-story_gallery_1.png - comics/my-story/my-story_gallery_2.png - ... - Generate videos for these gallery pages? [y/N]: - ``` -3. If you confirm, you are asked which pages to animate: - ``` - Which pages? (e.g. 1,3,5 or all) [all]: - ``` -4. Each selected gallery PNG is sent to the Veo API and the resulting 8-second MP4 is saved alongside the PNG: - ``` - comics/<slug>/<slug>_gallery_1.mp4 - comics/<slug>/<slug>_gallery_2.mp4 - ... - ``` - -**Model:** `veo-2.0-generate-001` - -**Cost note:** Veo is a paid Google Cloud feature. Your `GOOGLE_API_KEY` must have billing enabled and the Veo API activated in your Google Cloud project. The prompt will not appear if you pass `--video=false`. - -**CLI usage:** - -```bash -# Default: prompt appears after story generation (answer y/N interactively) -totalrecall --story words.txt - -# Skip the video prompt entirely -totalrecall --story words.txt --video=false -``` - -Video generation failures are non-fatal — any errors are printed as warnings and the already-generated comic, PDF, and narration remain intact on disk. - #### Batch file format Create a text file with Bulgarian words, optionally with English translations or Bulgarian definitions. The tool supports five flexible formats: |
