Prompt Studio 1.21 — Your creative workspace, connected.
A great result should be the beginning of your next idea — not a forgotten prompt buried somewhere on your computer.
Prompt Studio is a local-first Windows companion that brings prompt writing, a searchable media archive and creative planning tools into one workspace. It works alongside SwarmUI, ComfyUI and other generation tools. It is not restricted to MiniMax H3, FLUX, Krea or any single model, and it does not generate images or videos itself.
Explore what is new:
• Light Sphere — position a schematic lighting envelope around your scene and adjust its temperature, tint, strength and softness. Use the mouse wheel to orbit the 3D view, Shift + wheel to tilt and Ctrl + wheel to zoom.
• People and Pets — build scenes with multiple human figures, dogs and cats. Choose pet sizes and starting poses, adjust joints, and save the scene for later.
• Large Creative Workspaces — open the visual tools in separate tabs with larger previews and more room to work.
• Image and Video Workshops — work with reusable looks, text, logos and watermarks, then preview and export your results.
• Sound Workshop — explore 17 included WAV samples, import your own audio, organize themes and variations, and align sound with picture using waveforms and frame-based controls.
• Your Creative Archive — search readable prompts and recorded settings, browse daily thumbnail galleries, compare results and keep your best work close.
• Optional Local AI — choose installed language or vision models for prompt assistance. Responses and content restrictions depend on the selected model.
Getting started:
Download the complete Windows ZIP, extract everything into a writable folder and run START.cmd. Node.js 22 or newer is required; setup can offer a verified portable download.
Ollama and suitable models are separate, optional installations. Image understanding requires a vision-capable model. Video processing requires a configured FFmpeg installation.
Important: finish GPU image/video generation before requesting local image analysis. Running both together can exhaust available video memory and make the computer unresponsive.
The interface supports eight languages. Your personal media, archive index and generation models are not included.
This is a Public Beta. The lighting and pose tools provide schematic planning references; they do not guarantee exact generator compliance. Human OpenPose export remains human-only.
Free use and modification, including use while creating commercial work. Resale of the application or modified copies requires separate permission. See the included license and third-party notices.
If Prompt Studio helps you, a reaction, a short comment or a useful bug report helps shape what comes next. What would make your creative process easier?
Description
# Changelog
## 1.20.0 — 2026-10-28
Consolidated release of the working October 22–28 updates, compared with the published 1.19 edition. Generation and optional local AI models remain separate downloads.
### Archive and navigation
- File-modification-date presets and date ranges.
- Daily thumbnail gallery; first/last-page shortcuts and pagination at both ends.
- Startup and periodic automatic refresh of configured folders, with an opt-out.
- Floating back-to-top control and separate positive/negative copy actions in the media viewer.
- Editable Creator Passport from recorded metadata, explicit missing basics and clearer reproducibility limits.
### Prompt writing
- Prompt workbench with an original/working copy, conservative issue hints and optional AI handoff.
- Configurable SwarmUI, Stability Matrix, ComfyUI and custom tool launchers.
- Broader surprise prompt ideas, anatomy wording and model-aware negative suggestions.
- Optional 18+ artistic nude/boudoir theme, excluded from general random picks; non-explicit adult art only.
- Resolution presets, custom dimensions and framing choices for surprise prompts; remembered choices and clearer guidance to set dimensions in the generator.
- Classic saved prompt entries now preserve and search their negative text too, including JSON backups; old entries remain readable.
### AI assistance
- Local Ollama plus optional copy/open choices for external assistants.
- Image attachment and drag-and-drop for art prompts, restrained reinterpretations and image-to-video motion suggestions.
- Optional duration and target-workflow context for video ideas; no guarantee of model-setting compatibility.
- Separate remembered text/vision choices; model capability lookup rather than assuming vision from a name.
- Vision model selectors in the dedicated image-to-video and image analysis helpers; no longer limited there to Gemma.
- Prompt-only copying accepts common marker variants, JSON, headings and code blocks, while preserving the full response. Manual text selection takes priority.
- Optional experimental vision-profile instructions and an Ollama unload helper. Model weights are not included.
### Cleanup and appearance
- Explicit file-health/byte-identical-duplicate scan, recent-file exclusions, list/gallery review, checkboxes and select-all.
- Duplicate keeper preferences, batch review and protection of at least one exact copy; Recycle Bin or permanent deletion after review.
- Cleanup suggestions remain available after individual deletions.
- Readability adjustments, black/dark-blue appearance and eight-language UI updates.
### Distribution and limits
- Fresh English-first Windows package and separate GitHub source package with changelogs, upgrade instructions and local-AI memory guidance.
- Custom Free Use and No Resale License 1.0: commercial creative work allowed; software resale requires separate permission.
- No personal archive, media, downloaded Node runtime, AI weights or local launcher settings are bundled.
- Prompt Studio remains a companion, not a generator. Metadata completeness, accurate anatomy, exact reproduction, universal NSFW support and automatic clipboard insertion into third-party apps are not guaranteed.
- Some older translation/suggestion tools still use gemma3:4b. The optional vision profile was tried with an ordinary café image; explicit-image behavior was not validated.
## 1.19 — previous published baseline
Local prompt writing, searchable media archive, metadata recipes, comparison, ratings/categories, glossary, saved prompts and director timeline. See the repository history for that edition.