Repository navigation
Releases
Phoenix 1.10.1 - A new home
Phoenix is a creative's control room for local AI — a free, open‑source harness that drives your local generation toolchain from one browser workbench. Up to 1.10.0 it had grown tab by tab into one long page; 1.10.1 gives everything a home.
✨ Highlights
A new shell
A left rail with Create, Characters, Audio, Story and Library as their own workspaces, a Ctrl K command palette to jump anywhere, a Copilot panel beside your work, and a status bar with live providers — ComfyUI, VRAM, Blender, Unreal, jobs, orchestrator, vision and remote at a glance.
Settings, rebuilt
Six searchable sections that save the moment you change something. No more hunting for the one switch.
Library, regrouped — with a Trash
Assets, brushes, characters, images and sound takes sit in clearer groups, and brushes get folders (existing ones are migrated once). Deleting goes to a Trash with Restore — and replacing a brush keeps a snapshot, a failed brush save rolls back.
Storyboard → one MP4
Export board as video: the whole storyboard as one animatic in three aspect ratios, with Stop. Playback and export share one timeline; shots get camera tags, and a removed shot can be undone.
Levels
Library ▸ Scenes ▸ Levels — keep level layouts with Blockout / Movable / Detail layers.
And also
- A self‑added video workflow's frame rate now reaches the clip; the local / fal choice is remembered; reference keyframes can be stopped and time out.
- Skeleton assign refuses an empty selection, no more duplicated figures, brush runtime and fresh‑install scaffold kept in sync.
- Busy messages in plain words instead of codes.
- New troubleshooting entries: a missing Mixamo rig, LM Studio on Linux.
- The page is assembled from
web/src/at start instead of one hand‑editedindex.html. - Checked by the release sweep (two independent reviewers, three stages) and a 40‑step release live test.
📦 Built‑in workflows
Flux Klein · Z‑Image Turbo · SD 1.5 (image) · Trellis2‑GGUF (mesh) · Qwen Image Edit · SAM3 Isolate (i2i / edit) · LTX‑2 i2v (video) — each with an honest license badge (Apache‑2.0 · Apache‑2.0 · CreativeML Open RAIL‑M · MIT + model card · Apache‑2.0 · GPL‑3.0 · LTX‑2 Community License + Gemma Terms). Voice: Chatterbox (MIT).
Known limitations
Category overrides still don't reach the Edit tab; a second ComfyUI instance (endpoints.comfyui_bild) is set in phoenix-config.json only; a paired phone can't start paid jobs. Fixes from the release live test and machine‑readable marking of AI output (EU AI Act Art. 50) are planned for 1.10.2. After that the focus is game‑ready assets — see the ROADMAP.
Getting started
git clone https://github.com/laconova/phoenix.git
cd phoenix
node server.js
See the README for ComfyUI, Blender and Unreal setup, and the ROADMAP for what's next.
Status: alpha — early and evolving. Issues and feedback welcome, here or via laconova.com.
Phoenix 1.10.0 - From props to shots
Phoenix is a creative's control room for local AI — a free, open‑source harness that drives your local generation toolchain from one browser workbench. 1.10.0 takes Phoenix from props to shots.
✨ Highlights
Storyboards
Lay shots out as a timeline — image, prompt, 1–15 s each, reorderable — and send any Library image to a board with one click (images are linked, never copied). Keep several boards, grouped in the Library under Tools. Each shot carries its references, its keyframe and its clip.
Make video
Turn a shot's keyframe and prompt into a clip, right from the storyboard. Free and local through ComfyUI with the built‑in LTX‑2 i2v workflow (clips of up to about 3 s with sound; ComfyUI 0.27+, ~38 GB via ⬇ Acquire, 48 GB RAM) — or optional and paid through fal.ai (Kling 3 Pro) with your own key; Phoenix shows the estimated cost and asks before anything is sent, and a running fal job can be stopped. Bring your own local video model too: + Add workflow now takes a Video (local) stage (e.g. Wan 2.2) and reads fps, frame grid and sizes from the graph.
Reference images
Generate from images instead of text alone: Flux Klein takes up to two references, the new built‑in Z‑Image Turbo (fast, photo‑real) one, with a Change slider for how far it may move away. Reference runs honour the workflow's own settings and the seed lock.
Per‑workflow settings
CFG, steps, sampler, size, denoise and the other tunables of each workflow live in its Advanced section — per workflow, with optional per‑category overrides, a reset and a seed lock (workflow-params.json; 1.9 values from palette.json are migrated on first start). The Edit tab gets a CFG slider.
Use Phoenix from your phone (optional, off by default)
Phoenix publishes itself to your own Tailscale network over HTTPS (never public, port 7777 is never exposed). Pair a phone once by QR code and confirm it on the laptop; disconnect it there any time — and it stays disconnected. The phone can chat, watch jobs and browse the Library; API keys, MCP, Settings, the style palette, paid jobs and opening folders stay laptop‑only.
And also
- Images and Videos tabs in the Library — every generated or edited image and every clip is recorded.
- Clip sound — a per‑shot switch mutes or unmutes a local clip's own sound in every player (no re‑render, no ffmpeg).
- Chatterbox on this PC — the voice dispatcher ships: turn on Settings ▸ Voice ▸ Run voice engines on this PC and Acquire Chatterbox (Windows included).
- Acquire troubleshooting — disk full, gated Hugging Face models, interrupted downloads and a wrong ComfyUI folder each get an entry the error links to.
- Feedback straight from the app by mail, and the onboarding now helps you pick your brains.
- Hardening from the release sweep (two independent reviewers, three stages): durable phone disconnect, fal status handling that never drops a running request, local video that cancels a lost prompt.
📦 Built‑in workflows
Flux Klein · Z‑Image Turbo (new) · SD 1.5 (image) · Trellis2‑GGUF (mesh) · Qwen Image Edit · SAM3 Isolate (i2i / edit) · LTX‑2 i2v (new, video) — each with an honest license badge (Apache‑2.0 · Apache‑2.0 · CreativeML Open RAIL‑M · MIT + model card · Apache‑2.0 · GPL‑3.0 · LTX‑2 Community License + Gemma Terms).
Known limitations
Category overrides don't reach the Edit tab yet; a custom video workflow's fps field doesn't change the clip's frame rate; an open storyboard keeps the previous workflow's label after a switch; the Make‑video backend pick isn't remembered; reference keyframes can't be stopped. A second ComfyUI instance (endpoints.comfyui_bild) is set in phoenix-config.json for now. All of these are on the list for 1.10.1 — see the ROADMAP.
Getting started
git clone https://github.com/laconova/phoenix.git
cd phoenix
node server.js
See the README for ComfyUI, Blender and Unreal setup, and the ROADMAP for what's next.
Status: alpha — early and evolving. Issues and feedback welcome, here or via laconova.com.
Phoenix 1.9.0 - Bring your own brain
Phoenix is a creative's control room for local AI — a free, open‑source harness that drives your local generation toolchain from one browser workbench. 1.9.0 is about one idea: bring your own brain.
✨ Highlights
A provider per seat
Every role in Phoenix — orchestrator, troubleshooter, meta‑prompter, vision and category provider — now picks its own model source: your Claude subscription (Claude CLI), your ChatGPT plan (Codex CLI), a local OpenAI‑compatible server (LM Studio, Ollama, llama.cpp) or an API key. Put a big model where it matters and a small, fast one where it doesn't. Models that drive tools have to pass a tool test first (local ones also a short benchmark), and a non‑Claude vision seat has to pass a vision test — a colour card proves it really looked at the render before it may judge one. Keys live in .secrets/ only, never in the config, a response or a log.
Ask / Auto
Ask stops at the gates you switched on. Auto runs the job to the end within a step budget you set, and stops only when the render check doesn't clearly confirm a step.
Checkpoints + rewind
Before each scene‑changing step Phoenix saves a copy of your open Blender scene, together with the chat at that moment. Rewind to any checkpoint — and undo the rewind if you change your mind.
Phoenix as an MCP server
An external agent (Claude Code, Codex, Cursor) can call Phoenix as a tool: generate props, import them and check the result with Phoenix's render vision.
Remesh via Meshy (optional, paid)
Rebuild a staged mesh's topology in Meshy's cloud. A quad result gets its quads restored on import (GLB can only store triangles). Off by default; nothing is uploaded until you enable it and press the button.
And also
- GPU share (best effort) — before an image or mesh stage Phoenix asks a local LLM server on the same card to unload its model (waits for confirmation on the llama.cpp router).
- Resizable, collapsible panels; Settings split into General and Connections.
- Hardening from the release review: same‑origin protection on every state‑changing route, a strict argument allowlist for the Claude CLI, persistent fail‑closed bookkeeping for paid remesh jobs.
- The default orchestrator is now
opus— the alias that always follows the newest Opus.
Getting started
git clone https://github.com/laconova/phoenix.git
cd phoenix
node server.js
See the README for ComfyUI, Blender and Unreal setup, and the ROADMAP for what's next.
🎬 Watch what's new: https://youtu.be/EuxETcEaCvE
Status: alpha — early and evolving. Issues and feedback welcome, here or via laconova.com.
Phoenix 1.8.1 - Multi-engine voice + a library that ships with nothing
Phoenix is a creative's control room for local AI — a free, open‑source harness that drives your local generation toolchain from one browser workbench. This release makes Phoenix far easier to install and adds a voice‑engine picker.
✨ Highlights
A library that ships with nothing
Every built‑in workflow now carries an install manifest — its license plus the exact models and custom nodes it needs — so a fresh clone downloads no heavy weights. The Workflow Library shows a license badge and a real installed / not installed status per workflow, and ⬇ Acquire fetches only what a workflow you actually use is missing. Acquire runs as a cancellable background job with live progress, and it never deletes a directory it didn't create.
Multi‑engine voice
The Sound Design workbench gains an engine picker. CrispASR (the integrated default) runs against your own CrispASR speech server and works today. Chatterbox (MIT) and CosyVoice appear in the picker but route through a voice‑service dispatcher that ships in a later version, so they stay greyed out until then. /speak never silently falls back to an unavailable engine — it tells you what's missing.
📦 Built‑in workflows
Flux Klein · SD 1.5 (image) · Trellis2‑GGUF (mesh) · Qwen Image Edit · SAM3 Isolate (i2i / edit) — each with an honest license badge (Apache‑2.0 · CreativeML RAIL‑M · MIT + model card · Apache‑2.0 · GPL‑3.0).
Getting started
git clone https://github.com/laconova/phoenix.git
cd phoenix
node server.js
See the README for ComfyUI, Blender and Unreal setup.
Status: alpha — early and evolving. Issues and feedback welcome, here or via laconova.com.