Pick a tool. Tell it what you want. It makes it.

This is where everything gets created — images, video, voice. Nothing here needs approval, because you're the one creating it. Once something's ready, send it to the Admin Command Center to manage and publish.

If you want a finished video, start here
LIVE — TESTING

All-in-One Video

Describe the video you want in plain words — a topic, a length, a tone. This one tool does everything else for you: writes the script, finds or makes the visuals, records the narration, and puts it all together into one finished video file. Powered by OpenMontage, installed and running on the studio's infrastructure.

1Type what you want — e.g. "a 60-second explainer about RRSP deadlines, calm tone, real footage"
2It researches the topic, writes a script, and shows you the plan before making anything
3It builds the finished video — narration, visuals, captions, all in one file
4You get the file, ready to send to the Admin Command Center or upload yourself
Or, just want one specific thing? On demand
READY NOW

Picture from Words

Type a description, get a picture. Can also use Lumira's or Lunara's trained face so it actually looks like them.
If you want: a single image for a post, a thumbnail, or a visual to go with an article — use this.
⚠ Runs on gpu-studio, which gets stopped between sessions — start it first, or this link will time out.
SHELVED

Still Photo to Moving Video

Take one photo and make the face move naturally — blink, talk, turn slightly. Known as LivePortrait.
Decision: dropped after fragile dependency issues — revisit only if a pre-built container becomes available.
NOT INSTALLED

Voice — Speech to Text

Upload audio, get the written words out of it. Known as Faster-Whisper.
If you want: captions for a video, or to turn a recorded voice memo into a script — use this, once it's set up on gpu-studio.
FIX IN PROGRESS

Voice — Text to Speech

Type words, get them spoken aloud in a natural voice. Known as Chatterbox TTS — self-hosted, zero per-use cost. A model-loading bug is currently being fixed.
If you want: narration for a video, or to turn a written article into something listenable — this will be the tool, once the current bug is resolved.
SHELVED

Lip-Sync a Talking Video

Give it a face and an audio file — it makes the mouth match the words. Known as MuseTalk.
Decision: dropped after fragile dependency issues — revisit only if a pre-built container becomes available.
SHELVED

Keep a Face Consistent

Makes sure the same face looks like the same person every single time, across many images. Known as FaceChain.
Decision: dropped after fragile dependency issues — the trained Lumira LoRA already handles consistency well enough for now.
CONFIRMED WORKING

Train a New Face

Feed it about 40 photos of a person, and it learns their face well enough to generate new images of them. Known as Kohya SS — this is exactly how Lumira's face was trained (saved LoRA already in use).
If you want: to train a brand-new persona's face, or retrain Lumira/Lunara with better photos — this is the tool. Runs via direct SSH execution only — no browser address yet.
UNVERIFIED

Quick Trim & Edit

Cuts, trims, and assembles video clips without the full production process. Known as MoviePy and Auto-Editor.
If you want: a fast trim or simple edit, and don't need a full script-to-finished-video production — use this instead of the All-in-One above.
EXTERNAL TOOL — ACTIVE

Cinematic Camera Control

A separate, already-finished product — not something installed here. Real camera angles, lighting, and lens control. Known as Higgsfield.ai.
If you want: true cinematic camera control and polish — this is a hosted third-party service, used directly, no local instance to check status of.