FableML is an AI film studio on an infinite node canvas. Characters, places, shots, and motion, kept consistent from the first frame to the finale by a Production Bible.
early access · your work never trains public models
One canvas, every model: Sora 2, Veo 3, GPT-Image 2, Gemini Flash Image, GPT-4o, Claude Sonnet, Wan i2v, P-Video, Gemini Flash, one token price list.
the reveal
The film you just watched take shape is exactly how FableML works: every step is a card you can rewire, rerun, and reuse.
Mara
character
The lighthouse
location
continuity ledger · do not contradict
her lantern cracked in shot 3 · keep it cracked
coat soaked since the rowboat scene
00:01 · the moat
FableML keeps a Production Bible: your cast, props, locations, and a running continuity ledger updated after every shot. A prop broken in shot 3 is still broken in shot 12, and your lead's face survives every model you render her with, this project and the next.
image
video
00:02 · the canvas
Wire text, image, video, and audio nodes into a living graph. Every card is a shot; every wire is a decision. Tell the copilot the idea and it drafts the whole board, and nothing renders until you say so.
duration, what sora-2 really renders
references
1 image frame · video refs blocked at the wire
price, before you run it
shot 09 · 38 dream tokens
00:03 · the controls
Pick the model per shot and see exactly what it can do: real clip lengths, real reference rules, and the token price, all before you spend anything. The budget never surprises you.
showcase
Concept renders and product demonstrations from our own pipeline, labeled as exactly that. Creator work lands here as early access opens.
sc 01 · the lighthouse
Storm-low FPV pass around the keeper's light
sc 02 · mara, take 3
Same face, coat, and scar in every pose
sc 03 · canvas overview
One brief, seven scenes, 31 nodes wired by the copilot
the workflow
Seven connected steps, not seven separate tools. Everything you make in one step is a reusable asset in the next, and the bible carries the story between them.
00:00
Tell the copilot the story in a sentence, it drafts the whole board as wired nodes. Nothing renders until you say so.
00:01
Cast your lead once. Saved to the Production Bible with reference art, she survives every model and every episode.
00:02
Stage locations in the 3D world or describe them, saved places reload in any project, exactly as you left them.
00:03
Generate each frame with the model that fits it, with the bible's continuity ledger riding along in every prompt.
00:04
Animate image → video per shot. Real durations and reference rules are shown up front, so nothing renders short.
00:05
Inpaint frames, trim and reorder the timeline, regenerate a single shot without touching the rest of the cut.
00:06
Assemble the sequence and export the film. The bible remembers everything for shot 13, and for season two.
pricing
One Dream Token currency. Pick the plan that fits how you create — a render always costs what it says before you run it.
Ideal entry point for creators testing the waters — wire up the canvas and every AI model, no card required.
See what you can get
Generation:
Features you'll love:
The watermark comes off. Unlimited canvases and downloads for everyday creating and commercial work.
See what you can get
Generation:
Everything in Free, plus:
Built for creators shipping real work at scale — every model, including premium video, with founder support.
See what you can get
Generation:
Everything in Standard, plus:
questions
The short version of how FableML works and what it costs. Anything else, the team reads every mail.
An AI film studio on an infinite node canvas. You wire text, image, video, and audio nodes into shots, and a Production Bible keeps your characters, places, and story facts consistent across every render, every project, and every episode.
One internal currency across every model. Each render shows its token price before you run it, so a shot never costs more than it said it would. Plans include a monthly allowance and you can top up whenever you need more.
Sora 2, Veo 3, GPT-Image, Gemini, Claude, and more, all on one canvas. You pick the model per shot and FableML shows each one's real clip lengths and reference rules up front, before you spend a token.
No. Your work never trains public models, and we never route your content through a training-enabled endpoint. For enterprise customers, the whole provider layer can run inside your own boundary.
References keep a face consistent for one shot. The bible keeps the whole story consistent: a cross-project library of cast, props, and locations, plus a continuity ledger that updates after every shot. A lantern cracked in shot 3 is still cracked in shot 12.
We're onboarding creators in small waves. Request early access and we'll reach out with your invite. Free trial accounts get daily tokens to try the canvas before picking a plan.
enterprise & schools
Seat-based plans for teams and classrooms, isolated tenancy for your organization’s data, single sign-on, and a Production Bible your whole crew shares. We onboard enterprise customers personally.
Talk to usWe’re onboarding creators in small waves. Join the waitlist and we’ll reach out with your invite. Bring an idea, leave with a film.
Request early accessearly access · limited seats
talk to the team
We’re a two-person team onboarding creators, studios, and schools personally. For early access, partnerships, or enterprise questions, write to either of us.