Can I save a pipeline as a template?

Yes — save any canvas as a template and reuse it across projects. Parametrised inputs (prompts, reference images) swap per-run; node graph stays stable.

Does it cost more than individual generations?

No. Each node costs the same as running that model standalone. Canvas is a productivity layer, not a premium tier.

What if a node fails mid-pipeline?

The rest of the graph keeps running where possible. Failed nodes flag red; you can retry individual nodes without rerunning the whole pipeline.

Workflow Group

Node-Based AI Workflows.

Drag nodes, connect outputs to inputs, build branching AI pipelines visually.

Canvas is Stensyl's node-based workflow editor. Chain 24 node types across image, video, 3D, audio, and text into pipelines that execute in sequence or parallel. Character consistency nodes keep subjects identical across every step.

Open Canvas

Pipeline-grade AI creation, visually assembled.

24 node types

Every Stensyl model available as a node. Image gen, video gen, 3D gen, audio, text, upscale, edit, reference — all drag-and-drop.

Branching pipelines

Run variations in parallel, compare side-by-side. A/B test prompts, models, or parameters without rebuilding.

Discipline templates

10 pre-built templates: architecture visualisation, product reveal, brand identity, social campaign, and more. Start with scaffolding.

Open Canvas

Build complex pipelines without code.

Traditional creative automation means scripting. Canvas makes it visual: drag a generation node, connect its output to an upscale node, connect that to a video node, branch to a 3D node. What would take a developer an afternoon takes a designer 15 minutes.

Character consistency across nodes.

Character Reference nodes lock a subject identity across the whole pipeline. Generate a portrait, feed it as a reference into 10 downstream generations, and the subject stays consistent across every one. Essential for multi-shot film work and catalogue photography.

See 3D generation

Sequential or parallel execution.

Run nodes in sequence when each step feeds the next. Run them in parallel when you want variations to compare. Canvas handles the execution graph automatically — you just connect the sockets.

Compare, iterate, repeat.

Branched outputs render side-by-side. Pick the winners, mark the losers, rerun with tweaked parameters. The visual graph makes it easy to see what changed between runs and why one branch produced better output than another.

Models you can use.

Every image model

Any image model as a node.

Learn more →

Every video model

Any video model as a node.

Learn more →

Every 3D model

Any 3D model as a node.

Learn more →

Every audio model

Any audio model as a node.

Learn more →

Frequently asked.

No hard cap. Practical limit is execution time and credit budget — a typical pipeline has 4–10 nodes. Branching multiplies output, so parallel runs add up fast.

Ready to start? Jump straight in.

Every Stensyl plan includes every model on this page. From £8/month annual. No per-model fees, no surprise charges.

Open Canvas See Pricing