AURIFEXA Canvas

Why a node canvas beats one prompt at a time

In a chat window, every prompt is a dead end. Intermediate results scroll away, and any task with more than one step turns into copy-and-paste. AURIFEXA Canvas takes a different approach: an infinite canvas that runs in your browser with nothing to install. You place nodes for text, images, video, audio, and notes, then draw connections between them to build workflows you can actually see.

Those connections do real work. Each line feeds a node's output into the next step automatically, and inside any prompt you can reference upstream results directly with @image1 or @text1. Instead of a disposable conversation you get a reusable pipeline: build it once, run it as often as you like, and adjust a single node without touching the rest.

BYOK: your keys, your rates, zero markup

AURIFEXA Canvas is built around Bring Your Own Key. You connect the API key of the model platform you already use and pay that platform directly, at its own rates. AURIFEXA adds no markup and takes no cut per image or token. For text models, any OpenAI-compatible endpoint works, so you can pick providers freely and switch whenever pricing or quality shifts. The tool itself is free once you register; your only costs are what your own provider bills for actual usage. At high image volumes, the difference against services that resell model access adds up quickly.

The local ComfyUI route: images that never leave your machine

For privacy-sensitive work, this is the decisive feature. Connect the canvas to a ComfyUI instance running on your own computer and it becomes a clean front end for local generation: every image is produced on your own hardware, and the files never leave your machine. That materially simplifies GDPR assessments for unreleased product shots, internal drafts, or client material. You keep the overview and orchestration of the canvas while data and compute stay entirely under your control.

Groups, batches, and @references in real work

Three features carry day-to-day production:

• Groups: gather related nodes into one group and run the whole chain with a single click. A workflow you build once becomes the template for every following round.
• Batch generation: an image node produces up to 12 variants per run, tiered at 1, 4, 8, or 12 by membership level. Any result can be split off into its own node and refined separately.
• @references: @image1 and @text1 pull earlier outputs into new prompts precisely, with no manual carrying of content.

Three scenarios show what this looks like in practice:

1. E-commerce product series: a grouped workflow renders consistent series images for each product variant; batches supply alternatives, and you split off the strongest one for refinement.
2. Marketing campaign visuals: a text node develops the core message, and downstream image nodes pick it up via @text1 to produce multiple motifs and formats.
3. Interior concepts: starting from a reference image, @image1 drives variations of the same room in different styles, side by side on one canvas.

A queue that keeps you moving

Jobs run through a local task queue. Depending on your tier, 1, 3, 6, or 10 jobs run in parallel; anything beyond that lines up automatically and starts the moment a slot opens. Kick off a full series, then keep working elsewhere on the canvas while it processes.

Getting started

AURIFEXA Canvas is free to use after registration, with the interface available in English, German, and Chinese. If you already hold API keys or run ComfyUI locally, connecting either takes minutes, and scattered one-off prompts become a workflow that is visible, repeatable, and predictable in cost.

Frequently asked questions