Higgsfield Canvas Tutorial: Build Video Chains Step by Step

Higgsfield Canvas is Higgsfield’s node-based board for chaining image and video models into one workflow. This Higgsfield Canvas tutorial covers the steps, costs and templates: connect an image, a prompt and a model, then run the chain to get a clip. A Starter plan costs $19 per month (checked 5 Nov 2026) and includes 270 credits per month.

What is Higgsfield Canvas?

Canvas is a workspace inside Higgsfield where each step of a job is a node on a board, and you connect the nodes in the order they should run. One node can hold an image, the next can hold a prompt, and the next can hold a video model. When you run the board, the output of one node feeds the next node, so a product photo can become a scene, and that scene can become a moving clip, all in one place. Higgsfield describes it on its own pages as a node-based board for chaining models in sequence, and the same idea appears in Higgsfield’s own how-it-works guide.

The official Canvas page at higgsfield.ai/canvas shows the idea in one picture: a product image of green shoes leads to an image of a person wearing them, then to a frame with motion blur, then to a final video frame. A prompt node sits beside the frames, and a second node carries the line “Clone That’s cool!” The page also lists two starting points above the board: “All Canvases” for boards you have already made, and “Templates Quick Start” for boards that are ready to copy. You need to sign in before you can view your own canvases, because the page shows “Sign in to view your canvases” to signed-out visitors.

How to use Higgsfield Canvas: step-by-step

The steps below follow one simple chain: a product photo becomes a scene, and the scene becomes a short video. Each step lists its credit cost from the official Higgsfield pricing page, checked 5 Nov 2026. Steps that only arrange the board cost nothing; only the steps that generate an image or a video spend credits.

1. Open Canvas and sign in (0 credits). Go to higgsfield.ai/canvas and sign in to your Higgsfield account. Your saved boards appear under “All Canvases”. If you are not signed in, the page only shows the sign-in prompt.
2. Start from a template or a blank board (0 credits). Pick a template under “Templates Quick Start” if you want a ready-made chain, or start blank if you want to build your own. A template is a starting point: you still swap in your own image and prompt.
3. Add your first image node (0.12 to 2 credits). Upload a product photo, or generate one inside the chain. A Soul 2.0 image costs 0.12 credits per image, and a Nano Banana Pro image costs 2 credits per image, according to the official pricing page (checked 5 Nov 2026).
4. Add a prompt node (0 credits). Write what should happen in the scene in plain words: the setting, the action and the camera move. The prompt node is text only, so writing it costs nothing. Keep it to one clear action per node so the result stays close to what you asked for.
5. Connect the image and prompt to a video model node (about 8 to 60 credits per 5 seconds). Pick the video model for the clip. Kling 3.0 at 1080p costs about 8 credits per 5 seconds, Veo 3.1 costs about 40 credits per 4 seconds, and Seedance 2.0 at 1080p costs about 60 credits per 5 seconds, all checked 5 Nov 2026. The full Higgsfield models list explains what each model does best, and the text-to-video guide explains how the same models behave outside Canvas.
6. Run the chain and check the preview (cost of the generation nodes). Running the board generates each node in order. Check the preview before you add more steps: if the first image is wrong, fix it there instead of paying for a video built on a bad image.
7. Add a second step to the chain (2 to 14 credits per 5 seconds). Feed the finished clip into a lip-sync node to add a voice, or into a second video model for a new angle. Kling Lipsync costs about 2 credits per 5 seconds, and Higgsfield Speak 2.0 costs about 14 credits per 5 seconds, checked 5 Nov 2026.
8. Download the finished video (0 credits). The download itself costs nothing; only the generations in the chain spend credits. Save the clip from the video page, as explained in the Higgsfield download guide.

How much does a Canvas chain cost?

A Canvas chain costs the sum of its generation nodes. Arranging nodes, writing prompts and downloading the finished clip cost nothing. The table below uses the per-model costs on the official Higgsfield pricing page, checked 5 Nov 2026.

As a worked example, a chain that makes one Soul 2.0 image (0.12 credits) and turns it into a 5-second Kling 3.0 clip at 1080p (about 8 credits) costs about 8.12 credits in total. A Starter plan with 270 credits per month (checked 5 Nov 2026) pays for about 33 chains like that, before any other generation. A longer chain with a Veo 3.1 clip and a Speak 2.0 step costs far more, so check the preview after each node instead of running a long chain twice.

Canvas templates: where to start

If a blank board feels too open, start with a template. The Canvas page offers a “Templates Quick Start” section and a Templates search, so you can pick a board that already has the nodes in place and then swap in your own image and prompt. A template saves the setup time, but it does not save the generation credits: every image and video node in the template still costs the same credits when you run it. Treat a template as a layout to learn from. Run it once with a cheap model, such as Kling 3.0 at 720p for about 7 credits per 5 seconds, before you switch any node to a dearer model such as Veo 3.1.

Canvas vs the normal Studio

The normal Higgsfield Studio takes one prompt and makes one image or one video. Canvas takes a chain of prompts and models and makes a finished result in stages. The Studio is faster when the job is one clip. Canvas is better when the job has steps that depend on each other, for example when the video must show the same product that appears in the first image, or when the same character must appear in three scenes in a row. Both use the same credit balance and the same plan. The difference is control: in the Studio you accept the first result or try again, while in Canvas you can fix one node and re-run only the steps after it.

One limit applies to Canvas that does not apply to the main site. The official pricing page states that unlimited models and free generations on plans are accessible only via higgsfield.ai and are not accessible on MCP, CLI, Canvas or Supercomputer. In practice, that means a Plus or Ultra plan’s unlimited Seedance 2.5, Nano Banana Pro, Seedream 5.0 Pro and GPT Image 2.6 access does not carry over to Canvas: generations inside Canvas spend credits at the per-model rates in the table above. Plan the chain around credits, not around unlimited access.

Pros and cons

Pros

• One board holds the whole job, from the first image to the finished video, so you do not export files between tools.

• Each node can use a different model, so you can pair a cheap image model with a stronger video model.

• Templates give you a working chain to copy and change, instead of a blank board.

• You can fix one node and re-run only the steps after it, instead of starting the whole job again.

Cons

• Unlimited models and free generations do not work on Canvas, so every generation spends credits.

• A long chain adds up fast: a Veo 3.1 clip plus a Speak 2.0 step can cost more than 50 credits for a few seconds of video.

• A node board takes longer to learn than a single prompt box, and a wrong early node wastes the credits spent on the steps after it.

• The Free plan has 0 credits, so Canvas is not usable without a paid plan or bought credits.

• There is no native app, so Canvas runs in a browser and needs a steady connection while a chain runs.

Who should NOT use it

Canvas is a poor fit if you only need one quick clip from one prompt, if you are on the Free plan with 0 credits, or if you want unlimited-model access to carry over, because the official pricing page states that unlimited models do not work on Canvas. Use the normal Studio for single clips instead.

Frequently asked questions

Is Higgsfield Canvas free?

No. Canvas itself has no separate fee, but every image and video it generates spends credits from your plan. The Free plan includes 0 credits, so you need at least the Starter plan at $19 per month (checked 5 Nov 2026) with 270 credits per month to run a chain.

How many credits does a Canvas chain use?

It depends on the nodes: a Soul 2.0 image costs 0.12 credits, a 5-second Kling 3.0 clip at 1080p costs about 8 credits, and a Veo 3.1 clip costs about 40 credits per 4 seconds. A simple image-to-video chain costs about 8.12 credits in total, checked 5 Nov 2026.

Can I use Canvas on my phone?

Yes, in a mobile browser. Higgsfield has no native app for iOS or Android, so Canvas runs in the browser on a phone or a computer. The same account and credit balance apply, and a chain built on a laptop opens the same way on a phone browser.

Do unlimited models work on Canvas?

No. The official pricing page states that unlimited models and free generations are accessible only via higgsfield.ai and are not accessible on Canvas, MCP, CLI or Supercomputer. Generations inside Canvas always spend credits at the per-model rates.

Can I save and reuse a Canvas board?

Yes. Saved boards appear under “All Canvases” on the Canvas page, and templates under “Templates Quick Start” give you a ready-made chain to copy. You need to be signed in to view your own canvases.

Sources checked

• Higgsfield Canvas — https://higgsfield.ai/canvas
• Higgsfield pricing — https://higgsfield.ai/pricing
• Higgsfield home page — https://higgsfield.ai/

Checked 5 Nov 2026.