Higgsfield Automation — User Guide

Bulk image & video generation on higgsfield.ai, straight from a side panel.

This extension turns a list of prompts into a queue and runs them on higgsfield.ai for you — generating, downloading, and naming the results while you do something else. You pick your model, aspect ratio, duration, and resolution on higgsfield itself; the extension runs the queue in that page and handles the bulk work.

The four modes

Pick a mode at the top of the Control tab. Each works on a specific higgsfield page.

Entering prompts

Type one prompt per block, separated by a blank line:

• A prompt can span multiple lines — only a fully blank line starts a new prompt.
• You don't have to type them by hand — see Importing prompts below.

Power features

📄 Import prompts from a spreadsheet or text file

Above the prompt box are two buttons. Upload .txt loads a plain-text file as your prompts. Upload spreadsheet reads an .xlsx, .xls, or .csv file:

• It scans your sheet and auto-selects the column with the most text as your prompt column.
• A preview shows the first few prompts and the total count.
• You can switch to a different sheet or column before importing.
• On import, each row becomes one prompt (separated by blank lines automatically).

🔢 Outputs per Prompt — different for images vs videos

The Outputs per Prompt control means two different things depending on what you're generating:

⚡ Concurrency & pacing

• Concurrent Prompts — how many generations may be in flight at once. Submission stays orderly; this controls how many results are resolving/downloading in parallel before the next prompt starts.
• Random delay — a randomized wait between prompts (and between repeats of a video prompt), so runs are paced rather than hammering the page. Set a min and max in seconds.

🔁 Retries on failure

If a generation fails to start (no result appears), the extension automatically re-attempts that prompt up to your Max Retries setting before marking it failed. The queue shows Retrying ×n while it tries again.

🧩 Auto-add character images (I2I)

Turn this on in Image-to-Image and name your uploaded images after characters (e.g. alice.png, bob.png). Any prompt that mentions a character's name automatically uses that image as its reference; prompts with no matching name run as plain text-to-image. One prompt can pull in several characters.

↪️ Chain a prompt from the previous output (concat)

Start a prompt with >> to feed the previous prompt's output into it:

• Video: the previous clip's last frame becomes this prompt's start frame — great for continuous shots.
• Image-to-Image: the previous image becomes this prompt's reference.

Example (continuous video):

• a lone astronaut steps onto a red planet
• >> the astronaut plants a flag
• >> a wide shot of the colony at dusk

A chained prompt waits for the previous one to finish (so its output is ready); other prompts still run in parallel.

🖼️ Reorder & sort uploaded images

In the modes that take images, the upload rack lets you drag images to reorder them and sort by name. Because images are matched to prompts in order, reordering directly controls which image goes to which prompt. You can also drag-and-drop image files straight onto the rack. A warning appears before you run if some prompts will have no image (and will run as text-to-image).

💾 Downloads, folders & naming

• Auto-download — when on, each result is downloaded as soon as it finishes.
• Folder name — results are saved into this subfolder of your Downloads folder.
• Auto-rename — names each file by its prompt, e.g. 02_a-cat_<id>.png for the 2nd prompt — correct even when generations finish out of order. Turn it off to keep higgsfield's original filenames (the folder still applies).