Batch Image Generation
One of the most powerful applications of Planned Prompting in Browser Bro is automating and scaling image generation workflows. Whether you’re a digital artist, a dataset engineer, or running a creative agency, this feature unlocks next-level efficiency and control.
Everyone knows the key to image generation is the quality of your prompt, the planned prompting workflow therefore, is to focus on the most important part first.
Workflow Overview
- Write Your Prompt List: Create a simple text file with a list of prompts, each on a new line. These prompts can describe scenes, characters, moods, or variations. One per line. (It can be a really long line)
- Open Browser Bro and Open the Scripting page
- Switch the Editor to “Text” and paste your prompts into it.
- Copy the script from the field below the video, into the right most Shorthand Script Field
- Log into Invoke AI, set your settings.
- Press Play Script when ready
- Use Loop + Delay: Use the
LOOP
function to iterate through your prompt file and send each prompt to your image generation tool (e.g., InvokeAI, ComfyUI, Stable Diffusion WebUI). - Point & Automate: Target the browser input field and button with the visual selector. Browser Bro sends the prompts one by one, auto-submitting as needed.
- Relax or Optimize: Go do something else. Or tweak the script mid-run without restarting. You’re now running a reproducible image generation pipeline.
Benefits
- Massive Time Savings: Queue hundreds of prompts in minutes instead of manually entering them one by one.
- Creative Consistency: Run variant prompts with controlled variables for style, subject, or composition.
- Dataset-Ready: Generate training datasets or visual content for machine learning, eBooks, or game assets with reproducibility.
- Works Offline: Ideal for local-first setups using tools like InvokeAI, Stable Diffusion or ComfyUI on your own hardware.
- No API Keys Required: Works with local UIs — no extra services or costs needed.
The time-lapsed demo below is not of model accuracy and generation accuracy, its about the workflow we call “planned prompting”. This is a local instance of invoke.ai and Browser Bro working together to generate an icon set:
Invoke AI Example:
go|http://localhost:9090/
set|~AllPrompts|text editor
loop|each line|~AllPrompts|~ThePrompt
focus|id|prompt
type|id|prompt|~ThePrompt
// TOUCH FIELD
js|const input = document.getElementById('prompt'); if (input) { Object.getOwnPropertyDescriptor(window.HTMLTextAreaElement.prototype, 'value').set.call(input, input.value + ' '); input.dispatchEvent(new Event('input', { bubbles: true })); }
// CLICK INVOKE
js|[...document.querySelectorAll("button")].find(btn => btn.textContent.trim() === "Invoke")?.click();
// WAIT FOR THE IMAGE TO BE GENERATED, ADJUST BASED ON GENERATION TIME OF MODEL
wait|8|minutes
loop|end