🚀 LoRAtech Advanced Prompt Builder — going live today (Patreon)

I just finished building the biggest node I've put out yet, and quite possibly THE MOST DIVERSE ComfyUI node to ever exist, and it's dropping today. This one replaces manual prompt typing entirely with a full visual builder, and I want to walk you through what's actually in it before it goes live.

🧩 WHAT IT IS

One ComfyUI node, two modes, click-to-build prompts instead of typing them from scratch. No more staring at a blank text box trying to remember the right words for lighting or camera angle. You click through categories, the node assembles a natural-language prompt in real time in a live preview, and it feeds straight into your CLIP Text Encode.

🎭 CHARACTER & LORA PANEL

Sits at the top of the node regardless of mode. Type a trigger word, pick a LoRA from your models folder, set model and CLIP strength, and save the whole combo as a named character preset with its own thumbnail. Load a saved character any time and every prompt built from that point carries the trigger word automatically.

📸 INFLUENCER MODE

This is where it gets deep. Flip the mode toggle and the categories change entirely.

1. Clothing expands into Swimwear, Lingerie, Dresses, Gym Fits, Cosplay, Bedroom, and Fantasy

2. Pose and Camera split into focused subcategories like Selfie Styles, Framing & Distance, and Angles & Perspective

3. Background goes from a handful of studio backdrops to 8 full location categories (Home & Bedroom, Café & Dining, Shops & Beauty, Gym & Wellness, City & Streets, Nature & Outdoors, Travel & Luxury, Studio & Backdrop) covering 75+ real locations

4. Body, Face, Hair, and Makeup become their own dedicated categories, 43 separate physical trait fields, sourced and adapted from a proper character-generation dataset

You're not writing "pretty face" anymore. You're picking exact bone structure, skin tone, eye shape, and forty other traits individually.

💼 PROFESSIONAL / BUSINESS MODE

Your standard commercial toolkit. Nine categories: Medium, Subject, Clothing, Pose, Background, Camera, Lighting, Processing, Mood. Every option is backed by an actual reference thumbnail, so you're not guessing what "Rembrandt lighting" or "three-quarter angle" looks like, you're picking it visually. This is what I'd reach for on product displays, commercial shoots, anything that needs to look like it came out of a real studio.

🎲 THE RANDOMIZER

Four presets:

1. Randomize All — rolls every field in the node

2. Keep Character — rolls everything except physical appearance, so your character's identity stays locked while outfit, pose, and scene change

3. With LoRA — same as above but wipes the physical fields entirely, since your LoRA is already defining the look and you don't want the prompt fighting it

4. Custom Preset — check exactly which categories and subcategories are in play, everything else gets cleared

There's also a Random Seed checkbox that automatically re-rolls your last-used randomizer setup before every single queue, and it disarms itself the moment you touch a field by hand, so it never overwrites a manual choice by accident.

🔢 THE NUMBER THAT MATTERS

I sat down and actually counted the combination space, field by field, no rounding up for effect.

Professional mode: 173,454,177,595,449,644,413,186,867,200,000,000 possible combinations. That's a 36 digit number. There are an estimated 10 sextillion stars in the observable universe. Professional mode alone gives you roughly 173 billion times more combinations than there are stars in the sky. ✨

Influencer mode: 1,605,262,742,575,249,332,568,564,880,484,993,041,331,431,003,707,047,836,619,154,391,040,000,000,000,000,000 possible combinations. That's an 82 digit number. Scientists estimate there are about 10 quinvigintillion atoms in the observable universe. Influencer mode has roughly 16 times MORE possible combinations than there are atoms in every star, planet, and speck of dust in the observable universe, combined. 🤯

You could generate one image every second, nonstop, since the Big Bang, and you would not have made a dent. This node will never run dry.

🔧 WHY IT'S BUILT THIS WAY

The whole point is that "literally anything" is the use case. Professional product displays for client work. Randomized prompts for a character LoRA you already trained, to get infinite variety without breaking the likeness. Rolling the dice on totally random people until one of them sparks an idea for a brand new character LoRA. Or just hitting Randomize All and seeing what comes out, no plan, no direction, just seeing what the node gives you. It's built to cover every one of those without you switching tools.

🔥 This goes live today. If you're on Patreon, you're getting it before anyone else does.