If you feel like showing some love, awesome.
But the best kind of support?
Use the model. 🛠️
Post your results. 🖼️
Let the art speak. 🎯
🛠️ Mechnifying Glass MG-MK.1
A Detail-Oriented LoRA for Machines, Mechanics, and Metal Mastery
Description:
Mechnifying Glass MG-MK.1 is a precision-trained LoRA built specifically to enhance mechanical realism, complexity, and surface detail in AI-generated imagery. Whether you're designing robotic beasts, armored vehicles, exposed tech internals, or industrial landscapes, MG-MK.1 sharpens everything — from rivets and screws to cables, vents, and layered plating — without drowning your output in noise.
🎯 What It’s For:
Mechanical creatures (cats, dogs, dragons, birds, etc.)
Vehicles (cars, trucks, bikes, boats)
Industrial scenes and factory interiors
Robotics, drones, mechanical parts
Sci-fi props, armor, gadgets, machines
This LoRA excels at clean hard surfaces, visible mechanical structure, and intricate layering — ideal for concept artists, hard-surface sci-fi creators, and mech designers.
⚙️ Settings & Usage
Base Model: Flux
Sampler/Scheduler: Euler + Beta
Steps: 40
🧠 Flux Guidance (CFG):
2.5–3.0 → Ideal for prompts involving realistic human skin or organic realism
3.5–5.0+ → Optimal for mechanical, robotic, sci-fi, or fully inorganic subjects
🛠️ LoRA Strength:
0.20–0.40 → Subtle detail enhancement
0.50–0.80 → Balanced composition and mechanical features
0.90–1.00 → Strong, high-impact mechanical detail
⚠️ Not extensively tested above 1.00 — may yield creative but unpredictable results.
Description
⚙️ Settings & Usage
Base Model: Flux
Sampler/Scheduler: Euler + Beta
Steps: 40
🧠 Flux Guidance (CFG):
2.5–3.0 → Ideal for prompts involving realistic human skin or organic realism
3.5–5.0+ → Optimal for mechanical, robotic, sci-fi, or fully inorganic subjects
🛠️ LoRA Strength:
0.20–0.40 → Subtle detail enhancement
0.50–0.80 → Balanced composition and mechanical features
0.90–1.00 → Strong, high-impact mechanical detail
⚠️ Not extensively tested above 1.00 — may yield creative but unpredictable results.
FAQ
Comments (5)
#AwesomeSauce
Any plans to create the same for SDXL please? That would be fantastic...
I don’t currently have plans to make SDXL / Illustrious / Pony versions of my LoRAs. I don’t use those models myself, so I don’t feel confident releasing something I can’t properly test or support.
That said, I am experimenting with Z-Image Turbo, so there’s a real chance I’ll explore making LoRAs for that ecosystem in the future.
@Razane oh I see... In case You'd change Your mind, I am all into pure SDXL :) Still the best in my humble opinion
@Whatever123456 I began my AI work with Flux. I’ve tried SDXL, but I’m still learning how to properly handle CFG and negative prompts, so I’m sticking to the tools I know well for now rather than releasing something half-baked.
I really appreciate the interest, and yeah... SDXL can be incredibly powerful in the right hands. If I ever feel confident enough working with it, I’d definitely start by porting my most popular model first.