Type | |
Stats | 56 0 |
Reviews | (7) |
Published | Jun 16, 2024 |
Base Model | |
Usage Tips | Strength: 0.9 |
Trigger Words | (Jeorg) |
Hash | AutoV2 F6770CE1D1 |
UPDATED INFO FOR V3...
Version 3 is far more accurate and realistic in every aspect! I wanted to redo my first ever lora I created and I'm very pleased with the results.
Recommended Settings...
On A1111 for realism for this particular lora I recommend DDIM with Karras, (DPM++ SDE + Karras is also good),
Steps 25-40 (30 is the sweet spot)
CFG: 5-6
Size: 512x768
Hires: Denoising Strength: 0.30, Hires upscale: x2, Hires steps: 15, Hires Upscaler I highly recommend: 8x_NMKD-Superscale_150000_G also (1x-ITF-SkinDiffDetail-Lite-v1 or R-ESRGAN 4x is also good).
Adetailer: mediapipe_face_mesh_eyes_only and use the <lora:lora_perfecteyes_v1_from_v1_160:1> photo of perfect eyes, detailed perfecteyes subtle blue eyes" in the positive prompt.
Old info for V1 is below....
This is my first attempt at creating a LORA created via CivitAI so please be gentle. This is a creation of one of the best online male twink/twunk muscle models around Jeorg. Please generate responsibly.
I am using my own model PhotonicFusion which is exclusive to MageSpace but I have tested this LORA on other models and seems to get okay results. Let's call this a beta version because I need to better learn the various settings when making the LORA to fine tune it. As it stands you may need to do multiple gens to get a good result and you will need to use Hires Fix either upscale only or sometimes upscale with face fix between 15-30 usually.
I'm using PhotonicFusion exclusive male focused model on MageSpace
I'm using Hires. fix with 0,15 to 0,30
I'm usually using DDIM with Karras, Guidance 5-6 Steps 30-50
Just one trigger word of Jeorg I usually set the LORA strength between 0.7 to 0.9. He is a shape shifter sometimes he's a "skinny fit lean twink" others times he's a "muscular twunk" varying up your prompt should help get the type of body size you are looking for. Please keep in mind MageSpace has their own unique backend hardware and processing so my suggestions above work well on MageSpace but may not work so great on any other personal rig or platform so you may need to experiment...