A single Lora Locon to allow prompting of mini giantess size people that appear 6-30+ feet tall. The goal is to teach models to tweak the scale of the background and provide full body poses.
Flux 2 Klein 9b:
• Trained only using mesogts and macrogts as separate concepts, macrogts is maybe undertrained.
Z Image Turbo:
• Trained only macrogts and mesogts as separate concepts, macrogts is maybe undertrained.
Flux 1.D:
• Similar training to PDXL, trained all trigger words as separate concepts.
Illustrious XL:
• Trained similarly to Pony, possibly undertrained, not certain if it will actually produce better results than the SDXL base lora, but worth a shot.
Pony LyCoris:
• Re-trained on PDXL v6, see version details, as enabled flip augmentation which was missed during SDXL training.
• AIO Concept, trained as a single concept separated by tagging, prompts easier with more unpredictable sizes.
• Separate Concepts, separated sizes into their own concepts, more consistent sizes but may not trigger at as low strengths.
SDXL LyCoris What's New:
· Re-trained as a SDXL LyCoris using kohya_ss instead of dreambooth.
· Tested on SDXL base, Dreamshaper XL and Protovision XL
· Slightly suspicious of Kohya resizing to rank 64, so the much larger original trained version is included although all prompts are with the resized v1.
· For older versions and more details, checkout the pre-SDXL version.
Trigger words are tagged to better prompt approximate scaling using mg-sg guide:
• tallgts: Taller than average
• minigiantess: Giantesses 7' to 12' tall
• mesogts: Giantesses 12'to 50' tall
• macrogts: Giantesses 50' to 500' tall
Also training data is tagged by subject:
• solo: one subject
• couple: two subjects
• group: one minigts subject and a group of people