Disclaimer: This model is pretty wild, so don't expect perfect generations. Newer versions will probably fix the issues but over time. Future versions I will have after V0.8 will have, 350 to 1,000 image datasets of the images I get without a LoRA that will be trained on the current model version and put onto the model and released and other images on my page will be also in the dataset. Also V0.8 Went back to being illustrious V2.0 stable.
This is going to be a illustrious 2.0 stable finetune of my own. Also needs heavily tested for V0.9. If anyone can make a list of what this model can't do yet, I'd appreciate so I can put those things into V1.0.
Keep in mind you do not have to use any of the below stuff.
Recommended Settings V0.98:
Steps: 20+
CFG: 1 to 6, 7 and higher ONLY if you're doing abstract and want a overbaked effect or more contrast
Sampler/Scheduler: Whatever works best
Recommended Pos (you can add or remove):
masterpiece, best quality, absurdres, aged_upRecommended Neg (you can add or remove):
worst quality, low quality, artistic error, greyscale, monochrome, loli, shota, young, aged_downRecommended Settings V0.8 / 0.9:
Steps: 20+
Minimum CFG: 3+
Sampler/Scheduler: Whatever works best
Must have these in your Pos Tag as a base (you can add or remove):
best quality, aged_upMust have these in your Neg Tag as a base (you can add or remove):
low quality, artistic error, greyscale, monochrome, loli, shota, young, aged_downNote: Only because this is Illustrious V2.0 mainly with LoRAs
Recommended Settings V0.5:
Steps: 20+
Minimum CFG: 3+
Sampler/Scheduler: Whatever works best
Must have Pos Tags:
aged_upMust have Neg Tags:
loli, shota, child, young, aged_downNote: Only because this is Illustrious V2.0 mainly with LoRAs that only effect style mainly, and if you've seen the Danbooru tags you'd know why.
Recommended Settings V0.3:
Steps: 20+
Minimum CFG: 4+
Sampler/Scheduler: Whatever works best
Must have Pos Tags:
aged_upMust have Neg Tags:
loli, shota, child, young, aged_downNote: Only because this is Illustrious V2.0 mainly with LoRAs that only effect style mainly, and if you've seen the Danbooru tags you'd know why.
Description
Put slight NAI in here like 7% then simple merged in a Illu V2.0 Stable + LoRA mix I did, which some of that 7% of NAI Eps is probably gone but the merge version is how I want it to be.
Just hope to fix this to be by default not be a mix of ages without specific tags, on base generation eventually with LoRA training now for future versions.
FAQ
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.