High Quality Dreambooth (JP's repo) trained checkpoint of Arnold using HD film stills from the movies "Predator" & "Total Recall".
There are a couple LORAs of Arnold on Civitai, but imho for likeness this checkpoint will far out perform the Lora models. I'm going to stick w/ high quality checkpoints for people unless my attempts to extract to Lora are only marginally inferior. It's pretty easy to extract a Lora from these checkpoints on your own using the "Supermerger" extension for A1111. If you do make a Lora from this checkpoint (IE: if it works better w/ your workflow) please just keep it for personal use and don't share it out that way.
Use Hi-Res fix with a denoise strength in the .2 range for best results
Please post any images you created with this model that you enjoy - It's motivating to see people digging models
Trained locally w/ a 4090 using Joe Penna's Dreambooth repo (requires 24gb VRAM and in my experience out performs the extensions/etc).
Regarding the samples: I like to use controlnet w/ random movie still shots to put a model somewhere in frame or w/o something interesting going on. I'll usually grab some stills from Bing Images, or places like Shotdeck, throw them in a folder, and batch run them w/ either Canny (ending slider set 50 or below), or the Openpose model set a little below 100%. Due to this, some of the sample images won't be reproducible if you use the information Civitai breaks out. Also a few of the sample images have the original filename before I renamed the file to upload here - same exact thing.
High Quality / Dreambooth (JP's repo) trained using HD stills from mostly the movies "Predator" & "Total Recall"
1. The rights to reposted models belong to original creators.
2. Original creators should contact SeaArt.AI staff through official channels to claim their models. We are committed to protecting every creator's rights. Click to Claim
