Type | |
Stats | 221 |
Reviews | (23) |
Published | Mar 11, 2024 |
Base Model | |
Hash | AutoV2 B8E59D4C8F |
This model began as my own experimental training project. AmuzV1 (never released) was trained until I felt it was reaching diminishing returns.
With AmuzV2, I decided to start adding some Loras and a few Checkpoint merges. This helped improve the original quite a bit.
With AmuzV3, I have utilized several more Checkpoint models to merge. I have also experimented with WASasquatch's merge preset node for ComfyUI.
In the end, a few of my own Loras and Checkpoints were utilized and several from other creaters.
Here are the checkpoints, created by others, that I have used...
Tessilation https://civitai.com/models/279522/tessilation
Protovision https://civitai.com/models/125703?modelVersionId=265938
Juggernaut https://civitai.com/models/133005?modelVersionId=348913
Dark Arts Images https://civitai.com/models/321687/dark-arts-images
Realism from Hades https://civitai.com/models/193834?modelVersionId=376023
Zaxious https://civitai.com/models/324944?modelVersionId=364284
Ultima https://civitai.com/models/326287?modelVersionId=365741
AmuzV3 works with any Sampler and Scheduler, but I tend to use Euler_A or DPM++_2M generally with 20 - 30 steps and cfg around 5 - 7.
Experiment... have fun..