<h1 id="osmosis-drix-8gmphcu29">Osmosis - Drix</h1>Character LoRA.My ComfyUI Workflow: <a target="_blank" rel="ugc" href="https://civarchive.com/articles/14582">Comfyui Workflow: SDXL FaceDetailer, Upscaler, &amp; Batch Selection</a>Strength: 0.9 to 0.1 is the range. 1.0 seem to be the bestTrigger Word: osmosisdrix, robotSampler: Euler_AScheduler: KarrasSteps: 20-30CFG: 6-7<h3 id="tips-11pjdooko">Tips</h3>Try including both the trigger word osmosisdrix and robot in your promptsTraining tags are included in the files section so you can prompt better outcomes -&gt;May need to prompt "looking at view" because a lot of the training sets for this LoRA were looking away from the camera."1boy" may generate other characters from the Osmosis show.<h3 id="training-tche1wfnf">Training</h3>Base model: <a target="_blank" rel="ugc" href="https://civarchive.com/models/404154?modelVersionId=1767402">WAI-ANI-NSFW-PONYXL</a>. Good for prompting.All images were tagged using WD14. Touched up colors on a few images to reduce the greenish tint.

Osmosis - Drix

illustrious was trained longer and appears that it may give better results but it could just be that illustrious is "better". Who knows.Dataset: 39 images

Files with SHA256: 0893f8c4782f441f3f3f5de6b94da817683de841680eec32e49211a0202440ad