{ "id": 138717, "modelId": 126821, "name": "v1.0", "createdAt": "2023-08-11T15:17:34.568Z", "updatedAt": "2023-08-11T15:22:17.223Z", "trainedWords": [ "ren_dears", "bodysuit" ], "baseModel": "SD 1.5", "baseModelType": null, "earlyAccessTimeFrame": 0, "description": "
Trained on 70 images.
", "stats": { "downloadCount": 106, "ratingCount": 1, "rating": 5 }, "model": { "name": "Ren \u30ec\u30f3 / DearS", "type": "LORA", "nsfw": false, "poi": false, "description": "This LoRA seems very model dependent. It appears to work fine on models like MeinaUnreal and YesMix at high weight but is blurry on others like Anything or AnyLoRA. It also varies by prompt. NSFW generations seem to work well on most models. I recommend playing around with different checkpoints and weights.
Keyword is ren_dears. Standard outfit is bodysuit.
Issues
Forehead mark isn't accurate in generations and requires post-processing to add on (like in example images).
Sometimes hair is too long, sometimes hair is too short. You can add absurdly_long_hair, twintails, and antenna_hair for consistency.