Final version of the model Latex Reality. model trained on latex images, then merged with bondage / shibari models, merged again adding the weights for gas masks/better hoods with an incredibly higher success rate. add the suggested negatives <s>and use "perfect eyes" in the positive prompt.</s> I suggest that you use the Lora better eyes.

Latex Reality V3.0 (Final)

This version works better at 1024x1024 (512 works too but the eyes are often broken or overlapping, add "overlapping eyes, anime eyes, amber eyes" to the negative prompt) (add "solo" and "1woman" or "1 subject" to avoid having multiple subjects in the image at 1024x1024) This model works for all use cases. Use this negatives or u'll never obtain the same results. <a target="_blank" rel="ugc" href="https://sharetext.me/7uz1nox4w5">https://sharetext.me/7uz1nox4w5</a> Best sampling methods: DPM++ 2S A DPM++ SDE DPM++ SDE KARRAS PLMS LMS Consider trying it also for other things, not only latex, works really well for img2img or whatever photorealistic image

Files with SHA256: 5d54f55d5a46f6dfe43f2fb4eb90eb1fe98d7a1f8d59893ba1f500fe5cf19507