<a target="_blank" rel="ugc" href="https://www.instagram.com/karajewelll/?theme=dark">Kara del Toro</a> is an American model and social media influencer.Updated! 134-step TI trained on a dataset of 15 images with<a target="_blank" rel="ugc" href="https://imgur.com/a/d348H3k"> these settings</a>.Curious about my work process? I have summarized it <a target="_blank" rel="ugc" href="https://ko-fi.com/Post/My-TI-Training-Process-E1E1V8KLC">here</a>.Do you have a specific idea for a TI in mind?<a target="_blank" rel="ugc" href="https://www.buymeacoffee.com/jernaugurgeh"> Visit my website </a>and let me know.<h2 id="heading-386">Building a good prompt with my TIs</h2>You're obviously free to experiment, but bear in mind that my TIs are trained with a more or less fixed phrasing, that normally starts with:"photo of EMBEDDING_NAME, a woman"So I recommend always starting your prompt like that and then building the rest of the prompt from there. For instance, "photo of beautiful (kdt0r0:0.99), a woman, perfect blonde hair, (closeup portrait), (modern photo, party strapless dress), 85mm, (analog, cinematic, film grain:1.3), (at a red carpet event), ((detailed eyes)), (color picture:1.1), (looking at viewer)".Please also note that I'm using the <a target="_blank" rel="ugc" href="https://civarchive.com/models/58390?modelVersionId=62833">"add detail" LoRA</a> for my example pics. I recommend setting it around 0.5 for best results. CLIP Skip should always be set at 1 when generating pics with this TI.

Kara del Toro

Now this is more like it. 500 steps, retrained from scratch with a pruned dataset (since the face was not as accurate as I wished, I removed some distant shots). Still some slight issues with the eyes, so I might still revisit again in the future, but overall I'm quite satisfied with this version.

Files with SHA256: 9b4b60d0a2eb9b9b3bf40172aaaffc3eaf7fe2657865337b1f34d7e43d138415