Natalie Portman is a well known actress. There are many LoRA's and only a couple of embeddings. This is my attempt at creating a better quality TI.

This is a 120-step TI trained on a dataset of 15 images using JernauGurgeh's method of training embeddings. This was trained primarily using close-up images from the mid 2000s to late 2010's with a 5 instances of chest-up portraits, and 10 close-up portraits.

How to create a good prompt using this TI

JernauGurgeh recommends starting with this phrase specifically, but I also add "beautiful" in there for more consistent results with Jernaumix and other checkpoints in the sample photos. I highly recommend using ADetailer for anything outside of a close-up portrait.

"photo of beautiful n4talieportman, a woman"

If you have feedback about any limitations or suggestions to improve this, let me know. Please leave a review and some images if you find this embedding to be good quality!