Textual Inversion vs Lora
I've recently learned how to make Loras. I feel like they are easier to make and less can go wrong due to human error. Previously I had been making embeddings because that's all I knew at the time.
Is there any reason I should switch back to Textual Inversions for certain things?
Perhaps better question is, why do you like using one over the other for your prompts?
LoRAs train "harder", they have more weight than TI, so they can easily overfit. LoRAs are better to get the exact thing you want, whereas TI are better for general purposes. For example to train for a character or art style you definitely wanna use LoRA, you want that "overfit" benefit, but for let's say a certain hairstyle or clothing or theme, TI is better as the "overfit" benefit doesn't really apply :)