2.png)
<aside> 💡 At the time of writing, I do not use LoRAs normally when I generate, however it is a useful tool to up your image quality and prompt adherence if you are unable to prompt certain poses, specific characters/outfits or a style. I tested out 3 different LoRAs I found on civitai, and did a few tests on how much they impact the original prompt
</aside>
Here are a few that you can read if you want to learn more about the topic:
Using LoRA for Efficient Stable Diffusion Fine-Tuning
What is a LoRA and how do I use it effectively?
<aside> 📢 TLDR: Basically LoRAs allow for training of a model and makes the outputs be more specific to a target that you want to use. In many cases, characters or a certain style is involved in a LoRA.
</aside>
Here is a quick guide on how to use LoRAs, including downloading the models: Check out the page below to learn how to use them:
Website that is used as a source for the LoRAs Please note that there are some NSFW LoRAs, so please browse at your own caution Simply search a certain character and most likely there will be a LoRA for it
Civitai | Stable Diffusion models, embeddings, LoRAs and more
Modification includes adding the name of the LoRA under the prompt and the addition of trigger words to activate the LoRA. The changing of the strength of the LoRA results in the difference in influence the LoRA has in the final output.