Help·What are LoRAs?
What are LoRAs?
LoRAs explained
Small add-ons, big creative impact
LoRA stands for Low-Rank Adaptation. In simple terms, a LoRA is a small, lightweight add-on that modifies how a model generates images — steering it towards a particular style, subject, or aesthetic without replacing the whole model.
Think of a model as a general-purpose artist and a LoRA as a specialised technique they've learned. The artist is still the same, but the LoRA teaches them a new trick — maybe a particular art style, a character design approach, or a lighting technique.
How LoRAs work on Endlss
Select, combine, create
Choose your model first
LoRAs are compatible with specific models. When you select a model on the Generate page, only compatible LoRAs appear. WAN 2.1 14B supports the widest range of LoRAs.
Stack up to 3 LoRAs
You can combine up to three LoRAs per generation. This lets you layer styles — for example, a lighting LoRA with a texture LoRA and a colour palette LoRA.
Trigger words
Some LoRAs require a specific trigger word in your prompt to activate. When a LoRA has a trigger word, it's shown next to the LoRA name on the Generate page — just include it in your prompt.
Community vs Premium LoRAs
Access depends on your subscription tier
LoRAs on Endlss are split into two categories:
Community LoRAs
Available to all subscribers with AI generation access (Dali tier and above). These cover a broad range of popular styles and subjects.
Premium LoRAs
Exclusive to Van Gogh and Picasso tier subscribers. These are curated, higher-quality LoRAs that produce more refined and distinctive results.
What are AI models?
Understand the different models available and what each one is best at.
Generate an image from text
Step-by-step guide to creating your first AI image on Endlss.