cinemasetfreeWhat a LoRA is, and when to use one
A LoRA teaches an image model one specific thing. Use one when you need the same face or style in many pictures. For a single picture, a better description is quicker.
What it adds
An image model knows a lot about the world and nothing about your film. A LoRA is a small trained file that adds one thing, such as this actor’s face, this artist’s line or these uniforms.
You add it to a production design, together with an image model that accepts LoRAs. From then on, it applies to every picture that design makes with that model: characters, wardrobe, props, locations, scenes and storyboard panels.
Where you find it
The LoRA Integrations button is in the design editor. Open the Designs tab, press Edit on a design and go to the Image tab. The button only shows when the chosen Image Model can take a LoRA.
To find a model that takes one, type “lora” in the Image Model box. The model catalog also has a LoRA filter.
When a LoRA helps
The same face every time
A character LoRA keeps one person looking like themselves across dozens of pictures. A written description drifts from picture to picture.
A specific look
One artist’s or photographer’s style, when “in the style of” in a prompt only gets you close.
Something the model lacks
A period, a piece of gear, or a look the model never saw.
Not for one picture
Finding and tuning a LoRA takes longer than rewriting a prompt. Save it for things that repeat.
Try these first
A sharper description on the character or prop often fixes a generic result. A mood board sends real pictures along with each generation, which is often enough for a look you can show but not put into words.
Reach for a LoRA when both of those fall short and you need the result again and again.
Still stuck?Ask in the Discord