Glossary

Plain-language definitions of the AI image and video terms you meet in tendre.AI: LoRA, seed, inpainting, SDXL, credits and more.

LoRA

A small add-on file that teaches an image model a specific face, style or subject without retraining the whole model.

SDXL

An open-weights image generation model known for sharp detail and strong prompt following, widely used as a local base model.

Prompt

The text description you write to tell the AI model what image or video to generate.

Negative prompt

A list of words describing what should be excluded from the generated image, used alongside the main prompt.

Seed

The starting number that determines the random noise a model uses, so the same seed and settings reproduce the same image.

CFG

A setting that controls how strictly the model follows your prompt versus how much creative freedom it takes.

Sampler

The algorithm that guides how a generated image is refined step by step from random noise into a final picture.

Steps

The number of refinement passes a model runs to turn random noise into a finished image, affecting detail and speed.

Checkpoint

A saved, ready-to-use version of an image model's trained weights, used as the starting point for generation.

VAE

A model component that converts between compact latent representations and final full-resolution pixel images.

Inpainting

Editing technique that regenerates only a masked area of an image while leaving the rest untouched.

Outpainting

Technique that extends an image beyond its original borders by generating new, matching content around it.

img2img

Generation mode that starts from an existing image instead of a blank canvas to produce a new, related one.

text2img

Generation mode that creates a brand-new image directly from a written text prompt.

Upscaling

Process of increasing an image's resolution while adding realistic detail instead of just stretching pixels.

Latent space

Compressed mathematical representation an AI model uses internally to describe and generate images efficiently.

LTX-2.3

AI video generation model that turns a text prompt or image into a short video clip with synchronized sound.

Foley

Added sound effects that match on-screen action, generated to make a silent video clip feel alive.

Cloud credits

Prepaid balance that pays for generations run on remote GPUs through tendre.AI's optional cloud mode.