AI Art Glossary
Master the terminology of AI art generation. From diffusion models to CFG scales, understand every concept.
A
Aspect Ratio Parameters
The proportional relationship between image width and height. Different ratios suit different uses, and AI models may perform better at certain ratios.
AUTOMATIC1111 Tools
A popular web-based user interface for Stable Diffusion with extensive features and extension support. Also called A1111 or SD WebUI.
C
CFG Scale Parameters
Classifier-Free Guidance Scale - a parameter that controls how closely the AI follows your text prompt versus generating more freely. Higher values mean stricter adherence to the prompt.
Character Consistency Concepts
Techniques for generating the same character with consistent appearance across multiple images. A challenging problem in AI art with several solution approaches.
Checkpoint Models
A saved state of a trained AI model containing all its learned weights. In Stable Diffusion, checkpoints are the main model files that determine the overall style and capabilities.
Civitai Tools
The largest community platform for sharing AI art models, LoRAs, embeddings, and generated images. A hub for Stable Diffusion resources and inspiration.
CLIP Models
Contrastive Language-Image Pre-training - an AI model that understands relationships between text and images. Used to guide image generation based on text prompts.
ComfyUI Tools
A powerful node-based user interface for Stable Diffusion and Flux that allows visual workflow creation. Offers maximum flexibility through connecting processing nodes.
ControlNet Techniques
A neural network architecture that adds conditional control to diffusion models, allowing precise guidance through reference images for pose, depth, edges, and other visual features.
D
Denoising Strength Parameters
A parameter controlling how much an input image is modified during img2img or inpainting. Higher values allow more change from the original.
Depth Map Techniques
A grayscale image representing distance from the camera, where brightness indicates depth. Used with ControlNet to maintain spatial structure during generation.
Diffusion Model Models
A type of generative AI model that creates images by learning to reverse a gradual noising process, transforming random noise into coherent images guided by text prompts.
Dreambooth Techniques
A fine-tuning technique for training AI models to learn specific subjects, people, or objects from a small set of reference images. Creates highly accurate subject representation.
I
Image-to-Video Techniques
AI technique that animates a still image into a video clip with motion and dynamics.
Img2Img Techniques
Image-to-image generation - using an existing image as a starting point for AI generation. The output is influenced by both the input image and the text prompt.
Inpainting Techniques
A technique for selectively regenerating parts of an image by masking areas to be replaced while keeping the rest intact. Used for fixing errors, adding elements, or modifying specific regions.
IP-Adapter Techniques
Image Prompt Adapter - a method for using reference images to guide generation style, composition, or subject features without fine-tuning the model.
L
Latent Space Models
A compressed mathematical representation where AI models process images more efficiently than working with raw pixels. Each point in latent space corresponds to a possible image.
LoRA Techniques
Low-Rank Adaptation - a technique for fine-tuning AI models efficiently by training only a small number of additional parameters, allowing customization without retraining the entire model.
LyCORIS Techniques
An advanced alternative to LoRA that offers different fine-tuning methods like LoCon, LoHa, and LoKr. Can capture more complex styles and subjects with various precision/size tradeoffs.
S
Safetensors Concepts
A secure file format for storing AI model weights. Unlike .ckpt files, safetensors cannot contain malicious code, making them the recommended format for sharing models.
Sampler Parameters
The algorithm used to progressively denoise an image during generation. Different samplers produce varying results in quality, style, and speed.
Sampling Steps Parameters
The number of denoising iterations during image generation. More steps generally produce higher quality but take longer to generate.
Scheduler Parameters
An algorithm that controls how noise is added and removed across sampling steps. Different schedulers produce different quality characteristics at various step counts.
Seed Parameters
A number that initializes the random noise pattern for image generation. Using the same seed with identical settings produces the same image.
Style Transfer Techniques
The technique of applying the visual style of one image to another, making the content of image A look like it was rendered in the style of image B.
T
Text-to-Video Techniques
AI technology that generates video clips from written text descriptions (prompts).
Textual Inversion Techniques
A training technique that creates a new embedding (token) representing a specific concept, style, or object. The embedding can be invoked in prompts with a trigger word.
Token Concepts
The basic unit of text that AI models process. Prompts are split into tokens (roughly 0.75 words each), and models have a maximum token limit.
Transformer Models
A neural network architecture based on self-attention mechanisms. Used in modern AI models including CLIP, and increasingly replacing U-Net in newer diffusion models like SD 3.
U
U-Net Models
The core neural network architecture in diffusion models that predicts and removes noise during image generation. Named for its U-shaped structure with encoder and decoder paths.
Upscaling Techniques
The process of increasing an image's resolution while maintaining or enhancing quality. AI upscalers can add realistic detail that wasn't in the original.