AI Image Generator Glossary: 50 Terms Explained
Understanding AI Image Generation
AI image generation has its own vocabulary that can be confusing for newcomers. This glossary covers fifty essential terms in plain English, organized from basic concepts to advanced techniques. You do not need to memorize all of these to use AI generators effectively. Tools like ZSky AI handle the technical complexity behind the scenes. But understanding these terms helps you communicate with other AI art creators and make better use of advanced features when you need them.
Basic Terms
1. Prompt
The text description you type to tell the AI what image to create. Your prompt is the primary way you communicate your creative vision to the generator.
2. Generation
The process of creating an image from a prompt. Each time the AI produces an image, that is one generation.
3. Text-to-Image (T2I)
Creating an image from a text description. The core function of AI image generators.
4. Image-to-Image (I2I)
Using an existing image as a starting point and modifying it based on text instructions.
5. Text-to-Video (T2V)
Creating a video clip from a text description. Available in tools like ZSky AI's video generator.
6. Image-to-Video (I2V)
Animating a still image into a video clip. Upload a photo and the AI creates motion from it.
7. Aspect Ratio
The proportional relationship between width and height. Common ratios include 1:1 (square), 16:9 (widescreen), and 9:16 (vertical/mobile).
8. Resolution
The number of pixels in the generated image, expressed as width by height. Higher resolution means more detail and larger file size.
9. Seed
A random number that determines the specific starting point for generation. Using the same seed with the same prompt produces the same image, which is useful for reproducibility.
10. Negative Prompt
Text describing what you do not want in the image. Helps exclude unwanted elements, artifacts, or styles.
Model and Architecture Terms
11. Model
The trained AI system that generates images. Different models produce different visual styles and capabilities.
12. Inference
The process of running a trained model to produce an output. When you generate an image, the model performs inference.
13. Fine-Tuning
Additional training on a base model to specialize it for specific styles or subjects.
14. Checkpoint
A saved state of a trained model. Different checkpoints produce different visual results.
15. Latent Space
A compressed mathematical representation of images that the AI works with internally during generation.
16. Sampling
The process of progressively refining random noise into a coherent image during generation.
17. Steps
The number of refinement iterations during sampling. More steps generally produce more detailed results but take longer.
18. CFG Scale (Classifier-Free Guidance)
Controls how closely the generated image follows your prompt. Higher values stick closer to the prompt but may look less natural. Lower values are more creative but may drift from your description.
19. Scheduler/Sampler
The algorithm that controls how noise is removed during the generation process. Different schedulers produce slightly different visual characteristics.
20. Upscaling
Increasing the resolution of a generated image while adding detail. AI upscalers can enlarge images two to four times while maintaining or improving quality.
Creative Terms
21. Style Transfer
Applying the visual style of one image to the content of another.
22. Inpainting
Selectively regenerating a specific area of an image while keeping the rest unchanged.
23. Outpainting
Extending an image beyond its original boundaries, generating new content that seamlessly continues the scene.
24. ControlNet
A technique that uses reference images (like edge maps or pose skeletons) to guide the composition of generated images.
25. Depth Map
A grayscale image representing the distance of objects from the camera, used to guide spatial composition in generation.
Skip the Jargon, Start Creating
You do not need to know these terms to make amazing AI art. Just describe what you want and ZSky AI handles the rest.
Start Creating Free →Advanced Terms
26. Batch Size
The number of images generated simultaneously from a single prompt.
27. Denoising Strength
In image-to-image, controls how much the original image is altered. Higher values change more.
28. Embedding
A trained concept encoded as a small file that can be used in prompts to reproduce specific styles or subjects.
29. Hallucination
When the AI generates incorrect or impossible details, like extra fingers on hands or nonsensical text.
30. Token
A unit of text that the AI processes. Prompts are broken into tokens, and most models have a maximum token limit.
31. Clip Skip
Skipping layers of the text encoder to change how prompts are interpreted. Can affect artistic style.
32. VAE (Variational Autoencoder)
A component that encodes and decodes images between pixel space and latent space.
33. Attention
The mechanism that determines how much weight different parts of your prompt receive during generation.
34. Prompt Weighting
Assigning different importance levels to different parts of your prompt, usually using parentheses or numeric values.
35. Composition
The arrangement of visual elements within an image. Good composition guides the viewer's eye and creates visual interest.
36. Bokeh
The aesthetic quality of out-of-focus areas in an image. Often used in portrait prompts for blurred backgrounds.
37. HDR (High Dynamic Range)
Extended brightness and color range in an image. Including HDR in prompts can produce more vivid, detailed results.
38. Photorealistic
Images that closely resemble real photographs. A common style keyword in AI generation prompts.
39. Render
A 3D-generated image. Including render in prompts can produce images with a 3D-rendered look.
40. Artifact
An unwanted visual flaw in a generated image, such as blurry patches, color bleeding, or distorted features.
Business and Platform Terms
41. Credits/Tokens (billing)
Units of generation capacity. Many AI tools limit usage through credit or token systems.
42. Free Tier
A subscription level that costs nothing. Quality and limitations vary significantly between platforms.
43. Watermark
A visible mark added to generated images, often on free tiers, to identify the source tool.
44. API (Application Programming Interface)
A programming interface that allows developers to integrate AI generation into their own applications.
45. Commercial License
Permission to use generated images for business and profit-making purposes.
46. Content Policy
Rules governing what types of content can be generated. Varies by platform.
47. Stealth Mode
A privacy feature (notably on Midjourney) that hides your generations from public view.
48. Queue
A waiting line for generation during high-traffic periods. Free tiers often have longer queues.
49. Fast Mode vs Relaxed Mode
Priority levels for generation speed. Fast mode generates immediately, relaxed mode may queue.
50. Generation History
A log of your previous generations and prompts, stored by the platform for reference.
For more learning resources, see our prompt writing guide, our how AI generation works explainer, and our best AI for beginners guide.
Frequently Asked Questions
What is a prompt in AI image generation?
A prompt is the text description you provide to an AI image generator to tell it what to create. It can be as simple as a cat or as detailed as a Persian cat sitting on a velvet cushion in a Victorian library, oil painting style, warm candlelight. Better prompts produce better images.
What does text-to-image mean?
Text-to-image is the process of generating a visual image from a written text description. You type words describing what you want to see, and the AI creates a corresponding image. This is the core function of tools like ZSky AI, Midjourney, and DALL-E.
What is image-to-image generation?
Image-to-image generation takes an existing image as input and transforms it based on additional text instructions. You upload a photo and describe how you want it changed, and the AI creates a modified version. This is useful for style transfer, background replacement, and image enhancement.
What is a negative prompt?
A negative prompt tells the AI what you do not want in the image. If you want a portrait without glasses, you would include glasses in the negative prompt. Negative prompts help refine results by excluding unwanted elements, artifacts, or styles.
Do I need to understand these terms to use AI image generators?
No. Modern AI generators like ZSky AI are designed to work with plain English descriptions. You do not need to understand technical terminology to create great images. However, knowing these terms can help you fine-tune results and understand what is happening behind the scenes.
Start Creating, Not Studying
You can learn the jargon later. Right now, just type what you want to see and let ZSky AI do the rest.
Generate Your First Image →