OD glossary for beginners
Use these definitions while following the tutorials. A field is a labelled box or selector; a control is a button, menu, switch, or slider you can operate.
| Term | Meaning |
|---|---|
| AI | Artificial intelligence; here, software that generates replies or media from available input. |
| APD | Additional Personality Details. A character field for motives, behavioral traits, and written voice. |
| Art style | The picture's visual treatment, such as watercolor or photographic lighting. It does not define the person's identity. |
| Auto Memory Log | A native summary of conversation history. Inspect what it actually retained. |
| Backstory | Events before the current scene that help explain a character. |
| Base image | An accepted image used as the starting point for edits. |
| Canon | Facts established as true for this particular story. |
| CFD | Custom Face Details, where offered. A face-focused appearance field. |
| Character | A fictional person or creature represented in the chat or image. |
| Character card | A portable character definition, commonly stored in JSON or embedded metadata in a PNG. |
| Character sheet | A worksheet that organizes identity, behavior, relationships, and current state. It is not automatically a native import. |
| Chat | A conversation with one or more AI respondents. |
| Chat model | The model generating written replies. Its selector is separate from the image-model selector. |
| Checkpoint | A record of the facts needed to continue a particular accepted story state. |
| Comic panel | One picture or frame within a comic page. |
| Context window | The input capacity available to a model for a reply. A saved transcript and the context supplied to that reply are different; no universal OD token allowance is established here. |
| CPD | Custom Physical Details. A field for stable visible appearance. |
| Custom Instructions | Persistent requests about how replies should be written or constrained. |
| Display media | Images or other accepted media chosen to present a character or group. |
| ED | Extra Details. A field for relevant lore, rules, supporting cast, and systems. |
| Engine | In these tutorials, a set of roleplay instructions. The word does not mean installed software. |
| Evidence | An observed output or inspected source that supports a specific claim. A planned example is not a test result. |
| Event ID | A label distinguishing one consequential event from another, useful for avoiding duplicate rewards. |
| EXP | Experience points, an optional game resource whose earning rules must be defined. |
| Footer | A status block placed after a reply. It is text unless connected to an actual supported feature. |
| Free Play | A generation route without relying on a selected saved character, where offered. Supply the needed identity facts. |
| Gallery | The area holding generated or selected media. |
| Generation | Producing a new AI reply, picture, or other media from the available input. |
| Greeting | The character's opening message. |
| HUD | Heads-up display; in a text game, usually a visible status panel. |
| Image model | The model making or editing pictures. It need not match the chat model. |
| Import | Reading a supported external character-card file into the product's creation flow. |
| Inconclusive | A test with insufficient evidence to judge the chosen criterion; different from a pass, failure, or unrun test. |
| JSON | A structured data format with named fields, strings, numbers, lists, and objects. It does not run game logic by itself. |
| Lore | Background facts about the fictional world. |
| Lorebook | A collection of reference entries used by some character-card systems. Importing its data does not establish runtime trigger behavior. |
| LLM | Large language model. Here, the model generating written chat replies; an OD selector label does not by itself identify its underlying provider. |
| Metadata | Information stored with a file, such as the character text embedded in a card image. |
| Native feature | A feature provided through OD's actual controls, such as a State Tracker; not a local-tool panel or a prompt that resembles one. |
| Negative Prompt | A separate field, where offered, for undesired visual elements. |
| NPC | Non-player character: someone portrayed by the narrator or AI rather than controlled by the human player. |
| Persona | The identity you choose to play in a chat. |
| Personality | The character's motives, habits, traits, and ways of responding. |
| Pinned Memories | Selected important facts maintained for a continuing conversation. |
| PNG | An image-file format that can also contain card metadata. Not every PNG is a character card. |
| Preset | A reusable selection, such as an outfit, pose, or background. |
| Private | A visibility choice for personal use, not secure storage for real secrets. Read the actual control's description. |
| Prompt | Text asking the AI to produce something. |
| Prompt revision | A saved change to wording. Compare it with the previous version under matched settings and available context. |
| Public | A visibility or publication choice intended for community discovery, subject to the site's process. |
| Regenerate | Request an alternative version of a response or output. Check which version is retained. |
| Rendering | How a picture is drawn, painted, shaded, or photographed. |
| Roleplay | Participating in a fictional interaction through dialogue, actions, and consequences. |
| RPG | Role-playing game: roleplay with explicit game rules and outcomes. |
| Scenario | The opening situation, roles, setting, and immediate problem. |
| Scene prompt | Instructions for one image's outfit, action, framing, setting, and light. |
| SFW | Safe for work; used here for ordinary non-explicit examples and editing tasks. |
| State | Facts that can change, such as location, inventory, or a completed-event counter. |
| State Tracker | A native configurable state feature where available. Its actual updates still need inspection. |
| Stitching | Combining selected video clips into a sequence. |
| System prompt | A general term for application-supplied instructions. Writing SYSTEM in an OD field does not establish its actual message role or priority. Use the documented field jobs. |
| Timeline | The accepted sequence of story events, including which edited or regenerated version you kept. |
| Training | Learning or changing a model's parameters. These tutorial exercises teach the human author to write prompts; pasting a module into a field is not a model-training operation. |
| Token | A unit used by a language model. It is not interchangeable with a character or byte count. |
| UI | User interface: the site's visible pages, fields, and controls. |
| Unlisted | A sharing option, where offered, that differs from public catalog discovery. Check the site's description before sharing. |
| Voice | Can mean a character's writing style or the selected audio sound. The tutorials identify which meaning applies. |
| Voice sample | A brief example of a character's wording and tone, supplied as a reference rather than a line to repeat. |
| Worldbuilding | Defining the places, people, institutions, and rules of the fictional setting. |
Terminology follow-up: the company-hosted companion glossary, updated August 4, 2026 and retrieved with Firecrawl on October 31, distinguishes tokens, context, character definitions, and persistent memory. It describes vocabulary across platforms; its architectural generalizations do not establish OD's exact backend, retrieval method, or input limits. For OD's operational field map, use the memory guide and instruction guide.