Glossary
← Quick reference| Term | Meaning |
|---|---|
| Project | A body of work containing multiple shots. A project exists as long as it has at least one Shot — deleting the last Shot also deletes the project. |
| Automatic Project Naming | New projects start named Untitled (or Untitled_NN when taken). The first video uploaded into a still-default, empty project auto-renames it to the first 10 characters of the video’s filename with the extension stripped — e.g. Project_summer_ep01.mp4 → Project_su. Never overrides a user-chosen name, never re-triggers once the project has content, and is skipped during onboarding. |
| Shot | A single editable video segment within a project. The unit of rendering. Displayed as one entry in the shots panel and one row in the project timeline. |
| Render | An output produced from a source video inside a shot. Appears as a render track. |
| Track | A layer attached to a Shot. Two kinds: Image Track (the source video) and Render Track (a generated output). |
| Image Track | The original uploaded video. Always present, one per Shot. Also called “source track”. |
| Render track | A generated output, stacked below the source in the shot editor. A render track can be pinned. |
| Iterated track | A render produced using Edit this Render on a previous track. |
| Pin | Marking a render track as the representative render for a shot, shown in the Project Timeline. Only finished render tracks can be pinned. |
| Edit this Render | Action that uses a render as the input video for the next operation. |
| Reuse Settings | Action that loads a track’s settings into the Settings panel. |
| Project Timeline | The full sequence of all shots in a project, shown as a horizontal timeline with a playhead, ruler, and zoom controls. Each slot shows the pinned render (or source video if none is pinned). Also called “Global Timeline” in parts of the UI. |
| Video Model | An AI model that generates video (e.g. MagoV3, MagoV4, MagoV5, Seedance, Kling, GPT Image 2, Upscaler). |
| Credits mode | Pay-per-render mode. Each render consumes credits based on frame count, resolution, and the selected model, with unlimited concurrent renders. The alternative to Relaxed / Unlimited mode. |
| Relaxed mode | A rate-limited mode that doesn’t spend credits. Provides access to Mago video models and GPT 2 image model. Concurrent renders and maximum number of frames are capped unlike in credits mode. |
| Unlimited mode | Like Relaxed mode, but enables all the models and higher usage limits. |
| ControlNet | A conditioning signal (depth, pose, SoftEdge, Canny) that constrains the model’s output. |
| Context size | Frames per chunk in long renders. |
| Context overlap | Frames that overlap between adjacent chunks during chunked rendering. |
| Dynamic reference | Setting that regenerates the reference image between chunks. |
| Auto Prompt | Mago-generated prompt based on video and reference descriptions. |
| Descriptive prompt | A description of the desired result. Required by Mago models. |
| Instruction prompt | A directive to the model. Used by closed-source and image models. |
| Frame-perfect | Property of Mago models: N input frames produce N output frames, each corresponding to a specific source frame. |
| Img2Img (Image-to-Image) | The class of AI models used in the Modify Frame tab. Takes a source frame + prompt + settings and outputs an edited frame. Distinct from video-to-video rendering. Examples: GPT Image 2, Nano Banana Pro, Seedream 5.0 Pro — the full selectable list lives in the Tooltips reference. |
| Modify Frame | The UI tab where users apply Img2Img models to individual frames. “Modify Frame” is the product name; “Img2Img” is the technical model category. |
| Masking | A workflow that produces a black-and-white mask video using the SAM3 Mask model, to feed into Inpainting. |
| Inpainting | A video model (product label: Mago Inpaint) that edits only the region defined by a mask video. |
| Element / Element Pair | A reference object used by the element-swap Kling models (Kling O1 Pro, Kling 3.0 Motion Control, Kling O3 Pro). Supports two angles: a main reference image and an optional frontal image. |
| Drawn Mask | A red mask painted by hand with a brush on a single video frame inside Modify Frame, telling Seedream 5.0 Pro which region to edit. Available to all users on Seedream 5.0 Pro. Distinct from Masking (the SAM3 mask video used by Inpainting). GPT Image 2 keeps its separate “Mask image” upload field. |
| Reference Frame | An image input that acts as a style or content reference for a render. Order-independent — multiple reference frames can be added. |
| Key Frame | An image input anchored to a specific frame index in the output, guiding the model at that exact point. |
| Preset | Quick-start configuration that sets a model and baseline settings. |
| EXR | OpenEXR image format. Used in professional VFX pipelines for high dynamic range and lossless compositing. |
| Detail enhancement (Creative Upscaler) | Strength of reconstruction during upscale (formerly “Denoise”). Higher means more model freedom and more added detail. |
← Quick reference