The terms every AI video creator should know

Aspect ratio is the shape of a frame, the relationship between its width and height. It decides whether a shot reads as cinematic widescreen, a vertical social clip, or a square, and it should be set before you generate.

Camera motion is the movement of the virtual camera within an AI-generated shot, such as a dolly in, a pan, or a crane. Naming the move like a director briefs a crew gives a shot intent instead of random drift.

Character consistency is keeping the same face, wardrobe, and identity across every AI-generated shot instead of the character drifting scene to scene. It is the difference between a film and a pile of unrelated clips.

A cliffhanger ends an episode on an unresolved question, a reveal or a threat, so the viewer has to watch the next episode to find out what happens. It is the closing beat of every micro-drama episode.

A cold open drops the viewer straight into a moment of tension before any title or setup. In a vertical micro-drama it is the first two seconds of the episode: the shot that has to stop the scroll.

First-frame / last-frame video generation lets you pin the opening and closing frame of a shot, so the model fills in the motion between two images you control. It is the mechanism behind seamless AI transitions.

Frame chaining feeds the last frame of one AI video shot in as the first frame of the next, so motion and light flow continuously across shots with no still-image jumps or crossfades.

Image-to-image (i2i) transforms an existing picture with a prompt instead of starting from a blank frame. It restyles, edits, or extends a reference while keeping its composition, making it the tool for controlled revisions.

What is i2v? Image-to-video (i2v) animates a still image into a moving shot, using the frame as the first frame so motion inherits the composition you already approved. It is how a locked still becomes a cinematic clip.

A location plate is a reference still of a set, usually empty, reused as the reference for every shot in that place so the room, its layout and its light do not change between shots or episodes.

A LoRA is a small trained add-on that teaches a base model a specific subject or style, such as one character or one visual look, without retraining the whole model. It is a lightweight way to specialize outputs.

A lower third is an on-screen name or title card placed in the bottom third of the frame to identify a person, a place or a time. In the FXI workflow it goes on only after the color grade.

A micro-drama (microdrama) is a short vertical drama told in one to three minute episodes that each end on a cliffhanger, made for phones. The definition, the format rules, and how to make a micro-drama series with AI.

A negative prompt tells the model what to leave out of a generation. Where the main prompt describes what you want, the negative prompt names what you do not, from extra fingers to a specific color or style.

A seed is the number that sets the random starting point of an AI generation. What the seed does: reuse it with the same prompt and you get the same result, change it and you get a new variation. It is how outputs become repeatable instead of random.

A series bible is the reference document that keeps a show consistent: the premise, the recurring characters, the locations, the props and the rules of the story world, held identical across every episode.

Text-to-image (t2i) generates a still image from a written prompt alone, with no reference picture. It is the starting point for concept art, establishing frames, and reference stills in an AI film workflow.

Upscaling increases the resolution of an image or video while reconstructing detail, so a frame that looked soft at generation size holds up on a large screen or in an export. It is the finishing step before delivery.