Home › Glossary

AI Photo & Video Glossary

Plain-language definitions of the terms AI photo and video tools actually use — each with a concrete example and where it shows up in Hermosso.

• AI headshot

  An AI headshot is a professional-style portrait generated by an AI model trained on your selfies, instead of shot by a photographer. The likeness is yours; the studio, lighting and wardrobe are synthetic.

• AI model (vs engine)

  An AI model is the trained system that generates images or video; an engine is one runnable configuration of that model. One model can appear as several engines — for example a 1K and a 2K variant.

• AI persona (personal model)

  An AI persona is a private AI model trained on your own selfies so it can generate new photos of you in any style, setting or outfit. It is also called a personal model or an AI twin.

• Aspect ratio

  Aspect ratio is the width-to-height shape of an image or video — 16:9 widescreen, 1:1 square, 9:16 vertical. The right ratio is the one the platform it ships to is built around.

• Camera move

  A camera move is a deliberate motion of the virtual camera in a generated clip — dolly-in, orbit, whip pan, dolly zoom. Naming the move in the prompt is the strongest way to steer AI video.

• ControlNet

  ControlNet is a technique that forces an AI image model to follow the structure of a reference — edges, depth or pose — while changing style, materials or content. The geometry stays; everything else can move.

• Diffusion model

  A diffusion model is the AI architecture behind most modern image generators. It starts from pure random noise and removes it step by step until a picture matching your prompt remains.

• Image-to-image (img2img)

  Image-to-image (img2img) is generating a new picture from an existing photo plus instructions, instead of starting from text alone. The input photo anchors composition, pose or identity while the model changes the rest.

• Image-to-video

  Image-to-video is animating a still photo into a moving clip. The photo sets the subject and look; a prompt or camera move sets the motion.

• Inpainting

  Inpainting is regenerating a masked region of an image while leaving everything outside the mask untouched. You paint over what you want changed, describe the replacement, and the model fills only that area.

• Lip sync

  AI lip sync is re-animating a face's mouth so it matches a spoken audio track. The face stays the same person; only the speech motion is generated.

• LoRA

  LoRA (Low-Rank Adaptation) is a lightweight training method that teaches an existing AI image model one new thing — such as a specific person's face — without retraining the whole model. The result is a small add-on that steers every later generation.

• Negative prompt

  A negative prompt is text that tells an AI image model what to leave out — blur, extra fingers, watermarks, text. While generating, the model actively steers away from the patterns you list.

• Outpainting

  Outpainting is extending an image past its original borders — the model invents new surroundings that continue the scene convincingly. It is how a tight crop becomes a wide shot.

• Prompt

  A prompt is the written instruction you give an AI model describing the image or video you want. It is the main control you have: subject, style, lighting, composition and mood all come from the wording.

• Relighting

  Relighting is changing the light in a finished photo — direction, warmth, time of day — while the subject and identity stay intact. The AI re-renders the illumination, not the person.

• Resolution (4K, 8K, 22K)

  Resolution is the pixel count of an image or video, usually named by its long edge — 4K is roughly 4,000 pixels across. More resolution means more printable detail, at the cost of file size and compute.

• Seed

  A seed is the number that initialises an AI model's randomness. The same prompt with the same seed produces the same image; change the seed and you get a different variation of the same idea.

• Style transfer

  Style transfer is applying the visual language of one image — palette, grain, brushwork, mood — to another, while keeping the second image's content and composition intact.

• Talking photo

  A talking photo is a still portrait animated to speak — lips synced to a script or recording, with natural head movement and expression. One photo becomes a short presenter video.

• Text-to-image

  Text-to-image is generating a picture from a written description alone, with no input photo. You type what you want and the model creates it from scratch.

• Text-to-video

  Text-to-video is generating a moving clip from a written description alone, with no input footage. The model invents the scene and the motion — and on newer engines, the sound.

• Upscaling (super-resolution)

  Upscaling, or super-resolution, is increasing an image's pixel dimensions with AI that reconstructs plausible detail instead of stretching. A small or soft photo becomes large and sharp enough to print.

• Virtual staging

  Virtual staging is furnishing an empty room photo with AI — sofas, beds, lighting and decor rendered in while the architecture stays untouched. It replaces physical staging for listings at a fraction of the cost.

• Virtual try-on

  Virtual try-on is previewing a hairstyle, outfit or product on your own photo with AI, so you see how it looks on you before buying, booking or committing.