Knowledge baseSettings
Settings
Every editor setting on its own: what it means, when to switch it on and how it affects the price.
21 articles24 min read
01Aspect ratios: which one to pick1 min readAspect ratio is the shape of the frame: width relative to height.02Image resolution: 1K, 2K, 4K2 min readResolution is how many dots the image has. More dots mean a bigger file and more visible detail — and a higher price.03Video resolution1 min readResolution is the number of pixels in the frame. 720p is the standard for quick drafts and social feeds; 1080p gives a sharp picture for final videos and big screens.04Video duration and price1 min readVideo models are billed per second of output. A 10-second video costs twice as much as a 5-second one.05Quality: basic, high, ultra1 min readThe level of finish. The higher it is, the more the model works on details: skin, fabric, reflections, small text.06Prompt adherence2 min readA slider from 0 to 1: how strictly the model follows your text.07Edit strength1 min readA slider from 0 to 1: how much to rework the source photo. 0 keeps it almost intact, 1 redraws it from scratch.08Prompt enhancement2 min readYou write "a cat on a windowsill". The model expands it itself: "a ginger cat sits on a wooden windowsill, soft morning light, blurred background, detailed fur". The image is drawn from that expanded text.09File format: PNG or JPEG1 min readThis is the file you get. The content is the same — the packaging differs.10Safe mode1 min readAn extra guard: before handing over the result, the model checks the image for disallowed content. If something is off, the generation is not delivered.11Fixed camera1 min readNormally the model moves the camera on its own: pushing in, orbiting, drifting. Turn on the fixed camera and the shot behaves like it is on a tripod: the camera stays put, only the subjects move.12Sound in video1 min readModels usually output silent video. Turn sound on and a soundtrack is generated together with the picture: footsteps, wind, street noise, ambient background.13Video references1 min readShort example clips. The model watches them and repeats the movement, rhythm and shooting style — not the content.14Audio references1 min readShort example sound files. The model listens and makes the video's sound similar in character: same atmosphere, same kind of noise, same musical mood.15Tail image1 min readNormally you give the model one photo and the clip starts from it. The tail image is a second picture: what the clip should end on. The model invents the transition between them.16Upscale factor1 min readHow many times bigger to make the image. If it was 1000 px wide and you pick 4, it becomes 4000.17Style (Ideogram)1 min readA hint about which direction to draw in: like a photo, like a poster, or whatever the model prefers.18Rendering speed (Ideogram)1 min readA choice between fast and pretty. The price does not change — the time and the level of detail do.19Web search1 min readNormally a model only knows what it learned during training. With this checkbox it looks things up on the web before generating.20Mood (Grok)1 min readSets how seriously or freely Grok interprets your prompt.21Pro mode1 min readA "make it better" checkbox: it enables an improved generation path. The price stays the same.
Other sections