Is the image generator free?

Yes, within your free credits. Every account starts with credits that work across image, video, and audio generation, with no card required. Paid plans add more credits, higher resolutions, and watermark-free exports.

Which model should I use?

It depends on the job: photographic realism, illustration, legible text, and product accuracy are different strengths. The composer keeps your prompt and references when you switch, so comparing two on the same brief is the fastest way to decide.

Can I keep the same person across several images?

Yes. Save a character from reference photos and mention it in each prompt. Describing the same person in words produces a different person every run.

What resolution can I get?

Generate at the model's native size and upscale afterwards — that reliably beats asking a generator for an unusual large size directly. The upscaler takes results to print size.

Can I use the images commercially?

Yes, under the terms of service, on a paid plan. You remain responsible for what goes into a prompt — in particular for not generating a real person's likeness without their agreement.

Can I turn an image into a video?

Yes. Any image in your library can be used as the starting frame of a video generation, or as both the first and last frame if you want to control where the clip ends.

AI Image Generator

Write a prompt or upload a picture, choose a model, and get an image you can keep working on — edit it, enlarge it, relight it, or turn it into video without leaving the library it landed in.

Text to image and image to image
Several models, one composer
Characters that stay the same person
Upscaling, relighting, and video in one place

Generate an image Browse templates

What you get on the free tier

Sign-up credits: Free
Image models: Several
Max output: 4x upscaled
Watermark: None on paid

How to generate an image

The same three steps whether you are starting from a sentence or from a photograph. The second one is where most of the quality difference comes from.

01
Start from words or from a picture
A prompt when nothing exists yet; an upload when the subject is real and has to stay recognisable. Those are genuinely different jobs, and picking the wrong one is the most common reason a result disappoints.
02
Choose the model and the shape
Models specialise: photographic realism, illustration, legible text in the frame, faithful product geometry. Set the aspect ratio you will publish in before generating, so you are not cropping the composition away afterwards.
03
Generate, compare, finish
Runs land in your library so you can judge several side by side, then upscale, relight, or animate the one that worked without exporting it anywhere.

What this studio adds over a single prompt box

Model switching without rewrites

Your prompt, references, and settings carry when you change model, so running the same idea through two models is a click rather than a rebuild.

Characters that stay one person

Save a character from reference photos and mention it in every prompt. Words alone give you a different face on every run, however careful the description.

References, not just adjectives

Attach an image for style, composition, or a product's real shape — the difference between asking for 'our packaging' and showing it.

Editing as a first-class mode

Background replacement, restyling, cleanup, and frame extension all work on a picture you already have, keeping the parts that were already right.

Finishing tools in the same place

Upscale to print size, relight a shot that was lit wrong, or reframe — without exporting to another tool and re-uploading.

One step to motion

Any image in your library can become the first frame of a video, which is the shortest route from a still idea to a moving one.

What people make here

Product imagery

One real photograph of the product, then every background, season, and format a campaign needs — with the product still itself.

Social creative

Posts, stories, and thumbnails generated in the ratio each surface wants, tested as variants rather than designed once.

Concept and mood

Look development for a campaign or a scene, explored in a dozen directions before anything is commissioned.

Editorial illustration

Article and deck imagery in a consistent house style, generated instead of licensed.

Headshots and team photos

A single background and lighting treatment applied across photos taken in five different rooms.

First frames for video

Stills generated deliberately, so the clip that grows out of one starts on the right composition.

Prompt or upload — which input to use

The input decides more than the model does. This is the short version of when each one wins.

	Text to image	Image to image	Template
Best for	Subjects that do not exist yet	A real product, place, or person	A look you have seen and want to repeat
What you supply	A written prompt	A picture plus an edit note	A photo, usually of a face
How much survives	Nothing — it is invented	As much as you choose, by strength	The template's composition
Consistency across a set	Save a character or reference	Reuse the same source picture	Built into the template
Time to a usable result	A few iterations	Usually one or two runs	Usually one run

What an AI image generator is good at, and where it still is not

Image models have become genuinely reliable at a narrow, valuable set of things: a single well-composed frame, a described subject in described light, and a style held consistently across a run. Where they remain unreliable is anything requiring exactness — text that has to read correctly, a logo that has to be the logo, a product whose proportions customers would notice being wrong, a specific person's face. The practical response is not to prompt harder but to change the input: start from a photograph when the subject is real, and add exact text in a layout tool afterwards.

Consistency is a setup problem, not a prompting problem. Teams that get a coherent set of images are not writing better descriptions each time; they have saved the character, the reference, and the prompt that worked, and they re-run that combination with the subject swapped. The moment you find yourself describing the same face in words for the fourth time, the answer is a saved character.

Model choice does more work than another round of adjectives. Photographic realism, clean illustration, legible in-frame text, and faithful product geometry are four separate strengths and no model leads on all of them. Because the composer carries your prompt and references across a model switch, running the same brief through two models costs a click and settles the question with evidence rather than opinion.

It is also worth being clear about what the finishing tools do. An upscaler generates a larger picture consistent with the small one rather than recovering information that was never captured, which makes it excellent for texture and unsuitable for anything evidential. Relighting changes how a scene appears to be lit rather than physically relighting the objects in it. Both are enormously useful when you know which of those two things you are getting.

Finally, the boundary between still and moving image has effectively disappeared. Generating a still deliberately, approving the composition, and only then animating it produces better video than prompting for a clip and hoping the opening frame is right — which is why this hub and the video hub are two doors into the same library rather than two products.