Creative Controls are settings you pick instead of sentences you write. Choose the camera angle, the lighting, the pose, the composition and the style from a list — the way you would set up a real shot — then save that setup and load it again on the next product, the next campaign or a different AI model. This is the part that decides whether your second image matches your first.
AI images are normally made by writing a description and accepting whatever comes back. Controls change where the decision happens: the things a photographer would set on the day — lens, light, angle, pose — become settings you choose, keep and reuse, instead of adjectives you retype and re-interpret every run.
Camera angle. Key light. Pose. Instead of typing "low three-quarter angle, soft light from the left" into a prompt and hoping the model reads it your way, you pick those two things from a list. The model receives them as settings rather than as a sentence to interpret.
The controls active on the current generation, sitting above the prompt: which model, what angle, what light, how strongly each one applies. Change one of them and everything else stays where you put it.
Keep the whole bar under a name. Open it next month on a different product and the shot comes back the same — same angle, same key light, same treatment. That is the difference between a good result and a repeatable one.
A prompt is a request written in words, and every run interprets those words again. Controls set the variables directly, so the parts you fixed stay fixed and only what you change, changes.
The claim on this page is repeatability, so here it is directly: one saved control setup, applied to unrelated subjects. Same camera position, same key light, same treatment — the products change, the shoot does not.
Four different products, one control setup. Nothing about the lighting or the angle was re-described between them.
More than 100 ready-made controls, grouped the way the interface groups them. Combine them, adjust how strongly each applies, and leave everything else untouched.
Where the camera sits and how tight the frame is. Lock a low three-quarter shot or a tight crop and hold it across every generation in the set.
Where the light comes from and how hard it is — golden hour, soft studio, hard side light. Match a campaign look without re-describing it in prose each time.
What the subject is doing and how they read — walking, mid-turn, hands occupied, adjusting a hem, and the expression that goes with it.
Upload reference images, attach them to a control and save it privately. When the library does not cover your case, you build the control you need.
The library is organised by what each control changes in the frame, so you pick by outcome rather than by prompt wording.
Creative Controls are not a feature of one screen. The same setup follows the work from a single image to a batch of several hundred.
Direct the shot before you generate it, then rerun with one variable changed instead of a new prompt.
Camera movement, motion and action are controls here too. The video set is tuned for motion, but it works the same way as it does on images.
Expose a control as an input on the canvas and it becomes a parameter of the whole workflow, not of one node.
Apply one approved setup across a batch. There is no cap on batch size — how far a run goes is a question of credits, not of limits.
Mujo AI connects the major providers rather than betting on one — OpenAI, Google, BytePlus, xAI, Midjourney and Kling — and new models are added as they launch. Creative Controls sit above all of them, so switching model does not mean rebuilding your setup.
See what's includedGPT Image 2.5, GPT Image 2, Nano Banana 2, Nano Banana Pro, Nano Banana Light, Midjourney, Seedream 5.0 pro, Seedream 5.0 lite and Grok Imagine.
Veo 3.0, Kling 3.0, Seedance 2.5, Seedance 2.0, Seedance 2.0 Fast and Seedance 2.0 Mini, plus Grok.
When a provider releases something worth using, it appears in the same selector — your saved controls keep working.
The point is not the first image. The point is being able to get it again next month, on a different product, without remembering what you typed.
Add a prompt, a product photo or a reference image.
Set camera, lighting, composition, action and style with controls.
Run it on whichever model suits the job.
Keep the whole setup under a name — that saved bar is what you reload.
Load it next time, share it, or run it over a batch.
Controls matter the moment one image stops being enough and a set has to hold together.
When a range has to look photographed in one session, the setup is the product. A saved control bar does what a lighting plan does on a real shoot day.
One person holding a visual identity across stills, video and social has no art director to keep it honest. The saved setup plays that role.
A separate control library per client means nobody has to remember which look belonged to whom, and nothing bleeds between accounts.
Every setup you keep is one you never have to reconstruct. The library is where a lucky result turns into a repeatable one.
The work below has one thing in common: it only counts if the twentieth asset looks like it belongs with the first.
Hold one camera and lighting setup across an entire product range, the way a photographer holds a lighting plan across a shoot day.
Generate a week of posts that read as one account, not as twenty different generators.
Take an approved key visual and produce the crops, ratios and channel versions from the same settings.
Keep a character recognisable across scenes, outfits and lighting instead of regenerating a stranger each time.
Give a clip the same framing and movement language as the stills it sits next to in the campaign.
Run one approved setup across a whole catalogue so it comes out consistent without a person checking each frame.
In most teams the knowledge of what actually works sits with one or two people. Controls move it into the workspace, where everyone can use it.
Publish a setup to the workspace so the next person starts from it instead of a blank prompt. Available from the first paid plan.
Approved looks stay approved, whoever is generating and whichever model they pick.
Decide which models, controls and projects each role can reach.
Wire a control setup into a Node Editor workflow and the whole process becomes repeatable.
You can open the control bar on a free account and see what it does to your own images before any of this becomes a decision.
20 credits for image generation. Full access to the control library; video generation is not included.
Video generation, and shared controls for a team, start at the first paid plan.
Three short guides: choosing a model, using reference images, and building a reusable generation workflow.
Worth knowing before you build a process on top of them.
Controls narrow the range, they do not freeze the pixels. What you get back is a family of results that belong together, not a duplicate.
The same control reads harder on one model than another. That is a property of the model, and it is why comparing the same setup across models is useful rather than redundant.
Controls direct a generation. They do not rescue a badly lit reference photo, and a poor input will show through whatever you set.
Visual settings in Mujo AI that let you set camera, lighting, composition, action and style as parameters instead of describing them in a prompt. You can save any combination as a Control Bar and apply it again later.
The set of controls active on a generation — model, angle, light, style and how strongly each applies. Save it under a name and it becomes a setup you can load on any future project or share with your team.
Yes, for the idea — what the picture is of. Controls handle how it is shot and treated, which is the part that normally drifts between runs.
Save the settings that produced the result you approved, then apply that saved setup to every other image in the set. Consistency comes from reusing the same parameters, not from rewriting the same prompt.
The setup carries over when you switch models, but models differ in how strongly they respond to a given control. Running the same setup on two models and comparing is part of how people use it.
That is what the library is for. A saved Control Bar can be loaded into any project, shared with the workspace, or attached to a Node Editor workflow.
Yes. The video set is tuned for motion — camera movement, action, pacing — but it is the same system and the same saved setups.
Yes. A free account comes with 20 credits for image generation; video generation needs a paid plan, as does sharing saved controls with a team.
A prompt library stores text you paste. A saved control stores the settings themselves, so it keeps working when you change model, subject or format. Mujo has both — 1500+ prompt presets sit alongside the controls.
Open the control bar on your next generation and see how much of the guesswork it removes.
