Where do I start on my first day in MITO?

Open Home and either pick one of the six ways in, like Make a product ad or Create an avatar, or choose Start from scratch for an empty project. Everything your team has made lives under Projects, and every single generation across every project lives under Generations. Your credit balance sits at the bottom left of the menu.

What are the three tabs inside a project?

Canvas is the open board where images, video, references and notes sit side by side. Generations is everything that project has produced, in order. Timeline is where clips get assembled, trimmed and exported. You will spend day one entirely on the Canvas.

What can the AI Director actually do?

Describe a film, ad, scene or idea in plain language and it takes it from there: it reads what is already in your project, writes the prompt, picks the model, sets the options and generates. It will also build a storyboard, animate and edit what it made, and create an actor for you, and it suggests the next move once it has finished.

Can I talk to Director instead of typing?

Yes, on paid plans. The wave icon beside the send arrow in the Director panel starts a call, and you describe what you want out loud. It asks follow-up questions as you go, so it works like briefing someone rather than dictating a prompt.

Does talking to Director cost credits?

No. Only generations spend credits, meaning images, video and audio. Asking Director questions, getting a second opinion on a shot or having it explain a step costs nothing, so there is no reason to ration the conversation.

Which model does Director use?

It chooses one for you. The selector sits on Auto, which resolves to a balanced default rather than the most expensive option, and you can always see what it picked. Overriding it is possible but rarely the first thing worth doing.

What is the difference between Auto and letting Director decide?

Auto is a setting inside the manual panel: you have opened the panel and written the prompt yourself, and you are leaving just the model choice to MITO. Letting Director decide means you did not open the panel at all, and it handled the prompt, the model and the settings together. Auto picks a model. Director runs the job.

When should I use the manual panel instead?

When you already know exactly what you need and Director’s output is nearly there. It is a finishing tool: the last ten per cent on a shot that matters, not the way to start. If you are not sure what you want yet, the panel will slow you down.

How do I open the manual panel?

Double-click anywhere on the canvas. You get a New image or New video panel with the model selector, an optional reference slot, the prompt box, and the size and resolution controls behind the chip in the bottom left.

How do I know what a generation will cost before I run it?

The credit cost sits next to the Generate button and updates as you change the settings. Video is priced per second of output, so the cost moves with clip length as well as with the model: a 10-second clip is 400 credits on Omni 1.1 Flash and 4,000 on Kling 3.0 Pro. Prices as of October 2026.

Can I give Director my own images as references?

Yes, and it is one of the highest-value things you can do. Put them on the canvas and Director reads them. If you want rules it follows on every generation rather than just this one, write them into a Document.

Your First Day in MITO: Find Your Way Around, Then Make Something

This is day one, and it is three things. Find your way around, make an image, then make that image move. By the end of it you will have made something, which is the only part that actually teaches you anything.

Describe what you want and Director does the rest. It reads your project, writes the prompt, picks the model, sets the options and generates. That is the route to use for almost everything. The settings panel is still there, and this post covers it, but it belongs at the end of a job rather than the start.

Find your way around first

Five minutes here saves you an afternoon later.

The tour, in one minute and twelve seconds. Everything below is the same ground in writing.

The menu on the left is the whole app. Four of them matter today. Home is where a project starts. Projects holds everything your team has made. Generations is every single thing you have generated, across every project, which is where you go when you know you made it and cannot remember where. Tutorials is short videos, including the ones these posts are built on. Your credit balance sits at the bottom of that menu, under Referral and Help, and it is worth knowing where it is before you start spending it.

Home asks you what you are making. Six ways in: make a product ad from a product photo, make a music video from a track and a mood, bring an idea to life from a script, build a character you can reuse from one photo, create an avatar from a face and a script, or start from scratch with an empty project. The first five are shortcuts that set the project up for a known job. Start from scratch is the one to pick while you are still learning, because nothing is pre-decided.

Inside a project there are three tabs. Canvas is an open board where images, video, references and notes all sit next to each other, which is where today happens. Generations is that project’s output in order. Timeline is where clips get cut together and exported, and that is day two.

Director is the round button in the bottom right. It is the thing that makes the rest of this post short. Open it, say what you want, and it reads what is already on the canvas before it writes anything.

That is the tour. The rest of today is making something.

What Director is actually good at

It reads the whole project first. Before it writes a prompt or generates anything, it looks at what is already on your canvas: your references, your earlier shots, your Documents. You are not starting from a blank prompt box every time, which is the thing that makes most AI tools exhausting.

It turns a sentence into a production. Not just one image. Describe a film, an ad, a scene or an idea and it will build a storyboard, generate the images, animate and edit what it made, and create an actor if the piece needs one. We put that to the test on a 30-second explainer, where one sentence came back as a script, a shot plan, character and location references, sound design and a timeline.

It picks the model so you do not have to. The selector sits on Auto and resolves to a balanced default rather than the most expensive thing in the list. You can see what it chose at any point.

It asks when the brief is thin. If what you have said leaves something important open, like who the piece is for or what format it ships in, it asks rather than guessing. Answering those two questions properly is worth more than any setting in the panel.

It tells you what to try next. Once it has generated, it suggests the follow-ups: another variation, another shot in the same series, a tweak to the one you like.

You can talk to it out loud. The wave icon beside the send arrow starts a call, so you can describe what you want in speech rather than typing it. More on that below. On paid plans.

And it is free to talk to. Only generations cost credits. Asking questions, getting a second opinion, having it explain itself: none of that spends anything, so there is no reason to be economical with the conversation.

Making an image

From a sentence to finished images, and what the panel looks like if you need it.

Tell Director what you want and look at what comes back. If it is close, say what to change in words. Most of the work is in the describing, not in the dialing.

So describe properly, in plain words. The difference between a weak result and a good one is usually four things in the sentence: the subject, the setting, the light, and the framing. “A woman ordering coffee” gets you a stock photo. “A woman ordering a cup of coffee, warm morning light through the window, shot close on her face” gets you a shot.

Note what that is not. No syntax, no weights, no model name, no comma-separated tag soup. Prompt craft is a real skill and the manual panel still rewards it, because there you are writing the instruction the model receives. Talking to Director, you are not: you say what you can picture and it writes the prompt. That is the point of it.

Or just say it out loud

Typing is not the only way in. The wave icon sitting beside the send arrow in the Director panel starts a call, and you can simply describe what you want.

It is a conversation, not dictation. You say what you are after, it asks the things it needs to know, you answer, and it goes and makes it. Closer to briefing a colleague than filling in a form.

Which matters more than it sounds. Writing a brief makes you compose: you tidy the idea into a sentence, and some of what you actually meant falls out on the way. Speaking skips that. You describe the thing you can see in your head, including the half-formed parts, and Director does the work of turning it into an instruction a model will follow.

Calling Director is available on paid plans. There is a full walkthrough of a session run by voice, from a spoken sentence to a finished 30-second film.

Making a video

An image into a five-second clip, once by asking Director and once by hand.

Same move. Describe the shot, or point at an image you already made and ask for it to move. Director handles which model runs it, and the model choice for video is a genuinely awkward decision to make cold, because the models differ on clip length, on price per second and on what they can actually do.

You can still see what it chose, and what the alternatives cost, in the selector:

Model	Clip length	Credits per second	Can do
Omni 1.1 Flash	3 to 10s	40	Start/End, Video, Audio
H3 Max	5 to 15s	85	Start/End, Audio
Seedance 2.5	4 to 30s	110	Start/End, Video, Audio
Kling 3.0 Pro	3 to 15s	400	Start/End, Multi-shot, Audio

Featured models, prices as of October 2026. The full families, Kling, Google, Seedance, MiniMax and Wan, sit underneath, with a Compare models link beside them.

Video is priced per second, which is the thing that catches people out. The same 10-second clip is 400 credits on Omni 1.1 Flash and 4,000 on Kling 3.0 Pro. Ten times the model, ten times the bill, every second.

And the capability column usually decides it before the price does. A clip that has to begin and end on specific frames needs Start/End. Several shots out of one generation is Kling’s Multi-shot. If the shot does not need either, the cheap model is not a compromise.

Starting from text, or from an image

Every video generator makes you answer this, and it is worth answering on purpose.

From text is one sentence straight to a clip. Fastest, and the model decides what everything looks like, so it suits a shot where the look is not load-bearing.

From an image means you generate a still first, get it exactly right, and then ask for it to move. The still becomes the clip’s first frame, so you have already settled the composition, the colour and the cast before any motion exists. For anything that has to match other shots, this is the one to use. The technique has its own guide: what a start frame is and what it controls.

Director takes either. Describe the shot, or point at an image on the canvas and say what should happen next.

One ceiling to know about. A single generation runs between about 3 and 30 seconds depending on the model, so anything longer is several clips assembled rather than one long render. That is a timeline job, not a prompting problem, and it is covered end to end in the explainer walkthrough.

The one moment to open the manual panel

Double-click the canvas and you get the full manual controls: the model, ten aspect ratios, 1K to 4K, an effort slider, a reference slot, and the live credit cost.

The moment that earns is specific. Director has given you something nearly right, it is going somewhere that matters, and you know exactly what needs to change. A hero image that has to be 4K rather than 1K. A shot that has to be 21:9 because that is what the edit takes.

Before that point it will slow you down, because it asks a dozen questions you do not have answers to yet. Director asks one.

It also hands you the prompt. In the panel the words you type go to the model as written, so getting what you pictured is a matter of knowing how to write for it. That is a learnable skill and worth having. It is just not a skill you should need in order to start.

The short version. Start with Director every time. Describe the idea, look at what comes back, say what to change. When a piece matters and only the details are left, open the panel and finish it by hand.

All of these walkthroughs are inside MITO too, under Tutorials, alongside the rest.

Before you close the tab

Three things, fifteen minutes, and day one is done properly.

Open MITO and find two things: your credit balance at the bottom of the left menu, and your Projects page. That is the whole tour and you only need to do it once.

Make one image by describing it. Start from scratch, open Director, and say what you can picture. Subject, setting, light, framing. Look at what comes back and say what to change, in words.

Turn that image into a five-second video, twice. Once by asking Director, once by opening the panel yourself. Doing both is the point: you will not feel what Director is saving you until you have seen the panel it is saving you from.

Tomorrow is editing, which is where a pile of clips turns into a piece.

Open the canvas and describe something →