Miora 0-to-1 Beginner Tutorial: Become a Designer in Five Minutes

Yesterday, Tencent's Miora was officially launched for everyone. No waiting list, no invitation code, and you get 1,000 points upon registration. I spent a whole day exploring the product from start to finish and wrote this guide.

This isn't a review or a sponsored post; it's a practical record of "how I suggest you use Miora if you're opening it for the first time today." I'll cover all the features, but I'll spend the most time on its most unique aspect—the Memory function. This is what truly sets it apart from all other image generation tools on the market.

First, What Is It?

Open miora.design, and the homepage says: The Agentic Creative Studio

Simply put, it's not just a "write prompt → get an image" gacha machine. You give it a requirement, and it breaks down tasks, calls different expert modules, and lays out images, videos, 3D, and UI all at once on an infinite canvas, which you can then continue to modify.

Produced by Tencent, it uses the same architecture as CodeBuddy / WorkBuddy, specifically designed for creative scenarios. It ranked #1 on Product Hunt on July 13 and opened to everyone on July 22.

You only need to know three words to get started:

• Memory: It remembers your aesthetic; it doesn't forget after one use.
• Specialist: It has a group of creative experts for different tasks (image generation, video, 3D, UI...).
• Skills: You can save workflows that work for you to reuse repeatedly.

All three of these run on the same canvas.

After Opening: What the Page Looks Like

After logging in (supports Google / GitHub / Email), you'll see a clean interface. There's a navigation column on the left, the chat area and canvas in the middle, and model configuration at the bottom. Let's go through them.

Left Sidebar: Four Entrances

There are four buttons on the left, corresponding to four capabilities:

Creation—Where you start working. Choose a scenario, write your requirements, click send, and it starts breaking down tasks and producing results.

Inspiration Library—Browse here when you lack inspiration. It contains cases made by others. If you like something, click "Use Prompt," and it will automatically put that prompt into your chat box for you to modify.

Skills—This is where your and the community's reusable workflows are stored. Built-in skills include brand visual packages, e-commerce content generation, illustration storytelling, etc. You can also save your own successful workflows as exclusive Skills.

Memory—This is where you view and edit what Miora remembers about your rules. This is a key focus later in this article.

Scenario Modes

After clicking Creation, it will ask you to choose a scenario. This step isn't just for show—choosing the right scenario helps it use the appropriate professional process to break down your requirements.

Current common scenarios include:

• Brand Design: Logo, brand visuals, IP image, full VI package.
• Film & TV Creativity: Storyboards, concept art, promotional visuals.
• E-commerce Advertising: Main images, posters, selling point graphics.
• Interactive Games: Characters, settings, game visuals.
• Web Applications: Web / App interfaces, landing pages.

Behind each scenario is a set of Skills preset by Tencent. For example, if you choose an IP design package, it won't just give you one image; it will follow a professional process to create three-view drawings, cards, emojis, posters, and promotional materials, filling the canvas at once.

Example: Based on a one-sentence scenario, after providing options for direction, it starts the full design process immediately.

Chat Area: Write Your Requirements Here

This is where you tell Miora "what I want."

The most common mistake for beginners is writing too little. "Draw a cute cat"—this kind of command won't yield good results in any AI tool. In Miora, you need to write a few more sentences.

A good requirement should include:

1. What you want to do (A poster? A logo? A set of brand visuals?)
2. Who it's for and where it's used (Social media? E-commerce main image? Offline materials?)
3. Style and feel (Minimalist? Cinematic? Japanese? Cyberpunk?)
4. Any restrictions (What colors or styles to avoid)

For example, instead of writing "Make a coffee brand logo," write:

1I am creating visuals for a plant-based light food delivery brand.
2Need: One logo, 3 social media images.
3Style: Clean, green, modern, no flashy decorations.
4Usage: WeChat Official Account and Xiaohongshu.

Also, a very handy feature: there's an "Enhance Prompt" button at the bottom right of the chat box. Click it after writing in plain language, and it will help translate your colloquialisms into structured professional commands.

Canvas: Where the Real Work Happens

The canvas is the heart of Miora.

It's not a dry image preview page, but an infinite workbench where you can freely place, combine, and edit. Images, videos, 3D models, UI designs, and text are all placed on it as "nodes" that you can drag around and modify at any time.

Crucially: Miora can understand what's on the canvas. It knows what's currently there and the relationships between them. So if you circle an image on the canvas and say "make this title bigger," it knows which image and which title you're talking about.

In short: Chat is where you give commands; the canvas is where the work gets done.

Local Editing: Point and Change

This feature is extremely practical and often overlooked.

Ordinary image tools require you to regenerate the whole image to change one detail, only to find other parts have changed, leading to an infinite loop. Miora doesn't do that. After generation, you can directly on the canvas:

• Pencil Circling: Draw freely like on paper, using different colors to mark "keep," "delete," or "move."
• Marquee Editing: Select an area to modify directly.
• Local Fine-tuning: Change text, angles, or relight only the specified part.
• Crop / Erase / Remove Background.
• HD Upscale / Outpainting: Pull multiple images to a uniform size in one step.

After circling, just use plain language to say how you want to change it, for example:

1The yellow marked line is the flight path of the butterfly; make it fly in this direction.

Or:

1Only change the title text to "Half price for the first week of opening," leave everything else untouched.

Bottom Configuration: How to Choose Models

There are two layers of settings at the bottom. Beginners often get confused, so let me clarify.

The first layer is the Agent base model, the "Brain." It's responsible for understanding your needs, breaking down tasks, and organizing the process. There are three levels:

• Standard: Enough for simple tasks, like generating one image or changing text.
• Pro: The default choice for daily use; can handle most tasks.
• Max: Use this for complex full packages, like a complete brand VI.

The second layer is the multimodal generation model, the "Execution Layer." These are the mainstream models that actually generate images, videos, 3D, and UI. It even helpfully labels the price. You can specify them yourself or choose "Auto."

Expert modules already online include: Image Generation, Audio/Video, 3D Modeling, Video Rendering, and UI Design. When you submit a requirement, the main Agent understands the goal, automatically disassembles it, schedules these experts, and generates a whole set of things on the same canvas while maintaining style consistency.

• Image Generation: Nano banana, ChatGPT Image 2

• Video Generation: Seedance 2.0, Kling, Q3, etc.

There are two Execution Modes:

• Ask: It confirms the direction with you before acting, suitable for when requirements aren't yet stable.
• Auto: It gets straight to work, suitable for when you've explained clearly and want quick results.

For beginners, I suggest this setup: Scenario: Brand Design or E-commerce Ads → Agent: Pro → Generation Model: Auto → Mode: Ask.

Inspiration Library and Skills: Don't Start from Scratch

The Inspiration Library has many ready-made cases. Just pick one you like → click "Use Prompt" → it populates the chat box → modify and run. It's better than staring at a blank box.

Skills go a step further. Instead of just a prompt, it gives you a whole workflow. For example, the "Brand Visual Package" Skill will follow a professional process step-by-step: Logo first, then main color scheme, then application scenarios, and finally promotional materials.

Once you've run a workflow you like, you can save it as an exclusive Skill for one-click reuse or sharing.

A subtle but brilliant detail: when you save a workflow as a Skill, Miora automatically bundles your current memory rules with it—colors, styles, and taboos all follow. Future projects using this Skill will start with your brand settings.

Export

Once finished, you can download individual assets or the entire asset package. Skills can be shared with teams or the community.

Memory Function: The Core of This Article

Now for the most important part. My biggest takeaway is that Memory is the true watershed between Miora and ordinary image tools. It determines whether you're using a "one-time gacha machine" or a "creative partner that understands you better over time."

What Exactly Does It Remember?

The official website defines memory as a "methodology-level memory mechanism." It says:

Most tools forget you after every session. Miora remembers — your brand colors, your tone, your creative rules, your "never do this." The longer you work with Miora, the more it sounds like you.

It doesn't just remember static assets like "which image you uploaded," but the "methodology" of how you design. Specifically:

• Aesthetic Preferences: You like minimalist, cinematic, or Japanese fresh styles.
• Brand Standards: Main colors are off-white and dark brown; the logo must be centered.
• Layout Habits: Large titles, ample white space, minimal decoration.
• Tone: Gentle, restrained, not flashy.
• Forbidden Rules: No anime style, no neon, no excessive retouching.
• Project Context: All subsequent materials for this brand follow the same visual set.

In short: It remembers "how you design," not just a "temporary asset."

Memory Is Not a Black Box

Many AI tools have "memory" or "preferences," but they are black boxes—you check an option and that's it. You can't see or change what it remembered.

Miora is different. The official statement is: Memory is not a black box; users can view and edit the memory board at any time.

This means you should develop a habit: After giving a memory command, check the Memory entrance on the left. Confirm if it:

• Remembered everything.
• Remembered correctly.
• Didn't mistake a temporary preference for a long-term rule.

If something is wrong, change it on the spot. Memory is inspectable and editable.

Why It's Worth Learning

Think about the pain points of ordinary AI tools:

1. Explaining the style from scratch every time.
2. Results "drifting"—wanting the same vibe but getting something different every time.
3. Difficulty unifying series content—covers, headers, posters, and videos all having different styles.
4. Having a mature "recipe" in your head that the tool can't grasp.

With memory, the process becomes:

1. You set a set of rules.
2. It remembers them.
3. Next time, you only say the topic.
4. It automatically follows your style.

As the official saying goes: "By the tenth collaboration, the results look like they came from your own hand."

Miora's memory is divided into Personal Memory and Project Memory. This ensures all designs fit your personal tone while allowing specific rules for certain projects.

Step-by-Step: Four Steps to Master Memory

1. First, produce a set of results you're happy with. Don't record from thin air. Get a result where the colors, layout, and feel are just right.
2. Explain clearly what you like and have it remember. Send a paragraph with the rules you want to use long-term.
3. Check the Memory board. See what it recorded. Did it miss anything? Did it get it wrong?
4. Next time, only state the topic to test if it really remembered. Send a new requirement without style rules, just the topic.

How to Write Effective Memories

Bad memory (too vague): Remember that I like a high-end style.

Good memory (actionable): Please remember: 1. Overall style: Minimalist, restrained, artistic. 2. Main colors: Off-white + Light coffee. 3. Layout: Large titles, centered, ample white space. 4. Prohibited: Anime, neon, over-decoration. 5. Usage: Priority for social media covers and posters.

Advanced: "Reverse Engineer" Style from Your Own Work

This is my most recommended path:

1. Drop your most satisfying finished image onto the canvas.
2. Ask it to analyze: color scheme, layout, font temperament, white space rhythm, and decorative elements.
3. Write the analyzed standards into memory.
4. Apply them directly in the future.

Memory vs. Skill

Memory is your "aesthetic manual"—it's about preferences and standards to make results look more like you.

Skill is your "Standard Operating Procedure"—it's a repeatable workflow to make the process faster and more stable.

Memory determines "how much it looks like you"; Skill determines "whether it can do a whole set."

Three Common Pitfalls for Beginners

1. Using without recording: Explaining the style every time is a waste of Miora's best feature.
2. Recording too vaguely: "High-end feel" is useless. Use specific colors and layouts.
3. Recording temporary preferences: Don't record a one-off experiment as a long-term rule, or your future results will become a mess.

First Time Use: Follow This Workflow

1. Open miora.design and register (get 1,000 points).
2. Click Creation, choose "Brand Design" or "E-commerce Ads."
3. Config: Agent: Pro, Model: Auto, Mode: Ask.
4. Write clear requirements (Goal / Usage / Style / Limits).
5. Review results, make specific edits on the canvas.
6. Write the memory immediately after you're satisfied—don't skip this.
7. Check the Memory board, then test with a new topic.

Who Is It For?

Suitable for: Designers, operators, influencers, marketers, and small teams who need visual consistency.

Not suitable for: Those who want purely random generation without writing requirements, or those needing professional-grade 3D models or long cinematic videos (current limit is ~30s).

Final Word

If you only remember one thing: Ordinary tools treat you as a stranger every time; Miora's memory makes it your creative partner.