『Tsumugi's🔨WishMallet』

One prompt. Any scene you can imagine.

Tsumugi falling from a cliff towards the sea

Hollywood-classvideo is yours.

Tsumugi’s source photo

Start withjust one photo.

Tsumugi in a motion-comic style video

Motion comicsare yours too.

Source photo of a 50-year-old Japanese man

A photo + a voice.Become anyone.

16-year-old Japanese schoolboy

Younger

16-year-old Japanese schoolgirl

Gender swap

16-year-old Black schoolgirl

New look

0-year-old Black baby girl

Baby

100-year-old Black elderly woman

Age 100

What if you were a cat?

Animal

And finally, your anime self.

Anime

🔒 No Cloud. No Uploads. No ComfyUI!

Try Free for 7 DaysCheck my PC

TSUMUGI’S LITTLE CINEMA

Give a photo motion. And a story.

Explore the cinematic effects you love. These eight playful “Tsumugi’s Extraordinary Everyday” Reels were actually created with 『Tsumugi's🔨WishMallet』 and published on Instagram. Each comes with ideas for your own first scene.

08REELS / CINEMATIC IDEAS

One prompt with 『Tsumugi's🔨WishMallet』!All eight videos were generated from a single prompt in WishMallet.

ACTUALLY CREATED WITH WISHMALLET

Tsumugi suspended in mid-air above the sea, from the original generated video.Watch this ReelPublished on Instagram

Playback includes audio.

Open on Instagram

SLOW-MOTION FALL

Don’t push me!

One falling moment. A movie scene.

A seaside “Don’t push!” turns into an unexpected action scene. Watch the expression in mid-air as time slows down.

The craft behind the effect

Falling action reminiscent of Die Hard

Live-action slow motion can involve overcranking—recording at a higher frame rate—alongside compositing. Stunt preparation and keeping a falling subject in focus take specialist work.

One prompt with 『Tsumugi's🔨WishMallet』!

With WishMallet, start with a photo and describe the fall. Explore a cinematic idea without assembling a camera rig.

See a prompt starting point
She falls in cinematic slow motion, with a surprised expression.

An idea to build on, not the original production prompt.

Output size
736 × 1280px
Requested length
15sec
References
1 image · 1 audio
Tsumugi leaning back to dodge suspended snacks, from the original generated video.Watch this ReelPublished on Instagram

Playback includes audio.

Open on Instagram

BULLET TIME

Snack time

Snacks freeze. The camera circles.

Incoming snacks turn an ordinary afternoon into an action movie. Tsumugi leans back as the camera moves around her.

The craft behind the effect

The shifting perspective made famous by The Matrix

Bullet time moves the viewpoint through an apparently frozen moment. The film combined an array of cameras with image processing to create its distinctive orbit.

One prompt with 『Tsumugi's🔨WishMallet』!

With WishMallet, describe what freezes and how the camera moves. You can explore the effect without a roomful of cameras.

See a prompt starting point
She slowly dodges floating snacks while the camera circles around her.

An idea to build on, not the original production prompt.

Output size
736 × 1280px
Requested length
12sec
References
2 images · 1 audio
A blue spirit appears beside Tsumugi and her friends in class, from the original generated video.Watch this ReelPublished on Instagram

Playback includes audio.

Open on Instagram

REALISTIC × ANIMATED

I forgot my lunch!

Live action meets 3D animation.

A blue spirit emerges from a lamp at lunchtime. Realistic people and an animated character share one scene, complete with dialogue.

The craft behind the effect

Human and CG character performances, as in Aladdin

Film production can combine performance and facial capture, 3D animation, lighting and compositing. Making characters share the same space involves many specialist steps.

One prompt with 『Tsumugi's🔨WishMallet』!

Choose reference images and voices in WishMallet, then use the guide to describe the entrance and dialogue. Start a shared scene without learning 3D production.

See a prompt starting point
A blue spirit emerges from a little lamp in a classroom and speaks to the friends.

An idea to build on, not the original production prompt.

Output size
480 × 864px
Requested length
12sec
References
2 images · 2 audio
A dark creature holding an apple beside a girl, from the original generated video.Watch this ReelPublished on Instagram

Playback includes audio.

Open on Instagram

PHOTOREAL CREATURE

Sign this!

An impossible presence, right beside you!

A request to sign a black notebook leads to an unexpected encounter. Look at the creature’s textures and its presence beside a person.

The craft behind the effect

A creature in a realistic world, reminiscent of Death Note

Photoreal CGI builds a believable creature; VFX compositing places it into filmed footage. Matching light, shadow and movement adds substantial work beyond the model itself.

One prompt with 『Tsumugi's🔨WishMallet』!

In WishMallet, start with a reference image and describe who the creature stands beside and what it does. Try an impossible encounter without complex modelling.

See a prompt starting point
A giant dark creature stands beside a girl, holding an apple.

An idea to build on, not the original production prompt.

Output size
736 × 1280px
Requested length
12sec
References
2 images · 2 audio
Tsumugi with a cat and dog on a sofa, from the original generated video.Watch this ReelPublished on YouTube

Playback includes audio.

Open on YouTube Instagram post

ANIMAL ACTORS & FUR

Luna and Leo

Animals take centre stage.

Luna the cat and Leo the dog join Tsumugi on screen. Sunglasses, playful gestures and fluffy fur give these pets their own little performances.

The craft behind the effect

Expressive animal performances, as in the live-action Lady and the Tramp

Getting an animal to perform on cue is challenging. CG animals also need specialist work on fur simulation, expressions and convincing body movement.

One prompt with 『Tsumugi's🔨WishMallet』!

Bring reference images of the person, cat and dog into WishMallet and describe their actions. Explore a shared scene without animal training or building CG fur.

See a prompt starting point
A cat in sunglasses dances proudly while a happy dog watches beside it.

An idea to build on, not the original production prompt.

Output size
544 × 960px
Requested length
11sec
References
3 images · 0 audio
Tsumugi walking through a vast desert, from the original generated video.Watch this ReelPublished on Instagram

Playback includes audio.

Open on Instagram

DESERT SCALE & ATMOSPHERE

I’m thirsty!

A vast desert landscape.

Tsumugi crosses a sun-scorched desert and encounters something surprisingly ordinary. An epic landscape meets a small everyday punchline.

The craft behind the effect

Desert scale reminiscent of Lawrence of Arabia

Long-lens compression and heat haze can make a desert’s scale and heat palpable. A live shoot also needs the right location, weather and placement of the actor.

One prompt with 『Tsumugi's🔨WishMallet』!

Start in WishMallet with one photo and describe the place, action and atmosphere. Set a story in an expansive landscape without travelling to a location.

See a prompt starting point
She walks through a vast, shimmering desert looking for water. A vending machine appears in the distance.

An idea to build on, not the original production prompt.

Output size
480 × 864px
Requested length
11sec
References
1 image · 1 audio
Tsumugi looking back at a busy waterfront, from the original generated video.Watch this ReelPublished on Instagram

Playback includes audio.

Open on Instagram

LARGE SETS & LIVING CROWDS

Summer blue, looking back

Even the background tells a story.

A waterfront, passing visitors and Tsumugi turning towards the camera. This summer moment feels like a theme-park visit. Look at the depth of the scene.

The craft behind the effect

Expansive worlds like those in Titanic and Gladiator

Large film productions combine physical sets, digital extensions and crowd work. Aligning the architecture, perspective and background movement takes careful preparation.

One prompt with 『Tsumugi's🔨WishMallet』!

Give WishMallet a reference image and describe the place and action. Explore an entire setting without building large sets or arranging a crowd of extras.

See a prompt starting point
At a lively waterfront theme park, a girl slowly turns around and smiles.

An idea to build on, not the original production prompt.

Output size
736 × 1280px
Requested length
6sec
References
1 image · 0 audio
Tsumugi playing the violin in a music room, from the original generated video.Watch this ReelPublished on Instagram

Playback includes audio.

Open on Instagram

MUSICAL GESTURES & EXPRESSION

Violin practice

Give your story a musical moment.

Tsumugi plays the violin in a music room. The bowing arm, the supporting hand and her expression come together in one short scene.

The craft behind the effect

Performance scenes like those in The Pianist and Ladies in Lavender

Live-action musical scenes require attention to hands, posture and timing with the sound. Musicians, rehearsal, filming and editing help create a convincing performance.

One prompt with 『Tsumugi's🔨WishMallet』!

With WishMallet, begin with one photo and an idea for the performance. Describe the bow movement and expression to explore a musical scene.

See a prompt starting point
In a music room, a girl plays the violin with smooth movements of the bow.

An idea to build on, not the original production prompt.

Output size
736 × 1280px
Requested length
10sec
References
1 image · 0 audio

Swipe or choose a thumbnail to explore the next Reel

@tsumugi_hautelab

All Reels were generated with MiniMax H3. Values are the creator’s generation settings; edited Instagram running times may differ. Zero audio references does not mean silent output. Results vary with references, prompts and settings.

From enjoying a scene to creating your own.Explore the three steps

INSIDE WISHMALLET

One photo.
A whole new you.

What if you were 16 again? Or 100? From age, gender and appearance to becoming a cat or an anime character, you can easily create “what-if” videos.

Explore the transformationsReal generated videos and a three-step guide
Original photo of the 50-year-old man

START WITH ONE PHOTOIt all starts with this photo.

Rejuvenation
Rejuvenation
Infancy
Infancy
Anime self
Anime self
See what happens next.

Even vague instructions are fine — AI automatically refines them into officially recommended prompts!

1Say what you want to see.

A beautiful female samurai quietly readies her sword in a storm of cherry blossoms

2Press one button.

Auto prompt

3A beautiful scene is ready.

The actual WishMallet screen: a female samurai among cherry blossoms, with other generated videos in the historyClick to play this one-shot video

MORE WAYS TO CREATE

Keep creating, after generation.

Keep a favourite expression. Share your work. Create another scene. Discover more ways to enjoy what you make.

WishMallet export screen with formats for Google Colab, Floyo and ComfyUI.

Export

More places to bring your ideas to life.

Take your favourite video settings to another PC or the cloud. Spend less time retyping and more time creating your next scene.

Explore the feature
WishMallet Best Shot Album with favourite moments from several videos arranged as photos.

Best Shot Album

Turn a moment into a collection.

A lovely expression. A scene you got just right. Save favourite moments as photos and enjoy them together in your own album. Enhance a chosen frame with AI and save a large PNG for a book cover or poster.

See album and high-quality save
WishMallet Video Editing after a mask is applied, checking a comic page where only the panel being read moves.

Motion Comic

Motion comics made simple.

Turn a single comic page into a vertical short where only the panel you are reading moves and speaks. Lay a mask image with that panel made transparent over the video, and the rest stays still—no specialist software required for TikTok or YouTube Shorts.

See how it works
Full Privacy Processing screenshot showing four target categories, processing strength, automatic review and advanced settings for methods and coverage.Full Review Mosaic Processing screenshot showing FANZA Doujin and DLsite options, strength, guideline details, additional areas and the automatic processing button.

Privacy & Review Mosaic Processing

Prepare your creation to be shared with care.

Hide faces, whole people, sensitive body areas or licence plates with Privacy Processing, or choose publishing-platform presets with Review Mosaic Processing. Both open from buttons in Video Editing.

Explore the feature

LESS SETUP. MORE CREATION.

You just wanted to make a video.

Does getting MiniMax H3 running feel harder than creating the video itself?

  • A day spent on setup.

    Python, CUDA, model folders. You came to make a video, but are still preparing.

  • Too many starting points.

    Which model? Which workflow? Research keeps getting between you and your idea.

  • Stopped by a wall of nodes.

    An unfamiliar web of connected boxes. It is hard to know where to begin.

A CLEARER WAY TO START

A simpler path to your first creation.

『Tsumugi's🔨WishMallet』 gives MiniMax H3 a Windows interface. Create through the app without building node workflows.

BEFORE

Setting everything up yourself

A puzzled creator surrounded by installation components, competing settings and tangled nodes, struggling to begin making a video.

Running a large AI model such as MiniMax H3 on a local PC takes more than simply installing it. WSL2, CUDA, Python, SGLang, quantization, GPU memory, and other technical components must work together correctly. Even experienced engineers may need several hours to half a day, or more than a day if problems arise. Without prior expertise, researching and setting it up can take several days to over a week. This is not an environment anyone can use immediately: building it requires advanced knowledge and time.

These are the settings and dependencies that need to be configured when building the environment.

The stack has to be built while checking compatibility across OS, GPU, CUDA, libraries, models, and VRAM settings.

A mismatched combination can fail to start, stop mid-run, or become extremely slow.

AFTER / WITH WISHMALLET

An interface for creating video

One clear path connects an automatic PC check, installation, a guided photo and prompt, and video generation. A smiling creator watches the finished cat video.
  1. Automatically check your PC
  2. Automatically install the full environment
  3. Write the description with Guided writing or Auto prompt
  4. Press the Generate video button

Automatic installation prepares the generation environment on first use. In-app diagnostics help you check readiness.

YOUR FIRST VIDEO

Prepare. Describe. Generate.

A clear sequence of actions takes you toward your first video.

  1. 01Prepare

    Begin with your PC.

    Automatically check your hardware, then follow the app to automatically install WSL2 and the generation environment.

  2. 02Describe

    A photo. An idea. Your words.

    Add a photo and a description in the Create tab. The guided prompt builder helps you put your idea into words.

  3. 03Generate

    Let your PC take it from here.

    Choose length and quality, then select Generate video. The finished video is saved and can be played in the app.

Rendering time depends on your GPU, video length and quality settings.

Read the guide

NO NODES. NO WORKFLOWS.

Your idea comes first.
No nodes to learn.

『Tsumugi's🔨WishMallet』 does not use ComfyUI internally. Its interface focuses on creating video with MiniMax H3.

No ComfyUI installation required

No node connections. No workflow hunting.

  • 01

    Find the words for your idea.

    Answer the guided prompt builder’s questions. “Auto prompt” can build a description from just a few words.

  • 02

    Keep working from where you left off.

    History stores results and settings. Load the same setup, make small changes and try another version.

  • 03

    Put your GPU to creative use.

    Generate video and audio on your Windows PC’s NVIDIA GPU. Turn compatible hardware into your local AI workspace.

A closer look at the engine

An example ComfyUI-based setup

  1. GUI
  2. Node Workflow
  3. ComfyUI
  4. Model

『Tsumugi's🔨WishMallet』

  1. Windows UI
  2. SGLang
  3. MiniMax H3
  4. Your GPU

SGLang runs MiniMax H3 inside WSL2 on Windows. ComfyUI is a powerful option for flexible custom workflows. WishMallet is for people who want to create with MiniMax H3 without starting by learning nodes.

BEFORE YOU BUY

Will it work on your PC?
Find out before you buy.

Use the browser-based compatibility checker to review your hardware before making a purchase.

Check my PC

Open the checker on the Windows PC you plan to use.

Think a GPU is too expensive?Could a JPY 500,000 GPU cost less?Compare cloud generation costs with a local GPU. Calculate the difference for your workload.Compare costs

Hardware at a glance

For local generation on a compatible NVIDIA GPU.

ComponentMinimum guidelineRecommended
OSWindows 11 (64-bit)Windows 11 (64-bit)
GPUNVIDIA GeForce RTX 30 series or newerRTX5060 Ti / 5070Ti 16 GB or newer
Video memory (VRAM)12 GB or more16 GB or more
System memory (RAM)64 GB or more96 GB or more
Free storageAllow around 120 GiB170 GiB or more
CPU6 cores / 12 threads or more8 cores / 16 threads or more

The GPU must support Compute Capability 8.6 or higher. Check the exact model and VRAM capacity. AMD and Intel GPUs are outside this app’s supported configuration.

WSL2 and the generation environment require initial setup. The help lists 105 GB of free storage; the checker uses around 120 GiB as a minimum guideline. Allow extra space for additional models and saved videos.

This is an approximate browser check. Some values, including RAM and free storage, may need manual input. The browser cannot verify drivers or WSL2 readiness. Use the app’s environment diagnostics for the final readiness check.

Help with environment preparation

YOUR OWN CREATIVE STUDIO

Your computer.
Your creative space.

A downloadable video creation app for your own Windows PC. Access MiniMax H3 through a focused, approachable interface.

  • 🔒 No data uploads. No cloud storage. No ComfyUI required. Everything runs locally on your PC.
  • 🎬 Turn a single photo into the video you imagine.
  • 🔰 Beginner-friendly: the “🧭 Easy creation guide” and “✨ Auto prompt”.
  • 🔁 Easily recreate videos using the same settings from your history.
  • ☁ Easy export for high-performance GPU cloud environments, with support for Google Colab & Floyo.
  • ★ Save your favorite video moments as images with the Best Shot Album feature.
  • 🛡 AI-powered automatic privacy protection.
  • ▦ AI-powered automatic mosaic processing for content review, compatible with Japanese adult content sales platforms.

『Tsumugi's🔨WishMallet』

Powered by MiniMax H3

Downloadable software for Windows 11

Start with 7 days, free.

Download WishMallet for free and try it for 7 days.

  • Windows 11
  • NVIDIA GPU
  • VRAM 12 GB or more
  • 105 GB or more free space
  • Up to 3 videos a day during the free trial
Check my PCDownloadBuy Full Version License

Gumroad sells the full version licence (an activation code). The app itself is downloaded free from this site; entering the code unlocks the full version.

GOOD TO KNOW

Before you begin.
Your questions, answered.

A few helpful answers before your first video.

Read the guide
Can I try it before buying?

Yes. There is a 7-day free trial: download WishMallet from this site for free and start it without an activation code. During the trial you can generate up to 3 videos a day. To keep using it afterwards, buy a full version licence (an activation code).

Do I need to install ComfyUI?

No. WishMallet does not use ComfyUI internally. It runs MiniMax H3 through SGLang, without node connections or workflow editing.

Do I need Python or command-line knowledge?

Everyday video creation happens through the app. Initial WSL2 and generation environment setup still needs to be completed using the instructions. Preparation or troubleshooting may require commands.

Will it run on any PC?

You need Windows 11 and a compatible NVIDIA GPU. The guideline is at least 12 GB VRAM and 64 GB RAM, with additional requirements for the exact GPU model and free storage. Check your PC before buying. Mac, AMD and Intel GPUs are outside this app’s supported configuration.

Where do I get my first photo and prompt?

Use your own photo or try the included sample photos and voices. Start with a short description; the guided prompt builder and “Auto prompt” help you write it. Only use materials you have the right to use.

How do I prepare the models?

Install the inference environment and MiniMax H3 models inside WSL2 during initial setup. The app’s environment diagnostics help identify missing prerequisites. Prompt formatting and Turbo also download additional data on first use.

Do I need an internet connection?

Model downloads, licence registration and updates require internet access. Video generation runs on your own GPU. This does not mean every feature is always available offline.

How long does video generation take?

It depends on the GPU, video length, quality settings and reference materials. The app displays an estimated generation time. Initial preparation or restarting the engine takes additional time; no fixed rendering time is guaranteed.

How many PCs can I register with one code?

The current help specifies one PC per activation code. For moving to another PC or re-registration, contact the support channel provided with your purchase.

Where can I find help?

The dedicated help page covers basic operations and common problems. Settings → Support lets you save diagnostic information to send to the support contact provided with your purchase.

MAKE ROOM FOR YOUR IDEAS

Build your own private studio
— free and unlimited.

Download it free and try it for 7 days. You can also check whether this PC can run it.

DownloadCheck my PC

『Tsumugi's🔨WishMallet』 · Windows 11 · Powered by MiniMax H3

© 2026 『Tsumugi's🔨WishMallet』