Copy the free prompts below. Or let the Flat 2D builder draw and edit the whole video with code, on your computer. No image credits.
Show me the prompts → Prompts free · the builder is $37 one time · Flat 2D onlyEvery frame drawn and edited by code. No image credits, no clip credits.
Paste one into Claude and it draws the picture with code. No image credits.
Draw one picture in a Flat 2D explainer illustration style, as code. Subject: [a lighthouse keeper on a stone pier at dusk, holding a lantern, a ship on rough sea behind him]. LOOK: every shape is ONE flat base colour built three ways: a soft lit gradient (about 14 percent lighter at the top left, 16 percent darker at the bottom right), a hard edged darker shadow band on the side away from the light, and a thin lighter rim on the side toward it. Each whole figure or big object gets ONE outline in #3B2626, 3 px, drawn round the outside only. Walls and floors are drawn in perspective with a few panel or plank lines. Finish the whole frame with a very light warm grain (tiny random dots at 4 percent opacity) and a soft vignette. No textures, no photos, no realistic shading, no thick black cartoon outlines. PALETTE: peach wall #F2C7A0, terracotta #C8613F, mustard #E8B04A, teal #2F7F86, coral red #E2553F, slate #2E3A4A, wood #C9955C, cream #F5E6D0, sky #9FD3E6 to #F6E1C2. PEOPLE: one friendly puppet design for everyone: about 7.5 heads tall, a round head, a big simple face (two dark dot eyes with a tiny white glint, short eyebrows that show the mood, a small curved mouth, soft blush ovals), a rounded body, long thin arms that bend at the elbow, mitten hands with a separate thumb and two short finger lines, slim legs, rounded shoes. Men: short cropped hair, a squarer torso. Women: hair to the shoulder or a bun. Generic light skin #F2C4A4 with its shadow band #DDA889 unless the subject says otherwise. LAYOUT: a medium wide shot. The main person stands in the near ground on the left or right third, about 45 percent of the frame height, turned three quarters toward the camera, with one clear action pose. The place is dressed in three layers: big props beside the person in the near ground, the main setting in the middle ground, small far shapes (hills, buildings, a coast) near the horizon line at about 58 percent of the height. Water is layered wave bands with small white crest strokes, never cloud shapes. Light comes from the upper left; glowing things (lamps, windows, a lantern) get a soft round glow. BUILD: one single self contained HTML file with a 1920 by 1080 canvas drawn with the 2D canvas API (no libraries, no images, no external files, everything drawn with code), page margin 0, the canvas scaled to fit the window at 16:9. Write small helper functions first (for example person(), shadow(), outline()), then draw back to front: sky, far, middle, near, people, foreground, finish. Use a seeded random number function so the picture is the same every time. SIZE RULES: decide the ground line first. Every person and thing gets its real height in metres and is drawn at one scale for its depth (a standing adult is 1.7 m, a door 2.1 m, a table 0.75 m, a car 1.5 m tall, a tree 6 to 10 m, a house 7 m): things farther away are smaller and higher on the screen, never bigger than a nearer person unless they are really bigger. Every foot stands on its ground line with a small soft contact shadow. Compute each hand position first and draw anything held so its handle sits exactly on that hand. Nothing passes through anything else. CHECK BEFORE YOU ANSWER: at the top of the script write a short comment table: each person and thing, its real height, its pixel height and where it stands. Fix anything whose size is wrong for its depth, any floating or cut off feet, any held thing not touching a hand, anything important cut by the frame edge, and any face covered. The subject in the brackets must be obvious at first glance. Output the HTML file only, shown as an artifact, with no explanation.
Draw one picture in a Flat 2D explainer illustration style, as code. Subject: [a baker handing a loaf to a customer at a market stall, bunting overhead]. LOOK: every shape is ONE flat base colour built three ways: a soft lit gradient (about 14 percent lighter at the top left, 16 percent darker at the bottom right), a hard edged darker shadow band on the side away from the light, and a thin lighter rim on the side toward it. Each whole figure or big object gets ONE outline in #3B2626, 3 px, drawn round the outside only. Walls and floors are drawn in perspective with a few panel or plank lines. Finish the whole frame with a very light warm grain (tiny random dots at 4 percent opacity) and a soft vignette. No textures, no photos, no realistic shading, no thick black cartoon outlines. PALETTE: peach wall #F2C7A0, terracotta #C8613F, mustard #E8B04A, teal #2F7F86, coral red #E2553F, slate #2E3A4A, wood #C9955C, cream #F5E6D0, sky #9FD3E6 to #F6E1C2. PEOPLE: one friendly puppet design for everyone: about 7.5 heads tall, a round head, a big simple face (two dark dot eyes with a tiny white glint, short eyebrows that show the mood, a small curved mouth, soft blush ovals), a rounded body, long thin arms that bend at the elbow, mitten hands with a separate thumb and two short finger lines, slim legs, rounded shoes. Men: short cropped hair, a squarer torso. Women: hair to the shoulder or a bun. Generic light skin #F2C4A4 with its shadow band #DDA889 unless the subject says otherwise. LAYOUT: a medium shot where two people share the frame, each about 50 percent of the frame height, facing each other across the main prop (a counter, a table, a stall) which stands between them at waist height (about 0.9 m). The thing being handed over is held in the giver's hand and touched by the receiver's reaching hand at the same point. Behind them a dressed background (shelves, crates, plants, signs, bunting, other stalls smaller and farther away), in front a few props overlapping the bottom edge. Sunny high key light from the upper left. BUILD: one single self contained HTML file with a 1920 by 1080 canvas drawn with the 2D canvas API (no libraries, no images, no external files, everything drawn with code), page margin 0, the canvas scaled to fit the window at 16:9. Write small helper functions first (for example person(), shadow(), outline()), then draw back to front: sky, far, middle, near, people, foreground, finish. Use a seeded random number function so the picture is the same every time. SIZE RULES: decide the ground line first. Every person and thing gets its real height in metres and is drawn at one scale for its depth (a standing adult is 1.7 m, a door 2.1 m, a table 0.75 m, a car 1.5 m tall, a tree 6 to 10 m, a house 7 m): things farther away are smaller and higher on the screen, never bigger than a nearer person unless they are really bigger. Every foot stands on its ground line with a small soft contact shadow. Compute each hand position first and draw anything held so its handle sits exactly on that hand. Nothing passes through anything else. CHECK BEFORE YOU ANSWER: at the top of the script write a short comment table: each person and thing, its real height, its pixel height and where it stands. Fix anything whose size is wrong for its depth, any floating or cut off feet, any held thing not touching a hand, anything important cut by the frame edge, and any face covered. The subject in the brackets must be obvious at first glance. Output the HTML file only, shown as an artifact, with no explanation.


Swap the part in [square brackets] for your own idea before you paste. Free to use, as often as you like. Want the whole video, moving and timed to your voice? That's the builder.
No AI video tool, no image tool. Code draws every frame on your own computer.
Use the prompts as much as you like. The builder is for when you want the whole video in Flat 2D.
This is the builder inside Claude, on your computer.
An example run. Your shot count depends on your script.
You know the bill: every picture costs credits, and every redo costs again. Code draws them for nothing.
You want your videos in Flat 2D, the same every time, without learning animation.
Give it the script and the voiceover and get the edited video back.
For Flat 2D videos in 16:9, made on your own computer.
Every scene is drawn on your own computer, so there's nothing to pay per picture, per redo or per clip.
The look, the colours, the camera moves and the pace stay the same, so your channel looks like one channel.
Use it on every video you make in that style. It adds up from video two.
A one minute video can need 15 to 30 pictures. Then the redos when a face changes or a scene comes out wrong, and every redo costs again.
Then the evening lining it all up on a timeline. It works. It's just paid for picture by picture.
"How many credits does one video take?"
"My character looks different in every scene."
"Can I do it cheaper?"
Every scene plus the redos, all paid for. Then an evening lining them up.
Code draws every scene and puts the video together. Nothing to pay per picture.
The builder, locked to Flat 2D. It makes videos in that one look, every time. Ask Claude to set it up and it installs everything itself.
A built-in voice with 12 voices to pick from. No voiceover yet? It reads your script out loud with a voice built into the builder, so you need no voice subscription. Your own recording works too.
Background music and sound effects, mixed in for you. A soft music bed picked from your script's feeling, plus small sounds on cuts, signs and footsteps. Made by code on your computer, nothing to license. Ask for another mood, quieter music or none at all.
Finished, edited videos from your script. Timed to your voiceover, at the style's own pace.
No pictures to make at all. Code draws every scene in your style, so there is nothing to generate.
A clean video. No captions burned in, so you decide on captions yourself.
A short setup video. Install and first video, start to finish.
Help by email. If it won't run on your computer, reply to your delivery email and I'll help you get it working.
Updates included: when I improve the look, you get the new version at no extra cost. Same place you got it, nothing to buy again.
I run Monsguide on YouTube, where I show people how to make videos with AI. I build tools like this one and use them on camera before anyone gets them.
Describe your character once: who they are, what they wear, the one thing people remember. Paste it in and the builder draws them the same in every video.
Works for you, a mascot, or the people in your story.
Included free with your purchase.
One style, locked in. Use it on every video you make in this style.
You land on your delivery page right after you pay: the download, the setup steps and the tutorial. The link comes by email too, and it always brings you back.
Updates included: when I improve the look, you get the new version at no extra cost.
Digital product. All sales are final. Needs a computer and a paid Claude plan.
Want a different style? View all styles →
One style, $37, one time.
You land on it right after you pay, and the link comes by email. Download the builder and watch the short tutorial.
In the Claude desktop app. Then give it your script and voiceover. No voiceover yet? It reads your script out loud with its built-in voice. Your video lands in a folder.
Your delivery page opens right after you pay.
If it won't run on your computer, reply to your delivery email and I'll help you get it working.
Digital product. All sales are final.
Views, subscribers and income depend on YouTube and on you, so nobody can honestly sell you those.
Starting with the big one: will it work for me?
Your script, your voice, Flat 2D, drawn with code. Every video.
I need this. I'm in, Monique! → Prompts free · the builder is $37 one time · Flat 2D onlyWhen you fill out a form on this site, you give us your contact information so we can deliver what you asked for. You'll also receive emails from Monique and TuberAlchemy with steps, tools, and useful, relevant offers. By submitting a form, you consent to receiving these emails. Every email includes an unsubscribe link.
If you buy something from us, like a custom skill build, we also use your information to take your payment, build what you ordered, and deliver it.
From our forms and surveys: your name, your email address, and your answers to any questions we ask.
When you buy: your name, email address and billing details such as your country. Card payments are handled by Stripe, our payment provider. We never see or store your full card number.
For a custom build: what you send us in the build form, such as channel links, reference images, thumbnails, and notes about your voice and style.
When you visit: like most websites, basic visit information (pages viewed, device type), and which link or email brought you here, so we know what's useful.
To deliver what you asked for, send you our emails, take payments, build and deliver your order, give you access to anything included with it (for example a personal Video Sync access code), and improve what we build.
Trusted service providers process this information on our behalf to make that possible: our email and customer platform, Stripe for payments, our website and tool hosting, and Skool for the community. They act on our instructions and for no other purpose.
Anything you send us for a custom build is used only to build your skill. It is not given to anyone else, not published, and not used in a tutorial. We will never feature your channel without asking you first.
Our pages use the Meta pixel, a small piece of code from the company behind Facebook and Instagram. It tells us anonymously that someone visited a page or filled out a form, so we can show relevant ads to people who have already been here. We do not use it to send them your name or your email address.
If you would rather not be included, you can turn off ad personalization in your Facebook or Instagram settings, or block tracking in your browser. Either way, it does not affect anything you have asked us to send you.
Our community runs on Skool, which is a separate service with its own privacy policy covering anything that happens once you are inside it.
We keep your contact information while you're on our list or a customer. Purchase records are kept for as long as tax law requires, which in Canada is generally six years. Build files are kept so we can handle your round of changes and any questions, and you can ask us to delete them at any time after your build is delivered.
TuberAlchemy is run from Canada. Some of our service providers store or process information in the United States and other countries.
We do not sell your information. We do not rent it. We do not share it with anyone for their own marketing.
Every email includes an unsubscribe link, and it takes effect immediately. You can also ask us for a copy of the information we hold about you, ask us to correct it, or ask us to delete it entirely. Reply to any of our emails or contact us at the address below and we'll take care of it. We may need to keep purchase records that the law requires us to keep.
For any questions about this policy, email us at monsguide@mail.tuberalchemy.com.
If this policy changes, the current version will always be posted on this page, with the date above updated to match.
© 2026 TuberAlchemy. All rights reserved.Flat 2D Code Video Builder
The Flat 2D Code Video Builder and a setup tutorial, on a private delivery page.
A computer and a paid Claude plan. Works for looks code can draw: flat 2D, stick figures, maps, infographics, paper documentary and motion text.
This is a digital product and all sales are final. If it won't run on your computer, reply to your delivery email and I'll help you get it working.
Views, watch time, subscribers, revenue and monetisation are not promised or covered.
For your own channels. Not to be resold, shared, redistributed or repackaged.