How I Made This Viral Stop-Motion Animation (Step by Step Full Tutorial)

Handmade looking animations like this can stop people scrolling and gain millions of views, but you don't need to cut out hundreds of paper pieces to make one.

In this tutorial, I'll show you how I built this full paper cutout animation inside Love Art starting with AI generated images, editing the design

directly on the canvas, and then turning those scenes into a finished animated short. Let's get into it.

short. Let's get into it.

The link in the description will bring you straight to the Love Art homepage.

Once you've signed up, you can start [music] creating straight away. For this

project, I'm going to select GPT image two [music] from the model picker here.

And below that, Love Art gives you a selection of prompting templates, [music] which are really useful if you need a bit of inspiration or if you want to take one of these, [music]

customize it, and use it as a starting point for your own project. But for this animation, we're going to start [music] by creating a simple opening title scene that establishes the visual style for

the full piece. [music] So, instead of starting with a complicated prompt, I'm asking for a wide paper cutout scene with a textured background, colorful [music] handmade shapes, arrows,

mountains plants stars abstract details, a small paper [music] cut character, and the title accessing flow state in the center. These are some of the visual elements I plan to reuse

throughout the full composition. So,

this first image gives us a strong foundation to build from. As you can see, once the image is generated, it [music] opens directly inside the Love Art canvas. And this is where Love Art

Art canvas. And this is where Love Art starts to feel [music] very different from a standard AI image generator because the image isn't just a final output. It becomes something we can keep

output. It becomes something we can keep working with. [music] From here, we can

working with. [music] From here, we can speak directly to the agent, organize the project on the canvas, [music] make edits, generate new scenes, and eventually turn those scenes into video.

So, [music] let's open the chat and configure a few things before we continue. Initially, I'm going to make

continue. Initially, I'm going to make sure this section is set to agent, [music] which allows Love Art to dynamically generate results based on the prompt, but I'm also going to lock in the models

I want to use for this project. For

images, I'll be using GPT image 2, and for video generation later on, I'll be using SeaArt 2.0. Now,

[music] before we move on, there's one change I want to make to this opening image. I

prefer the word activating [music] instead of accessing, so I'm going to click on the image and select the edit text function. As you can see, Love Art

text function. As you can see, Love Art identifies the text inside the image, which means I can change the wording directly and rerun the edit [music] without having to rebuild the whole

image from scratch. And there we go, same visual style, but with the updated title. [music] Now, let's move on to the

title. [music] Now, let's move on to the next scene. I'll open the agent chat

next scene. I'll open the agent chat again, [music] and the first line I'm going to add is really important. Take

style reference from Then I can click on the opening image, [music] and Love Art will reference that image inside the chat. This means the agent can use it as

chat. This means the agent can use it as visual context for the next [music] generation. From here, I can describe

generation. From here, I can describe the next scene I want to create. For

example [music] create a 16:9 image of a dark gray paper cut figure sitting at a desk looking overwhelmed. Around [music] the

overwhelmed. Around [music] the character, include scattered mobile phones, clocks, thought bubbles, and other visual symbols of distraction. The

scene should show the character trying to work while surrounded by noise and interruption, while keeping the same handmade paper cut out style [music] as the reference image. As you can see, the

agent analyses the image we referenced [music] and continues the generation using GPT image 2. Once the image has generated, you can make adjustments if needed. I found [music] the quick edit

needed. I found [music] the quick edit function really handy here. For example,

you could ask [music] Love Art to zoom out slightly from the character, so we can see more of the full body sitting at the desk.

And this [music] is the benefit of having the canvas and agent working together. You're not just generating

together. You're not just generating isolated images, you're building a connected [music] sequence and refining it as you go. So,

now that we're creating multiple scenes, it's a good idea to start organizing everything. You can hold shift on your

everything. You can hold shift on your keyboard, select the images you want to keep, and then [music] set them to auto layout. This is one of the features I

layout. This is one of the features I really liked, because while you're [music] still iterating on the individual frames, you can keep everything organized as a rough timeline [music] for the final composition. You can even

press play to get a quick sense of how the sequence [music] is starting to flow before we move into animation. At this

point, it's a good idea to create a skill [music] based on the actions we have just taken. Create a new chat, turn thinking mode on, and repeat [music] the same process of generating an image.

Once you've completed the task, [music] create a skill based on the actions just taken.

Now we have a repeatable skill that can be accessed [music] whenever we want to create this style of image.

From here, continue the same process until you have all the key frames for your final video. Now starts the fun [music] bit. Let's start animating these

[music] bit. Let's start animating these scenes. For this section, I'm going to

scenes. For this section, I'm going to switch from agent [music] to the video generator, and then change the model to Seedance 2.0. The reason I'm switching

Seedance 2.0. The reason I'm switching into the video generator is [music] that it gives me more control over how the final animation is created. Inside the

settings, I'm going to choose frames, which allows [music] me to set a start frame and an end frame for the generation. I can also choose the aspect

generation. I can also choose the aspect ratio [music] resolution, and duration in seconds. For

this project, I kept everything in 16:9, [music] and for the audio settings, I set audio to on. One thing I found really

to on. One thing I found really important was to [music] tell the model not to add music or voice over.

Otherwise, you can end up with different clips having random voices or background music [music] which doesn't really work when you're stitching the full sequence together.

So, for the first transition, [music] I'm going to select the start frame, which in this case is the blank canvas, [music] and then select the end frame, which is the opening title scene. Then

in the prompt, I'll describe [music] how I want the animation to transition.

For consistency across the full composition, I'm asking [music] for a clear handmade stop motion effect around 8 to 12 frames per second with tiny

wiggles and slight rotations as the paper elements [music] settle into place. For the audio, I'm keeping it

place. For the audio, I'm keeping it simple. No music and no voice [music]

simple. No music and no voice [music] over. Instead, I'm asking for subtle

over. Instead, I'm asking for subtle paper sliding, small paper taps, gentle [music] cardboard wobble sounds, and soft handmade stop motion movement sounds. [music] For your first

sounds. [music] For your first generation, I'd recommend creating it in a lower resolution first, just to check that the concept is working and the model understands the type of motion

you're looking for. Regarding

resolution, you have two options. You

can generate in a lower resolution and upscale [music] the video afterwards.

But for this particular style, I actually found that upscaling didn't work as well because it made everything look a little too clean and polished, [music] which took away from that rough

textured paper aesthetic. So, what I preferred to do was generate [music] the first version in 480p to test the idea.

Then once I was happy with the result, [music] use the remix function and switch the resolution to 1080p, which is the highest option I could get from SeaDance [music] 2.0. For the next

generation, it's important that the start frame is the same as [music] the end frame from the previous generation.

That helps make the transition between clips feel as seamless as possible.

[music] Then, you just continue this process until you've animated the full composition. And eventually, you should

composition. And eventually, you should [music] have a frame timeline that looks something like this.

Once you're happy with everything, you can download the videos and bring them into your video editor.

Now, one thing I really liked from the original design was the cutout style font. So, I used font for future

font. So, I used font for future projects. Just open font generator from

projects. Just open font generator from the bottom toolbar, [music] select the image from your canvas, and Love Art will generate a reusable font based on

that [music] style.

Love Art is also currently offering 45% off when you sign up using the link [music] in the description. So, if you want to try this workflow for yourself, you'll find the link below. Now, [music]

let's take a look at the full short film we've made. Flow state is that rare

we've made. Flow state is that rare feeling where the work stops feeling like well, work.

At first, most creative work feels noisy.

But flow usually starts when the noise is reduced to one clear task.

It happens when the challenge is hard enough to hold your attention, [music] but not so hard that it overwhelms you.

When those two things line up, your attention starts to lock [music] in.

You stop watching the clock.

You stop judging every move.

You just follow the next step.

And the more you move, the easier it becomes to keep moving.

So, accessing flow is not about waiting for motivation.

It is about designing the conditions where focus has somewhere to go.

And if you'd like to open the door to creating sound designs manually, this is where I'll show you how.

Watch this video for the full tutorial and start building your own flow one frame at a time.