I Made a Whiteboard Animation Video With AI — Here’s the Exact Process (Free)
I used to think whiteboard-style videos, the ones where a hand draws everything out while a voice tells the story, needed a real illustrator and hours of editing.
Then I actually sat down and tried to make one using nothing but two free AI tools. It took less time than I expected, and it didn’t cost a rupee. Here’s exactly what I did, step by step, so you can copy the same process.
The Two Tools You Need
ChatGPT — for writing the script and the scene prompts.
Flow AI (Google’s Omni Flash model) — for actually generating the animated video clips.
That’s it. No paid plan on either one is required to get started.
Getting the Script and Prompts From ChatGPT
Open ChatGPT and give it a simple instruction: write a whiteboard-animation script broken into short scenes, and generate an AI video prompt for each scene.
You’ve got two paths from here. If you already have a script written, paste it in and ask ChatGPT to turn it into scene prompts. If you don’t have anything written yet, just hand it a topic instead.
I tried this with a topic I found genuinely interesting: “The Taj Mahal isn’t only a love story — it’s also a story about the emperor’s cruelty toward the laborers who built it.”
ChatGPT wrote out a full scene-by-scene script almost instantly, and along with it, gave me a ready-made prompt for each scene, including how long each clip should run. That last detail turns out to matter quite a bit later.
Turning Each Scene Into an Actual Video
Grab the prompt for scene one and head over to Chrome. Search for “Flow AI” — it’s Google’s video generation tool, and it should be the first result. Open it and start a new project.
Set your project up like this:
Choose Video, not Frames — switch the mode to Ingredients.
Pick your aspect ratio: 9:16 if you’re making Shorts or Reels, 16:9 for a regular long-form video.
Set the number of outputs to 1, so you’re not burning through your credits generating variations you don’t need.
For the model, choose Omni Flash. I’ve used Kling and Grok’s video tools before, and honestly, Omni Flash holds up well against both — which is impressive for something free.
Paste in your prompt. One thing worth doing: if ChatGPT says a scene is 6 seconds long, set the actual clip duration to 8 seconds instead of matching it exactly. That small buffer keeps the narration from getting cut off mid-sentence.
Hit generate, then be patient. My first clip stalled at around 6% for a bit — that’s just the tool being in high demand, not a bug. It picked back up on its own after a short wait.
When the clip finished, I was genuinely surprised by the result. The narration lined up with the drawing almost frame-for-frame, right down to small touches like a heart appearing on screen at the exact moment the voiceover mentioned love. Download it, then repeat with each remaining scene — you can actually queue multiple scenes back-to-back instead of generating them one at a time.
Watch Out for This One Detail
Here’s something I only noticed a few scenes in: not every clip generates with the hand-drawing animation visible. My first two scenes had it. A few after that didn’t, which looked slightly inconsistent once I watched them back to back.
The fix is easy. Go back to ChatGPT, tell it which scene is missing the hand, and ask it to revise the prompt to include it. It’ll give you a corrected version — copy that into Flow AI, regenerate, and download. It’s a small extra pass, but it’s worth doing before you move to editing.
Putting It All Together
Open a video editor — I used VN, which is free on both phone and desktop. Start a new project, import all your downloaded clips in the right order, and drop them onto the timeline.
Now, the watermark. You don’t need any special tool to remove it. Add a text layer with something simple, like your channel name, set the font to white with a red background, shrink it down, and place it directly over the watermark. Stretch that text box across the entire video length, and the watermark disappears without any complicated editing.
If you want to make the transitions feel less abrupt, add a short black or white transition between clips and apply it to the whole timeline at once. It’s a small detail, but it makes the final video look more finished.
Export at your editor’s highest quality setting, and it’s ready to publish.
So, Is It Actually Free?
Yes — and here’s why. New Flow AI accounts come with 280 free credits, which is roughly enough for one full video per day. Once those run out, switching to a different email gets you a fresh batch. Not the most elegant system, but it works fine while you’re starting out.
There’s also a Jio SIM offer currently active that adds an extra 1,000 credits if you qualify — I’ve gone into more detail on that separately since it deserves its own explanation.
Mistakes I’d Tell You to Skip
Don’t set your clip length to exactly match the script’s suggested duration — pad it slightly or the voiceover gets clipped.
Don’t skip checking each clip for the hand-drawing animation before moving on — it’s much faster to fix one scene now than to notice it after the whole video is edited together.
Don’t leave the watermark visible — the text-cover trick takes thirty seconds and makes the difference between a finished-looking video and an obviously free one.
What Stood Out to Me Most
The part I keep coming back to isn’t even the video quality, it’s how much this replaced. No animator, no voiceover recording, no manually timing a hand-drawn sketch to match narration. ChatGPT handled the writing, Flow AI handled the animation, and a basic free editor tied it together.
If you’re new to AI video tools, this is one of the more forgiving formats to start with, because the tools are doing most of the technical work. Your job is mostly just directing them.
MASTER PROMPT:
Prompt :-Topic: [ topic/script] Create a complete AI explainer video package in this exact style:
- Pure whiteboard explainer animation style
- Hand-drawn doodle illustrations
- White background only
- Marker drawing animation
- No realistic environment
- No visible realistic human character
- Simple black outline drawings
- Smooth whiteboard animation transitions
- Educational YouTube explainer vibe
- Minimal clean aesthetic
- Only voice-over
- Every scene duration only from: 4 sec, 6 sec, 8 sec, or 10 sec
- For every scene include:
1. Scene duration 2. Detailed AI video prompt 3. Voice-over dialogue
- Keep visuals highly engaging and easy to understand
- Add animated arrows, icons, labels, and doodles
- End with a strong conclusion scene