Watch the video with chapters & transcript →
AI TOOLS
In short: A video made in Canva and set as the HeyGen background replaces the plain white canvas, giving the Short a produced look without any editing software.
This workflow builds a complete avatar-led Short inside HeyGen: your own avatar, a script drafted with ChatGPT, a Canva-made background, your cloned voice and styled captions — assembled in the browser and rendered in a few minutes. Ani Björkström, a tech consultant, demonstrated it by producing a Short about Nano Banana ready for YouTube or TikTok.
Choose portrait orientation, then replace the default avatar with one of your own from "my avatars". One framing caveat: if you are using Avatar IV, the framing cannot be changed; turn Avatar IV off and you can pick between normal framing, a closer look, or a circle — the circle was the choice for this video, leaving room on screen for the background content.
The script is written outside HeyGen and pasted in — here it was drafted with ChatGPT on the topic of Nano Banana, then proofread before continuing. The background comes from the media panel: HeyGen offers a large library of free pictures and videos, or you can upload your own media. In this case the background was a short video created in Canva, uploaded and applied with "set as background" so the avatar sits over custom visuals instead of a white void.
Preview before generating, because the preview shows the runtime. The first pass here was 1 minute 7 seconds — over the one-minute ceiling for Shorts — so the fix was deleting part of the script text, which brought the video to a couple of seconds under a minute. Trimming words is faster than trimming footage: the avatar, voice and background all adjust automatically to the shorter script.
The voice attached by default was a stock voice that happened to be named Annie's voice — not the creator's own — so it was replaced under "my voices" with a voice imported from ElevenLabs. Captions are added from the captions menu; a blue-and-black style was chosen here. Then click generate, name the project, and submit.
| Production step | Setting used in the tutorial |
|---|---|
| Orientation | Portrait |
| Avatar framing | Circle (Avatar IV turned off) |
| Script | Drafted in ChatGPT, pasted and proofread |
| Background | Custom video made in Canva, set as background |
| Voice | Own voice imported from ElevenLabs |
| Captions | Blue-and-black caption style |
Rendering normally takes up to five or six minutes, and HeyGen emails you as soon as the video is created. From there the file is ready to upload to YouTube or TikTok.
No. The avatar, script, background, voice and captions are all assembled inside HeyGen; the only outside tools used were ChatGPT for the script and Canva for the background video.
Normally up to five or six minutes. You receive an email as soon as the video is ready, so you do not need to wait in the browser.
Yes. The media panel includes many free pictures and videos you can set as the background — uploading your own Canva creation is optional.
Read the full video transcript, with chapters, on the video page →
Want this working inside your finance team?