Four dollars for an explainer video is a great deal. Having a model write JavaScript canvas code that renders directly to MP4 completely sidesteps expensive, finicky video generation models. Every graphic element and timeline cue stays aligned through code. When you want to revise something, you tweak parameters directly without rerolling generations to test your luck.
Paying roughly $4 with delivery under two hours shows why this math works. Turning video creation into running code makes your spending obvious and your output stable. For structured content, I think generating assets with code is far more dependable than brute-forcing raw pixels.