A few months ago I kept running into the same problem as a content creator: I'd spend 1–2 hours in Canva or Photoshop on a thumbnail, upload the video, and watch it underperform.
The content wasn't the issue. The thumbnail was.
CTR (click-through rate) is the single biggest lever on YouTube reach — and it's almost entirely decided in the first 2 seconds someone sees your thumbnail in their feed. Yet most creators treat it as an afterthought.
So I built Miniagen.
— What it does —
Paste a YouTube URL or describe your video idea in text. Miniagen generates multiple thumbnail variations with AI. You pick the one that feels right, refine it, and export at 4K.
Three main flows:
· From a YouTube URL — it pulls your video context and generates thumbnail concepts around it
· From a text prompt — start from scratch if you don't have a video yet
· Edit an existing thumbnail — upload yours and improve it with AI
You also create a reusable avatar once. It stays consistent across all your generations so your channel looks like a brand, not a random collection of images.
— The hard part —
Getting multiple genuinely different layout variations (not just color-swapped versions of the same thing) took most of the iteration time. Creators need real options to A/B test, not the same thumbnail five times.
Avatar consistency across generations without fine-tuning per user was the other challenge. Still improving it.
— Where it stands —
1,200+ creators using it. YouTubers, short-form creators, agencies managing multiple channels.
Early feedback points to two things people care about most: speed (under 60 seconds from idea to export) and having real variations to test.
— Try it —
Free to start, no credit card needed. Would genuinely love feedback from anyone building content at scale — especially on what's missing.
This comment was deleted a month ago.
stop spamming your domain name they are rubbish otherwise I will report you