01 · Select
Start with a still or video
Use a local image, or scrub through a local video and capture the frame that best communicates the subject.
Private browser tool
Make scroll-stopping thumbnails from your own media
Import an image or capture a precise video frame, add focused copy, and export an exact-size PNG for YouTube, Shorts, Instagram or Facebook. Media processing stays on your device.
Used only for copy ideas. Your media is never sent to the model.
Optional: Optional local AI is not loaded. The first model load is about 376 MB on first use and requires WebGPU. It runs in a worker and is not required for editing.
Accent colour
YouTube video
Export: 1280 × 720 PNG
Choose a preset, then add your own image or video frame.
Platform dimensions included
Choose the placement before you design. The editor changes the working canvas and the exported file together, so a portrait cover is not merely cropped from a landscape preview.
| Placement | Export size | Ratio |
|---|---|---|
| YouTube video | 1280 × 720 | 16:9 |
| YouTube Shorts | 1080 × 1920 | 9:16 |
| Instagram Reel / Story | 1080 × 1920 | 9:16 |
| Instagram feed portrait | 1080 × 1350 | 4:5 |
| Facebook video | 1280 × 720 | 16:9 |
02 · Compose
Keep one clear promise
Edit the headline, kicker, colour, alignment and shading. Short, specific language remains readable on small screens.
03 · Export
Download the real dimensions
The PNG is rendered at the preset resolution—not the smaller responsive size shown on your screen.
What the local AI does—and does not do
It does
Turn your prompt into a concise headline, supporting kicker and accent-colour suggestion. After download, inference runs on your GPU in a dedicated browser worker.
It does not
Generate or inspect your media, guarantee clicks, or replace a human check for accuracy and brand fit. It is optional and never starts automatically.
Frequently asked questions
Does NexxiqAI upload my image or video?
No. The canvas editor, video-frame capture and PNG export run inside your browser. Your selected media is not sent to NexxiqAI.
Is the AI really free?
The optional copy assistant uses an open-source model in your browser. It needs a compatible WebGPU browser and a one-time model download of roughly 376 MB. The editor itself needs neither AI nor a model download.
Does the AI generate an image?
No. It suggests a short headline, kicker and accent colour. You provide the image or capture a frame from your own video, then the browser composes the finished thumbnail.
What image does the download contain?
The exported PNG uses the exact pixel dimensions shown in the selected preset, including the background, overlay, headline, kicker and optional logo.