What does the free version give me?
How does it decide what goes in the picture?
Your title goes to a language model that writes two short scene descriptions — a setting, a mood, a lighting direction — and each description is then rendered as a 16:9 image. The image model is deliberately instructed to put no words or letters in the frame, because its only job here is the backdrop. Your headline is added afterwards, as real type.
Why is the wording sharp instead of the garbled lettering AI images usually produce?
Because no image model touches the wording. Once the backdrop arrives, your title is drawn over it in your browser on a 1280 × 720 canvas in a real typeface — heavy weight, uppercase, white with a dark outline so it survives a busy photograph. That split is the whole trick: the model does the picture it is good at, and the text stays type.
Why two designs rather than ten?
The two are genuinely different reading patterns, not two crops of one idea. Center Bold darkens the whole frame slightly and sets the headline across the middle, which suits a single strong subject. Bottom Left keeps the top of the frame clear and drops the headline into the lower left over a gradient, which suits a face or a product that needs room. Ten variants of one layout would give a client nothing to choose between.
What is the Topic / Context field for, if the title is already there?
A title is often ambiguous on its own — "The Real Cost" could be a finance explainer or a documentary about fast fashion. The context field is passed separately to the concept writer, so it steers the setting without ever appearing in the image. If a client's video sits in a specific world — a clinic, a warehouse, a kitchen, a construction site — that is what belongs in this field.
What actually makes a thumbnail work on a client's channel?
One subject, read in a glance, at the smallest size it will ever appear. Most of the audience meets it as a small rectangle in a feed on a phone, not the large preview you approve it in. So: a single focal point rather than a collage, a clear tonal break between subject and background, and a headline short enough to stay large. If it only reads at full size, it does not read.
Should the headline on the thumbnail repeat the video title?
Usually not. YouTube already prints the title beside the thumbnail, so repeating it word for word spends the picture on something the viewer can read anyway. The stronger move is to let the two halves do different jobs — the title carries the subject, the thumbnail carries the tension, the number or the face. This tool draws whatever sits in the title field, so type the short thumbnail line there, not the full video title.
Why keep the bottom-right corner clear?
YouTube stamps the video duration over the bottom right of every thumbnail, and on a video the viewer has already started, a progress bar sits along the bottom edge as well. Anything placed there is partly covered. The Bottom Left layout puts the headline on the opposite side for exactly that reason; with Center Bold, keep faces and fine detail out of that corner when you choose between the two.
How do we keep thumbnails consistent across a client's channel?
Pick one of the two layouts per client and stay on it. A channel where every thumbnail shares a position, a tonal range and a headline length becomes recognisable in the feed before the words are read — worth more than each thumbnail being individually clever. Agencies running several channels at once feel this most, because inconsistency reaches them as a client asking why this week looks different from last week.
Can I move the text or change its colour?
Not in this tool. The two layouts are fixed, and the type is always white uppercase with a dark outline; the size fits itself between roughly 80 and 36 pixels and wraps to at most three lines, so a long headline shrinks rather than spilling off the frame. To change the words, edit the title field and run it again. Control over typeface, position and colour lives in the paid editor.
What size is the file, and what do I do with it once it downloads?
Every thumbnail is 1280 × 720 pixels — the exact size YouTube recommends — and it downloads as a PNG, ready to upload as the custom thumbnail on the client's channel with no conversion step. Keep the file rather than the browser tab: swapping the thumbnail on an already-published video is one of the few things YouTube still lets you change afterwards.
Can I use these on a client's channel?
Yes. The thumbnails you generate are yours to download and use however you like — no credit required, and nothing here is licensed per channel or per seat, so client work is fine on the free run as well.
Is the thumbnail all this does, or is there a video product behind it?
Will a client see that we used a tool?