YouTube Thumbnail Preview — See It in the Real Feed First
Drop your thumbnail draft into pixel-faithful mocks of YouTube's home feed, search results, up-next sidebar, mobile feed and Shorts shelf — at the real display sizes, in dark and light themes — then test it beside up to five real competing videos fetched live from YouTube, and run measured checks on title fit, upload spec and small-size legibility before you publish.
Loading one of your published videos as a variant is the fastest way to test a thumbnail refresh against the real title.
This is the shared title. Loading a live video as a variant, or the button below, gives a variant its own.
Test against real videos
Paste up to 5 YouTube links. Their real thumbnails and titles are pulled from YouTube's public oEmbed/images — view counts and durations are not public there, so none are shown for them.
Layouts use YouTube's typical widths and Roboto type sizes; the live site flexes with window size and experiments, so treat this as a faithful working preview, not a pixel-perfect copy. Neighbour tiles marked as samples are illustrative drawings; fetched videos show their real titles and thumbnails, with an initial-tile avatar — YouTube doesn't expose channel avatar images without its Data API key.
Checks for variant A
Every number below is measured from your actual image and title — spec rules from YouTube's own help pages, text fitting from a wrap model, image statistics from the pixels. They are heuristics for spotting problems early, not a prediction of your click-through rate; no tool outside YouTube Studio can know that.
Title fit — what each surface actually shows
Computed by wrapping your title at each surface's real column width and type size (two-line clamp, like YouTube). Front-load the hook: the tail is what disappears.
| Surface | Visible | What viewers read | Cut |
|---|---|---|---|
| Home — desktop | 100% | I Tested Every Thumbnail Trick So You Don't Have To | nothing |
| Search — desktop | 100% | I Tested Every Thumbnail Trick So You Don't Have To | nothing |
| Up next — sidebar | 100% | I Tested Every Thumbnail Trick So You Don't Have To | nothing |
| Home — mobile app | 100% | I Tested Every Thumbnail Trick So You Don't Have To | nothing |
| Shorts shelf | 94% | I Tested Every Thumbnail Trick So You Don't Have… | 3 chars |
How to Use
- 1Add your thumbnail: drag it in, paste it, or load one of your published videos as a variant slot (A, B or C).
- 2Refine the mock: title, channel name, avatar, duration badge, view-count line and verified badge — variants can carry their own titles.
- 3Paste up to five real YouTube links to pull competing thumbnails and titles into the same feed, then shuffle and compare at real sizes across home, search, up-next, mobile and Shorts.
- 4Read the checks: spec status, the numeric title-fit table, pixel and palette measurements, badge-corner busyness and the shrink strip — then export the mock as a PNG to share.
Key Features
- ✓
Real sizes, real surfaces
Your thumbnail rendered at YouTube's actual display widths — 360px desktop home cards, 360px search results, 168px up-next rows, a full-bleed 390px mobile feed and a 9:16 Shorts shelf — with YouTube's own Roboto type sizes and both the dark and light themes. A thumbnail that only works at full size in your editor doesn't work; this is where that gets found out, privately, before publishing instead of after.
- ✓
Test against real competing videos
Paste up to five YouTube links and those videos' genuine thumbnails and titles join your mock feed, fetched live from YouTube's public oEmbed endpoint and image CDN — no API key, no account, no scraping service. Shuffle the feed, glance at it the way a scroller does, and see which tile actually pulls your eye. Honesty is built in: oEmbed publishes a video's title, channel and thumbnail only, so fetched videos display no view counts or durations — tools that show you precise stats for a pasted link without an API key are inventing them.
- ✓
A numeric title-fit table, not just a visual clamp
Every preview tool shows your title cut off; this one measures the cut. A text-wrap model using each surface's real column width, font size and weight computes the exact percentage of your title that survives the two-line limit on desktop home, search, up-next, mobile and Shorts — plus the precise string viewers will read and the number of characters lost. If your hook's payoff sits at character 55, the sidebar will never show it; the table tells you in seconds, while the fix is still free.
- ✓
Three variants with their own titles
YouTube's Test & Compare validates up to three thumbnails, but only after you publish and only against real traffic. Variant slots A, B and C here each hold their own image and can carry their own title override, so the classic pre-publish questions get answered in the mock feed instead: close-up face or wide shot? Three-word title or six-word title? Does the red version still win when the feed shuffles? Flip variants inside the same crowded feed and the stronger draft usually announces itself.
- ✓
Measured legibility: pixels, palette, badge corner
The checks panel measures your actual image rather than admiring it: mean brightness, RMS contrast, saturation and colourfulness; the five dominant colours with their shares; a detail-retention score that downscales the image and measures how much structure survives at 168px and 120px; and the duration-badge corner enlarged with a busyness score, because art placed under the badge turns both into mush. A shrink strip renders the thumbnail at 360, 246, 168 and 120 pixels — with a grayscale toggle — so 'readable at sidebar size' is something you see, not something you hope.
- ✓
Colour-vision simulation and a face check
One toggle re-renders every thumbnail in the feed through deuteranopia, protanopia or tritanopia simulation matrices, catching the classic failure — a red-versus-green design that flattens into one brown smear for colour-blind viewers — before your audience does. And where the browser ships a built-in face detector (Chromium browsers do), the panel reports how many faces it found in your thumbnail, on-device; where it doesn't, the panel says so plainly instead of pretending a cloud model ran. Nothing about your image is ever uploaded for any of this.
Frequently Asked Questions
What does a thumbnail preview show that my image editor doesn't?
How do I test my thumbnail against a competitor's actual video?
Why don't fetched videos show view counts or durations?
Can I preview a new thumbnail for a video that's already published?
What size and format should a YouTube thumbnail be?
How much of my title will viewers actually see?
Can this tool predict my click-through rate?
Is my thumbnail uploaded to your server?
Your thumbnail will never be seen the way you designed it
Nobody encounters your thumbnail full-screen, alone, in a dark room, the way it looked in your editor while you polished it. They meet it at 168 pixels wide in a sidebar, between a video with two million views and a channel ten times your size. They meet it on a phone at arm's length in daylight, as the fourth tile in a scroll they are already halfway past. They meet it cropped into a vertical Shorts frame it was never composed for. A youtube thumbnail preview exists to close that gap between the design environment and the real one: this tool rebuilds the five surfaces a thumbnail actually lives on — desktop home, search results, the up-next sidebar, the mobile feed and the Shorts shelf — at YouTube's genuine display widths and type sizes, in both themes, and drops your draft into a living feed. The verdict stops being something your analytics deliver in a week. It is on the screen in about four seconds.
Test against real videos, with real data — and no invented numbers
Sample neighbours give instant context, but the decisive experiment is real competition. Paste up to five YouTube links — the videos ranking for your topic, the rival upload from this morning, your own previous episode — and their actual thumbnails and titles are pulled from YouTube's public oEmbed endpoint and image servers and placed beside yours in every mock. Shuffle the feed and scan it cold: whichever tile your eye lands on first is winning that impression, and it is not always yours. One disclosure matters here, because this category is full of quiet fiction: oEmbed returns a video's title, channel name and thumbnail — and nothing else. View counts, durations, like counts and upload dates are not available through it, which is why fetched videos in this tool show no statistics at all. Any preview tool that displays a precise view count for a link you pasted, without asking for a YouTube Data API key, is manufacturing that number. We would rather show you a gap than a guess. The same fetch powers the reverse move: load one of your own published videos as a variant, and you can trial a thumbnail refresh against its real, proven title before touching the live upload.
The title is half the thumbnail — so it gets measured too
Creators obsess over the image and then write a title whose payoff lands at character 60 — in a sidebar column that fits about 40 characters per line. Most preview tools show the truncation and leave the arithmetic to you. This one publishes a title-fit table: for each surface, a wrap model at the real column width and Roboto size computes the exact visible percentage, the verbatim text viewers will read, and the character count that was cut. The pattern it reveals is brutal and useful — search shows nearly everything, home shows most of it, up-next is where long titles go to die. The fix is almost always the same: move the hook to the front, let the tail carry the searchable detail, and re-check the table until every surface clears 90%. Pair that with the YouTube SEO tool for the keyword work and the pair of them cover the click decision from both sides — the promise in the title and the proof in the image.
Spec first, then the measurements rules can't make
The checks panel opens with YouTube's own gates, because failing them is unforced error: 1280×720 recommended resolution with 640px wide as the hard minimum, a true 16:9 frame, under 2 MB for uploads through the YouTube app (50 MB in Studio), JPG/PNG/GIF/BMP format, titles within the 100-character limit and ideally near 60. Then it measures what no rulebook can. Pixel statistics — mean brightness, RMS contrast, saturation, colourfulness and the dominant palette — describe the image the way a feed full of rivals will experience it: a dark, low-contrast thumbnail beside three bright ones is a tile people scroll past without ever consciously seeing. Detail retention downscales your image and measures how much edge structure survives at 168px and 120px; fine textures, thin outlines and small text evaporate at exactly those sizes, and the score notices before your click-through rate does. The badge-corner zoom enlarges the precise patch the duration badge will cover and scores its busyness — put a face, a logo or your third word there and both the art and the badge become illegible. A readiness score rolls the five measurement families into one number with every input published on the page. Treat it as a pre-flight checklist, not an oracle: it is a heuristic summary of measurable properties, and it is labelled that way, because no tool outside YouTube Studio can predict your CTR — anyone selling that prediction is selling a horoscope with a dashboard.
Variants, vision, and the workflow around the preview
Three variant slots mirror how YouTube's own Test & Compare thinks — A, B, C, each with its own image and an optional title override — moved to before the publish button, where changes are free. The colour-vision toggle re-renders the whole feed through deuteranopia, protanopia and tritanopia matrices: roughly one in twelve male viewers has some colour-vision deficiency, and the classic red-subject-on-green-background composition collapses for them into a single brown shape; thirty seconds with the toggle finds that before they do. When a subject sits off-centre for the vertical frame, fix the crop with the smart image cropper; when you are studying a rival's published art directly, pull the full-resolution file with the YouTube thumbnail downloader; and when the video is live and earning, the creator earnings calculator turns the views the thumbnail won into a revenue estimate. The preview itself exports as a PNG — the whole stage, theme included — for a client, a collaborator or your future self comparing drafts.
Private by architecture
An unpublished thumbnail is a trade secret — it announces your topic, your angle and your launch timing to anyone who sees it. So nothing here is uploaded. Your images are drawn to a canvas in your own browser and every measurement runs locally; the only network requests in the tool happen when you explicitly fetch a public video, and those go from your browser directly to YouTube's public endpoints. There is no account, no queue, no watermark, no weekly credit and no premium tier — the complete tool, installable to your home screen like an app, is free. A preview tool that demanded your unreleased artwork on its server before showing you a mock feed would be asking for more trust than a preview is worth.
Related Tools You Might Like
Handy utilities in Social Media Tools
YouTube Thumbnail Downloader
Download high-resolution YouTube video thumbnails in HD, 1080p, and 4K quality for free. Fast, direct downloads by video URL with zero ads.
YouTube SEO Tool
Free YouTube SEO tool: score titles, descriptions and tags with pixel-true title fit, chapter validation and keyword coverage — rubric published, no invented data.
Creator Earnings Calculator
Estimate YouTube video and Shorts earnings from views, RPM, and niche category CPM. Calculate potential creator income with our free online tool.