To make a podcast thumbnail for YouTube, decide who the draw is for this episode (you, or the guest's name), pair that with one short hook, and put both on a 16:9 image at 1280 by 720 or larger. In WThumb that is a link or a reference image, up to three lines of words, your own photo or none, and a wait of about thirty seconds.
The part that trips most shows is not the first thumbnail but the fiftieth. A weekly podcast needs one every week, often with a different guest and a Shorts clip on top, and the guest is someone whose photo you do not own. So the useful question is how to make a podcast thumbnail you can repeat.
This post is that repeatable version, built around one input choice: WThumb accepts one optional personal photo as your identity input for a fresh composition. For a guest-led episode, use the guest's name as the main text. Without a personal photo, WThumb supplies no replacement photo and instructs the model to remove people. Review the finished image before publishing.
Use the podcast thumbnail maker when you are ready to turn that episode hook into a fresh 16:9 thumbnail or a vertical clip composition.
How to make a podcast thumbnail for YouTube
The whole run for one episode, in order; the sections after this explain each choice.
- Pick a reference. Paste the YouTube link of a podcast thumbnail you like from a show in your area, or upload a reference image (PNG, JPEG or WebP, up to 12 MiB). WThumb fetches only that video's public thumbnail.
- Type the words. Up to three lines: the guest's name (or the show's name) on one line and a hook of a few words on another. Instructions in plain words are optional, and you can attach up to four extra images and point at them as @1 to @4.
- Add your photo, or do not. A personal photo (up to 8 MiB) supplies identity input for a fresh composition. Skip it to request a faceless design: WThumb supplies no replacement photo and instructs the model to remove people. Review the finished image for unwanted people or details before publishing.
- Tick 16:9, 9:16 or both. 16:9 is the episode thumbnail, 9:16 is for the Shorts clip. Tick both and one brief produces both.
- Wait, then check the words. A run takes about thirty seconds. Read every letter of the guest's name before you upload. If one thing is wrong, press the pencil on the thumbnail and describe the change, and WThumb makes a new version, for the cost of one credit per revision (1 credit = 1 thumbnail).
Output is a high-quality Full HD+ JPEG at 2048 by 1152 for 16:9 (1152 by 2048 for 9:16). As of September 2026, YouTube's Add custom thumbnails help page asks for a 16:9 image with a minimum width of 640 pixels and recommends 3840 by 2160, so WThumb's file is YouTube's shape, above the minimum and below the recommendation. It uploads without issue and is sharp at the sizes YouTube serves thumbnails at; it is not the 4K-sized file the page names. The same page also says a channel can only upload custom thumbnails once its account is verified, and YouTube's Verify your YouTube account page does that with a code sent to a phone number.
One thing WThumb does not make: as of 2026, YouTube's Create a podcast in YouTube Studio page says a podcast set up in YouTube Studio has its own square thumbnail, recommended at 1280 by 1280, uploaded on the podcast's Details page. That is the show's cover, separate from each episode's thumbnail, and a square you make elsewhere.
Guest or host: who goes on the thumbnail?
Before the words, decide who the draw is: not who is on the episode, but who makes a stranger stop scrolling. Three honest answers, each a different brief.
A solo or host-led show
If people come for you, your face is the constant and the hook is the variable. Add your photo every time, make the hook the largest words, and let the reference set the energy. The guest, if any, is a smaller line or left off.
A guest-led show, faceless
If the guest is the reason to click and you cannot get a photo you are allowed to use, request a faceless composition. Skip the personal photo, make the guest's name the largest words, and use the hook as the second line. WThumb instructs the model to remove people and supplies no replacement photo. Review the finished image to check that the name and hook lead and no unwanted person remains.
A guest-led show, with you in it
Lead with your own face and put the guest's name beside it. It gives the thumbnail a person without the guest's picture and keeps every episode recognisably yours. The face and the name are different people, so the words do the introducing: a line reading with Jane Smith does it in three words.
The personal photo slot is for your own identity input, not a second guest portrait. Use an extra image such as a book cover or object to help explain the topic, and use the guest's name in the words. This is a brief for a new composition, not a guarantee about which people appear; review the finished image before publishing. Pick a consistent approach for the show so viewers can recognize it.
Two people on one thumbnail: what a generator can and cannot do
The two-face split, host on the left and guest on the right, is the interview thumbnail everyone pictures.
WThumb has one personal-photo input, not two independently placed portrait layers. It uses that photo as identity input for a fresh composition, so it is not a precise two-photo layout workflow. Review the finished result for likeness and unwanted people. For a controlled two-photo split, use a tool with layers: two cut-out photos placed by hand in Canva, Adobe Express, Figma, Photoshop or an equivalent. Canva and Adobe Express have a free tier and Photoshop is paid; check what a free tier exports first.
What WThumb can do for a two-person episode:
- You plus the name. Your face from your photo, the guest as the largest line of text, the hook beneath. The closest thing to the split a single-photo generator honestly offers.
- Faceless plus two names. No photo, two names on two lines, a short hook as the third. Suits a show with two regular hosts and a strong wordmark.
- The reference's idea, recast. A two-person interview reference still works: its energy and the role each element plays carry over, and you ask for the second role to be the guest's name, or an object that stands for the topic, attached as @1.
If you need both faces in the picture, take the split to a tool with layers rather than spend a free credit finding out. If a name is enough, the generator is the shorter route, and the Shorts version comes from the same brief.
A podcast thumbnail formula you can repeat every week
A formula is a set of inputs you do not have to rethink. For a podcast, four of them, fixed once:
- One photo. Take one good photo of yourself and use it every episode: front-on or three-quarter, plain background, a face that reads small. Saved personal photos remain until you delete them or your account. Transient reference and extra images expire seven days after last use. Private prompt and instruction content is redacted after 30 days. Keep your own copy of the photo too.
- One text pattern. Line one, the guest's name. Line two, a hook of two to five words. Line three, optional, the episode number or the show's name. Same order every week, so the words fall into the same jobs even though the design around them is new. If the number stays, the episode number on a YouTube thumbnail shows where a label like EP 14 sits.
- One kind of reference. Last week's episode by link, or the podcast thumbnail you liked at the start. The reference sets the subject, the visual idea, the energy and the role of each element, so the same kind gives every run the same brief.
- One instruction line. The plain-words field is where the colors go. Dark background, white name, red hook, calm not shouty: that is enough. Write it once, paste it every week.
Then two habits per episode. Tick 16:9 and 9:16 in one submission so the episode and the clip come from the same brief. And ask for two designs rather than one, then pick the one closer to last week's.
Fix the inputs. Let the design be new. Pick the closer one.
Your words are checked by a moderation service before anything is generated; a refused request is never generated and costs nothing. And new accounts get two free credits, no card needed, which is enough to try the formula. Paid plans are there when you need more.
Keeping fifty episodes looking like one show
The intuitive method is wrong: paste last episode's link and get last episode's layout back with a new name in it. That is not what happens.
The reference is a brief, not a template. Pasting last episode's link carries over the subject, the visual idea, the energy and the role each element plays: a person, a name, a hook, a mood. What comes out is designed fresh, with its own framing, typography, color treatment and graphic structure. The idea carries over; the design is new. Fifty episodes from fifty runs are fifty designs.
So consistency does not come from the layout repeating, because it does not. It comes from what you control:
- The same face. One photo, every week. A viewer who sees the same person in the same photo does most of the recognizing for you.
- The same words in the same jobs. Name, hook, show. When the text always plays the same roles, differences in design read as variation inside a series rather than a different channel.
- The same palette, asked for in words. If your show is black and yellow, say so in the instructions every time. Color is the easiest thing to match across a row of thumbnails.
- The pick. Two designs, and you choose the one that sits better next to last week's. Over a year that choice keeps a row of them looking like one show.
There is no editor for nudging an old thumbnail into line: no canvas, no layers. Instead there are revisions: the pencil on a finished thumbnail takes a change in words, one credit each. If episode twelve came out cooler than the rest, that is the moment to type warmer, more orange, before it goes up.
The clip version for Shorts from the same brief
Many podcasts cut a clip or two for Shorts from every episode, and each clip wants a vertical thumbnail. Tick 9:16 alongside 16:9 and the same brief produces both.
What changes in the vertical version is the order, not the idea: words up top, subject beneath, less sitting side by side. It is designed for its frame rather than cropped from the landscape one: a 9:16 slice keeps under a third of a 16:9 frame's width, and the design lives in the two thirds you would cut.
- Swap the hook for the clip's line. An episode hook says what the hour is about; a clip hook is the one sentence the clip is built around. If they differ, run two briefs. If the same hook fits both, one brief with both formats ticked is quicker.
- Keep the name. A Shorts viewer meets the clip cold, so the guest's name and the show's name do more work here than on the episode thumbnail.
As of 2026, YouTube's Create a podcast in YouTube Studio page notes that Shorts created to support a podcast will not appear in YouTube Music, so the clip's audience is on YouTube itself. It also says to add podcast features to a playlist only if it contains full podcast episodes, so the clips stay out of that playlist.
Podcast thumbnail ideas when the episode has no obvious image
A podcast episode is an hour of talking. Some weeks the guest's name is enough; some weeks the topic has no picture attached to it. Ideas that work as a brief, each as words plus one instruction:
- The claim. Words: the guest's most quotable sentence, cut to five. Instruction: make the quote the largest thing on the thumbnail. The faceless interview thumbnail at its plainest.
- The object. Every topic has a thing: a book, a bottle, a chart, a chess piece. Attach a picture of it as @1 and say put @1 beside me or, faceless, make @1 the subject. It gives the design something to build around that is not a face.
- The guest's work. If the guest has a book, a film or a product, ask them for the image they use to promote it, attach it as @1 and put it in. Their work does the job their face would have, and it is an image they want seen.
The ideas share one rule: the words are the thumbnail. Pick the reference for its energy, keep the text pattern, and let the design come out new. By episode fifty the show will look like one show because the inputs never moved.



