Write the caption · Describe the scene · Get a finished meme
Most meme tools give you a fixed template and a text box. This one draws the whole thing: you say what is in the picture and exactly what the top and bottom lines read, and the caption comes back rendered into the image in proper meme lettering — white, heavy, black-outlined. No template to pick, so the joke is not limited to the pictures someone else uploaded.
Both were generated on Kavel while building this page, to test the part that usually breaks: small text inside a picture. Every output is AI-generated, so yours will differ in face, framing and lettering weight — read these as what the tool can hold.
The scene was described as a ginger cat in a paper party hat behind a single lit candle, and the caption was quoted exactly: ANOTHER YEAR over SAME NAP SCHEDULE. Both lines came back in the heavy white-on-black-outline lettering the format lives on, and the cake and hat landed where the description put them.
Open the image studio
The second run pushed on the scene rather than the text: a golden retriever, a crooked hat, confetti mid-air. IT IS MY BIRTHDAY and WHERE IS EVERYONE both rendered cleanly at that length. Where it strains is longer lines — past about six words the letterforms start to soften.
Try your own captionIt turns a written joke into a finished image: you supply the caption and a description of the scene, and it renders both together — the animal or person, the party hat, the cake, and the lettering across the top and bottom.
The older kind of meme generator is a template picker: it hosts a few hundred familiar images and lets you type into two boxes. That works until the joke you have needs a picture nobody uploaded. Here the picture is described in the same breath as the caption, so a specific gag — your colleague's cat, a very particular sort of tired — has somewhere to live.
Creative engine
Describe the scene, quote the two caption lines, pick a square canvas. The credit estimate appears before you generate and a failed run is refunded automatically (unless it broke the content policy).
Put the caption in quotes exactly as you want it read. The model is drawing letterforms, so it can only draw the words you actually gave it — "my name at the top" produces something, but not your name.
We stepped fourteen test runs from two words a line up to ten and the lettering stayed sharp the whole way — a twenty-three word caption rendered cleanly. What goes wrong instead is omission: the longest run drew every letter perfectly and silently dropped three words from the middle of a sentence. Read the result back before you send it.
The caption is part of the image rather than a text layer on top, so changing a word means generating again. That is the trade for not being stuck with a template.
For a flattering birthday portrait instead of a joke, the birthday photoshoot generator is the other side of the occasion, and the movie poster generator is the other tool here built around rendered text.
Three things to state outright. Leave any of them vague and the model fills the gap with something generic.
Write: top line reads "ANOTHER YEAR", bottom line reads "SAME NAP SCHEDULE". Asking for "a funny caption" hands the joke to the model, and the model is not the one who knows your friend.
Three to five words holds up. The format itself rewards this — a meme that needs a paragraph has already lost — so the technical limit and the comic one point the same way.
Name the subject, the prop and the room: a ginger cat, a paper hat, a kitchen counter. Concrete nouns produce a specific picture; "funny birthday scene" produces the most photographed one.
Three cases where the joke needs an image no template library has.
The funniest birthday message is usually about something only six people understand. A template cannot show that; a described scene can.
A generic animal in a party hat is affectionate without being personal, which is often exactly the register a work chat needs.
Recognisable meme templates carry other people's baggage and licensing questions. An original picture with your own caption avoids both — the same reason a sticker set is easier to use commercially than a borrowed one.
What renders, what does not, and what it costs.
Short lines usually do — three to five words render reliably in our runs. Longer lines, unusual punctuation and names with odd spellings are where letterforms distort. If a line comes back wrong, shorten it rather than adding instructions about the text.
Not on this page, which draws the whole picture from a description. If you want to work from a real photo, start from an upload in the image studio instead and describe the change you want — but be careful with someone's face and a joke in the same image.
Use one if a template already fits. This exists for the case where it does not: an inside joke, a specific animal, a scene nobody uploaded. The trade is that the caption is part of the picture, so editing a word means one more run.
Past roughly six words per line the lettering starts to soften in our tests. That happens to match what makes a meme land, so the practical advice — cut the line down — is the same advice a comedy editor would give.
A square works for most places these get posted, and it is what both examples on this page used. Pick the shape before generating rather than cropping after, since a crop can cut into the caption.
Free credits come with the account and cover the first runs. After that a meme costs credits like any image, the estimate for the model and size you chose appears in the generator before you start, and a failed run is refunded automatically (unless it broke the content policy).
One credit pool covers Nano Banana images and Seedance and Veo video. The cost shows before every run, and failed jobs are not charged. Use a subscription for ongoing work, or a one-time pack when you just need to top up.
What's included
Secure checkout by Stripe. Card details never touch our servers.
What's included
Secure checkout by Stripe. Card details never touch our servers.
What's included
Secure checkout by Stripe. Card details never touch our servers.
What's included
Secure checkout by Stripe. Card details never touch our servers.
One-time top-ups — buy extra credits any time you run low.
What's included
Secure checkout by Stripe. Card details never touch our servers.
What's included
Secure checkout by Stripe. Card details never touch our servers.
What's included
Secure checkout by Stripe. Card details never touch our servers.
Charges appear as “KAVEL AI” on your card statement. You can cancel any time from Settings → Billing; cancellation takes effect at the end of the period you already paid for.
Powered by
Use one balance across every supported image and video model — eleven of them today, listed below — and check the credit cost before each request.
Video models
Image models
Model credit guide
You're only charged for successful generations. The exact estimate in the generator varies by model, length, resolution, audio, and number of images.
| Type | Model | Credit cost |
|---|---|---|
| Video | Seedance 2.5 | The long-take tier, priced per second: 5s 480p ≈ 211 credits, 5s 720p ≈ 473. A full 30s take runs ≈ 1,260 at 480p and ≈ 2,835 at 720p. Supplying a reference clip lowers the per-second rate. |
| Video | Seedance 2.0 | 5s 720p image-to-video ≈ 188 credits; text-to-video ≈ 308 credits. Scales with resolution and length. |
| Video | Seedance 2 Fast | Faster and lower cost. 5s 720p text-to-video ≈ 248 credits. |
| Video | Seedance 2 Mini | The cheapest Seedance tier. 5s 720p ≈ 154 credits, 5s 480p ≈ 72 credits. |
| Video | Seedance 1.5 Pro | Audio doubles the rate. 5s 720p ≈ 27 credits silent, ≈ 53 with audio; 1080p ≈ 57 and ≈ 113. |
| Video | Veo 3.1 | Billed per video, not per second (Lite tier). About 45 credits at 720p and 53 at 1080p. |
| Video | Kling 3.0 | Audio raises the rate. 5s 720p ≈ 105 credits silent, ≈ 150 with audio; 1080p ≈ 135 and ≈ 203. |
| Video | MiniMax H3 | Fixed 2K, no resolution ladder. Priced per second — a 5s clip ≈ 158 credits. |
| Image | Nano Banana 2 | Generate or edit from text and images. About 8 credits per 1K image, 12 at 2K, 18 at 4K. |
| Image | Nano Banana Pro | Consistent run times across generations. About 12 credits per 1K or 2K image, 21 at 4K. |
| Image | Nano Banana 2 Lite | Faster, simpler variant. A flat 8 credits per image at every resolution. Start here to test. |
| Image | GPT Image 2 | The lowest-cost image model here. About 3 credits per 1K image, 6 at 2K, 12 at 4K. |
| Image | Seedream 5.0 Lite | Flat pricing — about 8 credits per image whether you generate or edit. |
Three tools for the same kind of moment, in different registers.
Two short caption lines and a sentence about the scene. The lettering is handled.
A real edit made with this tool — drag your own photo in on the left, the prompt is already written, then press Generate.
