Meme Maker Online
Turn a photo into a readable meme with classic caption styling.
Two lines of type, drawn on the bitmap
Meme Maker puts a top line and a bottom line on the image and exports imageswift-meme.jpg. Font size is a percent of the picture height, from 3% to 16%, starting at 7%. A tall portrait at 7% gets a larger letter than a short banner at 7%, because the percent follows the height of that file, not a fixed point size. That is why a caption that looks right on a square post can look huge on a phone screenshot of the same words. Move the slider after you see the preview, and do not copy a “7” from one meme to another as if it were points.
The words are painted into the pixels in Impact (or Arial Black if Impact is missing), white fill with a black stroke, and the line is forced to capitals. They are not a text layer you can edit in another app. Download, and the joke is baked in. If a word is misspelled, change it here and export again. The previous JPEG will not update.
Suggest a caption is the exception to the local rule. That button sends the image to /api/ai-caption so a model can propose a top line and a bottom line. The prompt it uses asks for a short setup and a short punchline, about eight words each, tied to what is actually in the photo, without hashtags or emoji. You can ignore the suggestion and type your own. Typing does not contact the server. Only the suggest action does. The download itself is local.
| Control | Behavior |
|---|---|
| Top and bottom text | Drawn on the image. Empty is allowed if you only want one line. |
| Font size | 3% to 16% of image height, default 7%. |
| Suggest | POST /api/ai-caption. The image leaves the browser for that request. |
| Download | imageswift-meme.jpg, encoded locally. |
Read the picture before you ask for a caption
The caption model sees a reduced look at the photo and will invent a joke that fits a generic scene if the subject is ambiguous. A picture of a dog in a kitchen may get a kitchen joke that ignores the dog. Read both lines before you download. You are publishing the words, not the model. Do not suggest captions for images you are not allowed to transmit. The request includes the picture.
Impact-style captions fail when the background is busy behind the letters. This page does not give you a text box you can drag onto a quiet patch of sky. Top is top, bottom is bottom. If the subject fills both edges, crop a little headroom on Crop Image first, then add type. White type on a white shirt will vanish. The outline helps, and it is not a magic contrast control. If you cannot read the line in the preview, neither can anyone else.
Uses that match the two fields
- A reaction image with a setup on top and a payoff below. Keep each line short. At 16% a long sentence wraps into the faces.
- A label meme with only a bottom line. Leave the top field empty and check that the bottom line is not sitting on a caption that was already burned into the source photo.
- A draft you want help starting. Suggest, then rewrite the lines yourself. The suggestion is not saved as a separate file.
- A meme that must stay private. Do not press Suggest. Type the words. The draw and the JPEG stay on the device.
Steps
- Drop the image and wait for the preview.
- Type the top line, the bottom line, or both. Set the size as a percent of height.
- If you want a machine draft, use Suggest and wait for the fields to fill. Edit them.
- Download
imageswift-meme.jpg. Rename it if you will export another. The name does not include your caption.
Frequently asked questions
Does typing a caption upload the photo?
No. The text is drawn locally. The photo is sent only if you press Suggest a caption, which calls /api/ai-caption.
What does the font size percent mean?
It is 3% to 16% of the image height, starting at 7%. The same percent produces different pixel sizes on a tall photo and a wide banner.
Can I move the text to the middle?
No. One line is placed at the top and one at the bottom. There is no free position.
Will Suggest always be funny?
No. It is asked for a short setup and a short payoff about what is in the picture. Read it, rewrite it, or ignore it. You download the words you leave in the fields.
What file do I get?
imageswift-meme.jpg. The caption is part of the pixels, not a separate text file.
Can I edit the joke later?
Not inside the JPEG. Change the fields and export again from the original image. Re-encoding the meme JPEG just adds another compression pass.