How to

How to stitch subtitle screenshots into one image

Four ways to do it, what each one costs you, and the four mistakes that ruin a dialogue strip.

Stitching subtitle screenshots is a small job with an outsize failure rate. The frames are nearly identical, the strip you want is a few percent of each one, and every method for getting at it trades control against repetition. Here are the four that work, in the order people try them.

Method 1 — an image editor

Open every screenshot in Photoshop, GIMP, Affinity Photo or Paint.NET. Crop each one down to its caption strip, stack them on one canvas, align the widths, export.

You get total control: exact strip heights, per-frame adjustments, any output format. You pay with time and with attention. On ten frames this is a twenty-minute job, and it is easy to crop one frame half a line tighter than the others so the strip jumps as you read down.

This is still the right method when you need per-frame artwork — a caption overlaid on a different background, or a stylised frame around the whole thing.

Method 2 — a subtitle stitcher with a manual crop

Several web tools are built for exactly this. You upload the screenshots, the tool stacks them, and you drag a handle to set the caption band. The slice is a fixed fraction of the frame, and you correct it frame by frame.

You get a fast result and no software to install. You pay with one correction per frame, and with the tool's own ceiling on how many images one run can hold.

Method 3 — a stitcher that measures the band

This is the method QuoteStitch uses, and it is the one that removes the repetition rather than speeding it up. Instead of you telling the tool where the band is, the tool works it out: each frame is downscaled and converted to greyscale, rows are compared across all the frames, and the contiguous run of rows with the highest change score becomes the band.

You get a correct first attempt on almost every run, and no fixed percentage — so a two-line subtitle gets a taller strip than a one-line subtitle without any setting being changed. You pay nothing, but you do lose the per-frame control that method 1 gives you. The band is one measurement applied to every frame, which is exactly what you want for a dialogue strip and exactly what you do not want if each frame needs its own treatment.

Method 4 — skip the screenshots entirely

If you still have the video, the screenshots were never necessary. Drop the MP4, MOV or WEBM into QuoteStitch and it samples up to 320 points across the clip, finds the subtitle band, and keeps one frame per line of dialogue. The pausing and capturing disappears from the workflow altogether.

• Video inMP4, MOV, WEBM — anything the browser can play
• Images inPNG, JPG, WEBP
• Long image outPNG at 1x or 2x, at original width or 720 / 1080 / 1440 px
• SamplingUp to 320 points across the clip, so seek time stays bounded
• LimitNo fixed image count. Practical limits come from device memory.
• NeedsSubtitles burned into the picture, not a separate .srt file

The four mistakes that ruin a dialogue strip

• Cutting the second line. The single most common failure. A slice sized for one line cannot hold a two-line subtitle, and the joke lives in the second line. This is why the band is measured here rather than set.
• Scaling frames differently. If one frame is exported at 2x and the rest at 1x, the strip drifts sideways as you read down. Keep every frame at one scale.
• Leaving the player UI in. A progress bar sitting in the detected band is fine — it gets dropped. A progress bar sitting just above the subtitle is not, because it widens the band and pulls in the shot. Full-screen the player before capturing.
• Mixing scenes. Frames from two different shots in one image read as a mistake, and they also break the detection: two scenes mean two bands, and only one of them wins.

Frequently asked