What happens to a photo before it reaches your camera roll?
By the time an image reaches your camera roll it has been through several hands. It was shot, encoded, uploaded, re-encoded by a platform, streamed to your phone at whatever quality your connection allowed, and then captured off a screen.
What survives all that is the shape of the outfit. What does not survive is much of the detail that separates one garment from another at the same shape. Understanding what specifically has been lost is what turns a bad screenshot into a good one.
- Detail. Compression removes the fine texture first, which is exactly the part that separates a knit from a woven and one price point from another.
- Colour. Every platform, screen and grade shifts the hue. The colour you can see is the colour of the light in the room the video was shot in, filtered twice more on the way to you.
- Stillness. A frame pulled from moving video is smeared along the direction of the movement. The garment is there and its edges are not.
What should you do to a screenshot before uploading it?
None of these require an editing app. All four can be done on a phone in under a minute, and they are ordered by how much they give back for the effort.
- Pause on a held pose. Scrub back and forth for a moment where the person is still. One second either side of the pose you wanted is usually much sharper than the pose itself.
- Crop the interface out. Captions, buttons, progress bars and usernames are all things a model has to look at and discard. Removing them is the single highest-value edit available.
- Keep the whole garment in frame. A hem, a cuff and a shoulder tell you more than a close-up of the middle. Crop to the outfit rather than into it.
- Go back to the source if you can. A screenshot of a repost of a screenshot has lost most of what a match depends on. One step upstream is often worth more than any editing.
If you do only one, crop. Removing the interface and the background is the highest-value edit available and it takes about ten seconds.
How do you get a sharp frame out of a video?
This is the one habit that changes video screenshots the most, and it is almost the opposite of what everyone does instinctively.
Tapping pause lands you on whatever frame was playing, which is usually mid-movement. A moving subject in a compressed video is smeared along the direction of travel, and the edges of the garment, the ones a model needs, are the first thing to go.
Dragging the progress bar slowly lets you find the moments where the person is holding still: the top of a turn, a pose held for the camera, a pause before they speak. Those frames are noticeably sharper, and they are usually within a second of the moment you wanted.
There is more on this in finding an outfit from TikTok. If you are on an iPhone, the screenshot buttons themselves add a separate limit worth knowing about, covered in how to find an outfit from a screenshot on iPhone.
Should you crop out captions, buttons and usernames?
A screenshot from any social platform arrives carrying furniture: captions across the middle, a column of buttons down one side, a username, a progress bar, sometimes a sticker directly over the garment.
All of it is content in the image as far as any model is concerned, and all of it has to be understood and discarded before the clothing gets any attention. Cropping it away removes competition for that attention, which is worth far more than the tidier picture it also gives you.
Crop generously around the outfit and keep the extremities: a hem tells you a length, a cuff tells you a sleeve, a shoulder tells you how a garment is cut. Those are worth far more than filling the frame with the middle of a jumper. The reasoning is worked through in how to crop a photo for a better match.
Can you trust the colour you see on screen?
A film still has been colour graded on purpose. A phone video has been white balanced by an algorithm making a guess about the room. A red carpet photo was lit by direct flash. A concert photo was lit in whatever colour the show was using that night.
None of those tell you the colour of the cloth. Treat colour as a rough family rather than a specific shade, and let the shape of the garment carry the search. There is a fuller version of this argument in why lighting changes what you find.
Why can some outfits never be bought?
This is worth saying plainly, because a lot of frustration comes from not knowing it.
Costume for film and television is often made for the production, heavily altered, or sourced from stock that was never sold at retail. Clothing in games and animation is drawn rather than made. Stage outfits are frequently one-offs. In each of those cases no search will find the item, because there is no item.
What works instead is treating the image as a brief rather than a target. Break the outfit into named pieces, search each one separately, and assemble something that reads the same way. That is a real and achievable outcome, and it is a different one from finding the exact garment.
How do you find a better copy of the same image?
Before editing anything, ask how many hands the image has been through. A pin that was saved from a repost of a screenshot has been compressed four times, and no amount of cropping puts back what those four passes removed.
Finding the account it originally came from and screenshotting there is frequently worth more than every other technique on this page combined. It takes a minute and it starts you from a much better image.
What does each platform do to your screenshot?
Each chapter takes one place a photo comes from, names exactly what that source does to the frame, and gives the fix. A source only earns a chapter when it breaks an image in a way none of the others do. Most of the chapters below are still being written; until they are, how outfit search differs by platform covers TikTok, Instagram, Pinterest, film and television, your own camera, live video and marketplace listings in full.