Cropping is the one thing you control that reliably changes what comes back. Not lighting, not resolution, not which phone you own. How much of the picture is the garment.
Why does cropping change the result so much?
Snagfit describes each garment in words before it searches for anything. The description is built entirely from what is visible.
A coat that fills the frame gives it a collar, a lapel, a closure, a hem and a fabric texture. The description that comes out is specific, and a specific description finds specific results.
The same coat on a person standing across a room gives it a shape and a colour. The description that comes out is "long dark coat", and that matches almost every long dark coat for sale.
How tight should you crop?
Tight enough that the garment takes up most of the picture, with a little room left around it.
Do not cut into the garment. The hem, the shoulder line and the sleeve length are part of what identifies it, so slicing them off costs more than the extra background does.
A useful check: if you can see the top and bottom of the piece, and it fills most of the width, you are done. Stop fiddling.
Should you crop to one garment or keep the whole outfit?
This is a real decision and it depends on what you want.
Crop to one piece when you want that piece and the rest is noise. A photo cropped to just the boots gets a much better boot description than a full body shot does.
Keep the whole outfit when you want everything named. Snagfit returns up to eight garments per photo, but it can only find what is in the frame. Crop the shoes out and the shoes will not appear.
If you want two specific pieces, crop to include both rather than choosing.
What if there are several people in the photo?
In a group shot, Snagfit reads the main subject. If there are three people standing together, it has to decide which one you meant.
Cropping to the person you care about removes that decision. It is the single most common fix for a result that came back describing the wrong outfit.
What can you leave in the frame?
Captions, usernames, play buttons, progress bars, sound titles. All ignored, because Snagfit is looking for garments.
Crop them out only when they sit on top of the clothes. A caption running across the middle of a jacket is genuinely hiding the jacket. A username in the corner is not doing any harm.
Does cropping lose image quality?
People hesitate to crop because it feels like throwing away detail.
Every upload is downscaled to 1280 pixels on its longest side before it is sent, so unless you are cropping down to a tiny corner of the image, you are still above what the search actually uses. A tight crop from a recent phone screenshot has plenty left.
What is the rule in one sentence?
If the garment you want does not take up most of the picture, crop until it does.
Everything else in this post is detail on that. It is the highest-value habit in outfit search, it takes about five seconds, and almost nobody does it before their first few disappointing results.
What does a tight crop change, exactly?
It helps to see what actually differs between a loose crop and a tight one, because "a better description" is vague.
A person standing in a wide shot produces something like: long dark coat, black, knee length. Three facts. That description matches an enormous number of coats, so the results come back generic and shuffle around every time you search.
The same coat cropped to fill the frame produces something like: camel double breasted wool overcoat, notch lapel, below the knee, heavy weave, horn buttons. Six or seven facts, several of which are unusual enough to filter hard.
Nothing changed about the coat. The second description exists because the crop made those details visible, and a description search can only use what it was given.
Does zooming in on your screen count as cropping?
A small point that catches people out: looking closely at a photo does not change what gets uploaded.
Zooming in on your screen to inspect a garment feels like you have focused on it, but the file sent is still the whole picture, and the search still sees the person as a small figure in a wide frame.
The crop has to be an actual edit. On both major phone platforms it is the crop tool in the photo editor, drag the corners in, save. It takes about five seconds and it is the difference between a vague description and a specific one.
What order should you do this in?
A small workflow that removes most of the guesswork.
First, pick the frame. If it is a video, pause it and scrub for a moment where the garment faces the camera and is still. This is worth more than the crop itself, because a blurred frame cannot be rescued by cropping into the blur.
Second, decide what you want. One piece, or the whole outfit. Deciding after you have cropped is how people accidentally cut out the shoes they wanted.
Third, crop. Pull the edges in until the target fills most of the frame, with a little room around it.
Fourth, check the light. If the crop is now mostly shadow, go back and find a brighter frame rather than uploading a dark one. Colour is one of the fields Snagfit records, and a garment in shadow reads as a black shape.
Should you straighten a tilted photo first?
A minor point with a real effect on odd angles.
A badly tilted photo makes a garment's shape harder to read, because the silhouette is what identifies most pieces and a tilt distorts it. Most phone editors have a straighten slider next to the crop tool.
It takes a couple of seconds and it is worth doing on anything shot from a low angle or in a hurry.
How should you crop for each kind of garment?
Different pieces need different framing, and the differences are not obvious.
Outerwear. Include the full length, top of the collar to the hem. The length and the closure are the two things that separate a coat from a jacket, and both sit at the edges of the garment.
Tops. The neckline and the sleeve are what matter. A crop that cuts the shoulders off loses the two most identifying features.
Trousers and skirts. Include the waist and the hem. The rise and the length are the description, and a crop of the middle of a leg is just fabric.
Footwear. Side on, and cropped to the shoes alone. Shoes are the item that benefits most from their own separate crop, because they are the smallest thing in almost every photo.
Bags. The whole outline plus the strap. An angled shot from above hides the base, which carries a surprising amount of the shape.
Jewellery and eyewear. These are small and often need their own tight crop to register at all. Do not expect them to be found in a full-body shot.
Is it better to zoom before or crop after?
Both work, and one is better.
Zooming into a video before you screenshot uses the original frame, so you keep more real detail. Cropping afterwards works from the screenshot you already have, which is a smaller starting point.
In practice the difference is small, because every upload is downscaled to 1280 pixels anyway and a phone screenshot is usually above that. Do whichever is faster in the moment. Both are far better than uploading the whole screen untouched, which is the only choice that genuinely costs you.
When should you not crop tight?
Two cases where restraint is right.
When you want the whole look. Snagfit returns up to eight garments and can only find what is in frame. If you want everything named, keep the whole figure in and accept less detail per piece. Then take a second, tighter crop of whichever piece turns out to matter, which is the two-screenshot habit.
When the garment's shape is the point. An oversized coat is identified partly by how it hangs on the person. Crop so tight that you lose the silhouette and you have removed the thing that made it recognisable.
How can cropping make the result worse?
Cropping past the garment. If you zoom so far in that all you have is a patch of fabric, the shape is gone. Snagfit can tell you it is grey wool. It cannot tell you it is a coat.
Cropping out something you wanted. Decide what you want named before you drag the corners, not after the results come back.