2026-08-31

Making things from your pet’s photo — where the cutout fits, and what fur does to it

Etsy's own community team published a seller-facing trend note this summer — search data as of July 14, 2026, comparing the last three months to the same window a year earlier — and the pet numbers all point the same direction: custom cat portrait searches up 17% year over year (dog portraits up 7%), dog gift searches up 30%, and pet memorial searches up 30%. None of that is happening in a small market. The American Pet Products Association's current industry-stats page puts US pet ownership at 95 million households, with $158 billion spent on pets in 2024. Somewhere in that overlap — an enormous number of pet households, and fast-growing demand for things made from a specific pet's likeness — is a quiet workflow fact: almost every one of those projects starts by separating the pet from whatever it was standing in front of.

The shared first step, whatever the project is

A custom portrait mug, a sticker of your dog for a group chat, a framed memorial silhouette, a planner sticker of the cat mid-yawn — as products these have nothing in common, but as files they all begin the same way: the pet, alone, on a transparent background. The couch, the lawn, the kitchen floor have to go, because the next tool in line — a print-on-demand uploader, a messenger sticker importer, a photo-book layout — wants to place the animal on its background, not yours. (If you're commissioning a portrait from an Etsy artist, you can stop reading: you send them the ordinary photo and the isolating is their job. This post is about the do-it-yourself routes.)

Good news first: the model doesn't care that it's not a person

A reasonable worry is that a background remover is really a person remover that happens to be demoed on portraits. NearBG's model isn't: U2Netp belongs to a family of models built for salient object detection — finding the prominent subject in a photo, whatever it is — rather than for human segmentation specifically. A dog filling the frame, a cat on a windowsill, a rabbit on a rug are exactly the kind of "one clear subject" input it was made for. The failure modes that do exist have nothing to do with species, and everything to do with edges.

The honest hard case: fur and whiskers

This blog's hair post walks through the mechanics in detail, and every word of it applies double to pets: a segmentation model makes a per-pixel in-or-out decision, and a whisker — thinner than a pixel at normal photo resolution — is a genuine blend of subject and background that gets forced into one bucket anyway. Expect clean results along a dog's back, legs, and face, and expect roughness exactly where a pet is fluffiest: wispy ear fur, a tail mid-swish, whiskers against a busy background. Photo choice moves this more than anything else. A black cat on a dark sofa is the worst case (low contrast plus fine fur); the same cat against a bright plain wall, in even light, filling the frame, is a different tool experience entirely. Motion blur — a professional hazard of photographing animals — adds ambiguous pixels the same way fine fur does, so a sharp boring photo beats a charming blurry one for this particular job.

Where NearBG fits, stated narrowly

• It does the cutout step in any browser on any OS, on-device: the photo is processed inside the tab and never uploaded — no account, no server copy, nothing for a third party to retain. Pet photos feel low-stakes until you notice what's behind the pet: your living room, your street, your family at the edge of the frame.
• After the automatic pass, a touch-up brush (with a size slider) erases whatever the model wrongly kept — the patch of carpet under the belly, the toy it grabbed along with the tail — and a reset button reverts to the untouched automatic result. The brush only removes: whiskers the model already dropped can't be painted back, because the brush subtracts alpha, it doesn't restore detail that was never separated.
• The download is a PNG at the original photo's resolution. Keep it a PNG — re-saving as JPEG silently fills the transparency back in as a white box, the failure this blog's print-on-demand post documents from the printing side.
• One photo at a time is the honest granularity. A twelve-sticker pack of your dog means twelve passes; there's no batch mode. And everything after the cutout — turning it into a watercolor-style portrait, placing it on a mug template, resizing to a sticker platform's spec, laying out a die-cut — is explicitly some other tool's job.

A word about the memorial case

The fastest-growing number in Etsy's list — memorial searches up 30% — is also the one where the workflow reality is least forgiving, and worth stating plainly. When the project is a keepsake for a pet that has died, the photos you have are the photos there are. You can't reshoot toward better contrast or sharper focus, so the advice above inverts: pick the best of what exists, expect edge roughness if it's blurry or dim, and lean on the brush for the patches — knowing it cleans up excess but can't recover fur the mask never caught. On-device processing carries its own quiet weight here too: a photo like that is not one most people want to hand to an anonymous server just to get a silhouette back.

The honest summary

Every external fact above has a named source: Etsy's own community-team trend post for the 2026 search numbers (cat portraits +17%, dog gifts +30%, pet memorials +30%, data as of July 14, 2026), and the American Pet Products Association's industry-stats page for the scale (95 million US pet households, $158 billion spent in 2024). The claims about the tool come from its own shipped code: a salient-object model that handles non-human subjects as a matter of design, a fur-and-whisker limitation inherited from segmentation itself, an erase-only brush with a reset, one transparent PNG at original resolution per pass — and a hard stop where the mug, the sticker sheet, and the frame begin.