🐾 Softspawt

← Blog

2026-05-31

How to take the best photo for a custom pet pillow

The pillow can only be as good as the photo. A field guide to the lighting, angle, and timing that make AI-generated pet plush previews actually work β€” and the fixes when the photo is the only one you have.

The single biggest variable in how your custom pet pillow turns out is the photo you upload. Not the AI model. Not the printer. Not the fabric. The photo.

This is counter-intuitive because the product looks like it's mostly about the technology. In practice the technology is fine. The bottleneck is what we receive.

Here is a practical guide to the photo, with the trade-offs explained so you can decide what counts as "good enough" for what you want.

What the AI is actually looking at

When the model receives your photo it locks onto six things in roughly this order:

  1. Eye line. Where the eyes are pointing. This sets the entire pose of the pillow.
  2. Lighting direction. Whether the light is hitting from above, the side, behind. This determines what color the fur reads as.
  3. Distance and crop. How much of the pet is in frame. The closer the crop, the more pillow surface is filled with the pet.
  4. Sharpness. Whether edges are crisp. Blurry edges become guessed edges, which the hand-finisher then has to clean up.
  5. Background. What's behind the pet. We always remove it, but a busy background means more work and occasional bleed-through.
  6. Expression. Mouth open or closed, ears up or down, that exact look that means "her."

If any of these six is unclear, the AI guesses. The guesses are usually plausible but not always your pet.

The five photo principles that matter

Get to eye level

The most common mistake on phone photos of pets is shooting from human height down at the animal. This squishes the face vertically and makes the snout look shorter than it is. The pillow will inherit the squish.

Get on the floor. Or put the cat on the couch and shoot from the same level. The difference between a top-down photo and an eye-level photo is the difference between "looks vaguely like my dog" and "looks like my dog."

Use natural light from the side

Phone flash blows out pupils, kills color depth, and casts hard shadows the AI then bakes into the design. Window light from the side is dramatically better. If you're indoors, position the pet a few feet from a window, with the light coming across the face rather than straight on.

Avoid bright overhead light. It creates raccoon-eye shadows that the AI has to either preserve (bad) or invent its way around (also bad).

Fill the frame, but not too much

A photo where the pet takes up roughly 60–80% of the frame works best. Too small and the AI has to upscale, which softens detail. Too tight and you've cropped off the ear tips or chin, which the model then has to reconstruct.

A useful rule: you should be able to see all of the pet's head plus a little chest. Tail and back paws can be cropped without losing the pillow shape.

Sharpness matters more than resolution

A 12-megapixel phone photo that is in focus beats a 48-megapixel photo that is slightly motion-blurred. Tap to focus on the eyes before you shoot. If the pet is moving, use burst mode and pick the sharpest one.

Modern phone "portrait mode" with artificial background blur is fine for the AI. The fake blur applies only to the background and the subject stays sharp.

Wait for the resting expression

The expression in the photo becomes the expression on the pillow. Mid-bark, mid-yawn, mid-shake produce action shots that look weird as a still pillow. The expression you want is the one you see when you walk into the room and find them on the couch.

If your pet won't sit still: take twenty photos in two minutes, pick the one where the face looks most like them to you.

The seven common mistakes

These show up on at least a third of uploads. They are fixable, but you'll get a better preview faster if you avoid them.

  1. Shot from above. Squished face. Get to eye level.
  2. Flash on. Blown-out eyes, no color depth. Use natural light.
  3. Backlit silhouette. Bright window behind the pet. The face is in shadow. Turn around or move the pet.
  4. Mid-action. Tongue out, eyes closed, head turned. Wait for the rest.
  5. Far away. Tiny pet, huge couch. Move closer or zoom in before shooting.
  6. Heavy filter or beauty mode. Smoothed-out fur, fake colors. Turn it off.
  7. Screenshot of a screenshot. Compression artifacts compound. Send the original photo from your library, not a screenshot.

Special cases

Dark fur

Black dogs and black cats are the hardest to photograph because phones underexpose dark subjects. The fix is positive exposure compensation: tap to focus on the face, then drag the brightness slider up before you shoot. The fur should look almost gray. The AI can darken it back to true black; it cannot invent detail that was crushed by underexposure.

White fur

The opposite problem. White cats, white dogs, light coats all blow out highlights, especially in direct sun. Shoot in slightly dimmer indoor light, drag the brightness slider down, and accept that the fur in the photo will look more cream than white. The model can adjust the color back.

Two pets in one pillow

Send two separate clear photos of each pet. We can combine them into one pillow with both faces and a die-cut shape that flows. We cannot reliably split two pets out of a single photo where they're touching. The exception is a photo where they're already clearly separated by space and we can crop each one out cleanly.

Old printed photos you only have a scan of

Common for memorial pillows. Photograph the print under a window with your phone (not a flash), straight on, with the print flat. Make sure the scan or photo of the photo is sharp. The AI will treat it like a low-resolution original and work with what it has. Some detail is genuinely lost in old prints; the hand-finishing pass will clean what it can.

Action shots

A dog mid-jump, a cat mid-pounce. These photos are emotionally important to you and rarely make good pillow designs because the body is contorted out of pillow-shape. Use them as reference for spirit. Use a more static photo as the actual design source.

When the preview tells you the photo isn't enough

The free preview is the truth-teller here. If the photo is genuinely insufficient, you will know in fifteen seconds. The face will look slightly wrong. The pose will be off. Something will feel like "not quite her."

When that happens, you have three options.

  1. Take a better photo if the pet is still around. Most of the time, this is what we recommend.
  2. Try a different photo you already have. Sometimes the photo you assumed was the best one is not the best one for this purpose.
  3. Send what you have and write a description telling the AI what the photo isn't capturing. "The eyes should be brighter blue." "Her left ear flops, the right one stands up." The model will use the words as additional data.

Photo quality matters less than you think for one reason

If you order from a pay-first custom pillow brand, the photo quality is locked in before they touch the design. You won't know it wasn't enough until the pillow arrives.

With a preview-first AI flow, you see the design before you spend anything. If the photo wasn't enough, the preview tells you. If it was, you order. The downside risk of a sub-optimal photo is your time, not your $64.

This is why we keep saying take a better photo if you can β€” but also: don't stress about it if you can't.

Try a preview with what you have

Upload the photo you have. Watch the design generate. Iterate as many times as you want, free, until you decide one is the right one.

If it isn't, you've lost five minutes. If it is, you've ordered a pillow that already looks like your pet, on screen, before the order goes through.


Ready to see your pet as a plush pillow?

Start designing β€” free preview

Related reading