How to Add a Realistic Shadow After Removing the Background

I still do manual masks occasionally — for a hero shot that's going on a billboard, or a really tricky glass-on-glass product. Outside of that, the math doesn't work. A modern segmentation model trained on millions of images sees realistic, anchored cutouts more often than any individual designer ever will. It knows what hair looks like at the edge of a face. It knows what fabric does where it meets a chair. And it doesn't get tired at image 47 of 50.

What manual masking still wins on is the absolute worst-case images: a black coat against a black couch, a glass bottle against a glass shelf. Those are real, but they're rare. For 95% of what product photographers actually shoot, AI is now the right default.

Pitfall one: the cutout has a faint colored halo. Cause: the original background bled into the subject's edge. Fix: redo with a tool that decontaminates. BG Clear does this automatically; some others don't.

Pitfall two: hair looks chunky or missing strands. Cause: the model was given a low-resolution source. Fix: re-upload a higher-resolution copy. Almost always works.

Pitfall three: the export has a watermark. Cause: you're using a free tier that watermarks free exports. Fix: switch tools.

Pitfall four: the file size is huge. Cause: alpha PNGs are big by nature. Fix: keep the PNG as master, export a JPG for the destination. For realistic, anchored cutouts specifically this happens a lot.

1. Open BG Clear. No signup screen, no email wall.

2. Drag the photo of realistic, anchored cutouts onto the upload area. JPG, PNG and WebP all work, up to 10 MB.

3. Wait about five seconds. The AI runs an InSPyReNet segmentation pass plus a ViTMatte refinement for soft edges.

4. Preview against transparent, white, black, or any of the preset colors. Pick what your downstream surface needs.

5. Hit Download. You'll get a full-resolution transparent PNG (or a flattened JPG if you picked a solid color).

That's the whole thing. If anything's wrong with the cutout, you'll usually see it in step 4 — at which point you can reupload a higher-resolution source rather than fighting with the result.

Source resolution matters more than anything else. The AI extracts cleaner edges from a 3000-pixel source than from a 600-pixel one — not because the model is different, but because it has more pixels to work with at the boundary. If the cutout looks soft, re-upload a higher-resolution version before reaching for any other fix.

Lighting matters second. Soft, even, front-facing light gives the AI clear contrast at the subject edge. Harsh side light creates shadows the AI sometimes interprets as part of the subject. For realistic, anchored cutouts, daylight from a window or a single soft box is enough; no studio kit required.

Resolution and lighting together cover maybe 80% of cutout quality. The remaining 20% is the model itself, and at this point all the major free tools are using broadly similar architectures.

The PNG is your master file. From there, product photographers typically split it three ways. First, into wherever the final asset lives for the primary use case. Second, into Figma, Canva or Photoshop for ad creatives and social posts that need different framing. Third, into a folder you'll come back to in a month when someone needs the same subject on a different background.

Keep the PNG. Always. Flatten it onto a colored background only when you're exporting for a specific destination that needs JPG. The transparent master gives you every future variation for free.

Most tools, BG Clear included, have a small number of knobs. The two that actually matter are the source resolution you upload and the background color you composite onto. Resolution drives edge quality. Background color drives whether the cutout looks naturally lit or weirdly floating.

For realistic, anchored cutouts, start with white. White is the most forgiving — any color bleed disappears against it. Once you're confident the cutout itself is clean, then experiment with brand colors or photo backgrounds. People who flip those two steps spend a lot of time fighting halos that aren't actually there.

Above ~50 images a day the UI stops being the right tool. You don't want to be drag-and-dropping a hundred files. The API takes a URL or upload binary and returns a transparent PNG, runs the same model as the browser tool, and integrates with whatever build script or CMS pipeline you already have.

For product photographers this matters specifically because realistic, anchored cutouts tends to come in batches — a shoot day, a campaign refresh, a catalog update — and 200 images at once is a different problem from 5 a week.