2026-08-11

Why renders are full of cutout people, and how to make your own with a real photo

Search "free PNG cutout people for renders" or "scale figure PNG" and you'll land on the same small cluster of sites architecture and design students have been trading links to for well over a decade — flat, transparent-background photos of real people, meant to be dropped into a 3D render or a floor plan to give it scale and life. It's a genuinely established practice, not a niche trick, and the reason people eventually go looking for how to make their own cutout instead of downloading someone else's is worth understanding before you try it. Here's what a cutout person actually is, why the field cares about it as much as it does, what actually separates a convincing one from an obviously fake one, and exactly what a browser-based background remover like NearBG does and doesn't do to get you there.

What a cutout person is, and why renders are full of them

In architectural visualization, a flat PNG of a person — camera-facing, transparent background, placed on a vertical plane inside an otherwise fully modeled 3D scene — is usually called a scale figure, or more informally a cutout. Architizer's own reporting on the practice puts the reason for using them plainly: "Humans in architectural renderings are essential to show the scale, activity and usability of spaces," and notes that skipping real figures for simplified, ghosted silhouettes can undermine an otherwise photo-realistic render. A building rendered with nobody in it doesn't read its own scale; a person standing in the doorway does that instantly, without a dimension line anywhere in the frame.

The site that started it — and why more people want their own now

The de facto origin of the whole genre is a Swedish site called Skalgubbar, created by Teodor Javanaud Emdén — who later went on to work at the architecture firm BIG — starting with photos of himself and his friends. Architizer's writeup calls it "the first and best website with quality cutout images" and notes, with a bit of surprise, "how many large architectural offices use Skalgubbar images within their renderings." That popularity created its own problem: because so many firms drew from the same small pool of free libraries, the same handful of people started showing up in unrelated projects all over the world, and the demographic mix of those libraries skewed narrow. Architizer points to that directly — some renders ended up "populated by wealthy young people entirely unreflective of the area's true demographics" — which is exactly why alternative libraries built around ethnic and demographic diversity emerged afterward, and exactly why the more durable fix for a specific project, a specific client team, or a specific community is a cutout made from an actual photo of an actual person relevant to that project, not another download from the same shared pool everyone else is already drawing from.

What actually makes a cutout look convincing, or immediately fake

A transparent-background photo doesn't automatically read as part of the scene — practical rendering guides are specific about what breaks the illusion. One widely used cutout-people guide puts it plainly: the figure's camera angle has to roughly match the render's own viewpoint ("if your view is eye-level, avoid top-down or low-angle figures"), its lighting direction has to plausibly agree with the scene's ("if your render has sunlight from the right, your cutout should reflect that"), and — the point that matters most for a background-removal tool specifically — the cut edge itself has to be clean: "Poor cutouts — low resolution, bad edges, inconsistent lighting — will immediately stand out." Perspective and lighting are decisions you make when you shoot the photo and place the figure in your scene. Edge quality is the one variable a background-removal tool actually controls, and it's the exact failure mode this blog's post on hair and matting covers in more depth — fine, wispy edges are where any segmentation-based tool is weakest, cutout figures included.

What NearBG actually does here, and what stays on you

NearBG removes the background from a single photo entirely inside your browser — no upload — using the U2Netp segmentation model covered in this blog's first post, and hands back a transparent PNG. If the automatic pass leaves a stray patch of background near the edge, or a fragment that should have been cut away, the touch-up brush erases it directly on the canvas; it only removes pixels, it doesn't restore detail the model never captured, and a reset button reverts to the untouched automatic result if a cleanup pass goes wrong. That's the entire job a cutout-people workflow actually needs from a background remover, and it's a genuinely useful shortcut over hunting through a shared library for a figure that happens to match your client, your team, or your project's own demographic. But it's worth being precise about where NearBG's job ends. It doesn't warp, rotate, or otherwise correct the camera angle of your source photo to match a specific render's perspective; it doesn't adjust lighting or color-grade the cutout to match the scene it's headed into; and it doesn't composite the result onto anything. Placing the transparent PNG into your render, floor plan, or slide deck — and getting the angle and light right when you shoot the source photo in the first place — both stay squarely in your hands.

A practical checklist before you shoot the source photo

• Shoot at roughly the eye-level, camera-facing angle most renders actually need — matching perspective after the fact is on you, not on the background remover.
• Use even, directional light you can consciously match, or at least not obviously contradict, against your scene's own lighting later.
• Pick a plain, high-contrast backdrop for the shoot itself — fewer ambiguous edge pixels for the segmentation model to guess wrong on, especially around hair and loose clothing.
• After automatic removal, zoom in on the silhouette and touch up any stray background patch or detached fragment before you drop the cutout into your scene.
• Treat the transparent PNG as raw material, not a finished asset — placing, scaling, and light-matching still happen in whatever 3D, CAD, or slide tool the cutout is headed into.

The honest summary

Cutout people have been a real, load-bearing part of architectural and design visualization for well over a decade — popular enough that a handful of free libraries built a small industry around them, and popular enough that relying on the same shared pool eventually produces the same faces in unrelated projects, and renders that don't reflect who a space is actually for. Making your own from a real photo fixes that at the source. NearBG's part in that is specific and honest: automatic background removal plus a manual touch-up brush, entirely in the browser, producing a transparent PNG with a clean edge. Getting the angle right when you shoot, matching the light, and placing the result in your scene are separate steps this tool doesn't take for you — not a hidden gap, just the honest edge of what a background remover, on its own, actually does.