AI Portraits with Real Bokeh Background

Bokeh that looks like a lens made it. Not a filter.

Circular highlight bokeh, depth-accurate falloff, aperture-based blur gradient. The background separation that costs $2,000 in camera equipment, without the camera.

Background separation through shallow depth of field is one of the most coveted qualities in portrait photography. When the background dissolves into soft, circular blur while the face remains sharp, the brain reads 'expensive camera, skilled photographer'. This effect requires large-aperture prime lenses, a Canon 85mm f/1.2, a Sony 85mm f/1.4, or similar. ThePortraitOS simulates this optically accurate depth-of-field effect from a single selfie, modelling the specific aperture and focal length physics that create genuine bokeh rather than background blur.

Technical advantage

The human visual system is highly sensitive to optical bokeh patterns.

The human visual system is highly sensitive to optical bokeh patterns. We recognise the circular highlights, the specific quality of out-of-focus rendering, and the way foreground elements transition differently from background elements. These cues signal expensive optics, which signals professional photography, which signals someone who takes their visual presentation seriously. Gaussian blur fails this test because it lacks the optical specificity that our visual system expects.

Standard AI vs ThePortraitOS

The 0.1% technical difference.

Background blur in standard AI generators is applied as a Gaussian filter, a mathematical operation that produces uniform, smooth smearing without any relationship to optical physics. Real bokeh has specific characteristics: circular highlights (the shape of the aperture), variable intensity based on distance from the focal plane, and directional quality based on the lens design. ThePortraitOS models these optical properties from first principles, producing blur that looks like it came from a large-aperture lens rather than a filter.

What the output looks like

What a bokeh background portrait looks like.

Background separated from the subject with circular highlight bokeh visible in any point light sources (windows, lamps, environmental light points). The blur gradient follows the depth plane, objects at the same distance as the face are sharp; objects behind them are increasingly blurred based on their distance from the focal plane. The transition from sharp subject to blurred background is smooth rather than abrupt.

Common questions

Why does AI bokeh usually look fake?

Because most AI systems apply simple blur filters rather than modelling optical physics. Real bokeh has circular highlight shapes (from the aperture), variable intensity by depth, and specific directional qualities from the lens design. Filter-based blur has none of these characteristics. ThePortraitOS simulates the aperture and depth-of-field physics of specific lens configurations.

Which aperture does ThePortraitOS simulate?

The default simulation is 85mm f/1.8, the most widely used portrait configuration for background separation. You can also choose f/1.2 for more extreme bokeh or f/2.8 for a subtler, more controlled separation. The focal length can be set between 50mm and 135mm.

Does the bokeh adapt to the lighting in the portrait?

Yes. The colour and intensity of the bokeh is derived from the lighting setup applied to the portrait. A warm cinematic lighting setup produces warm-toned bokeh; a cool studio setup produces neutral-toned bokeh. This ensures optical consistency throughout the image.

Start now

One selfie. 15 seconds. 8K studio portraits.

20 portraits for $29, one-time. Credits never expire. Your identity model is stored permanently so you can generate new portraits at any time.

Create AI Portrait → Select plan