Shallow focus is a photographic and cinematic technique in which only a narrow plane of the image is rendered in sharp detail, while elements in front of and behind that plane appear softly blurred. This effect is achieved primarily through a wide aperture (a low f-number), which reduces the depth of field, the range of distances within a scene that appear acceptably sharp. By isolating the subject from its surroundings, shallow focus directs the viewer's attention, creates a sense of depth, and can convey psychological or emotional states. In the context of Generative AI image and video generation, shallow focus is a commonly requested visual style, invoked through descriptive keywords in a Prompt to emulate the look of professional photography and cinematography.
The technique has its roots in the early history of photography, where the limitations of large-format lenses and slow film emulsions naturally produced a narrow depth of field. As camera technology advanced, photographers began to use shallow focus deliberately as an expressive tool. In the 20th century, portrait photographers such as Yousuf Karsh and Richard Avedon employed wide apertures to separate their subjects from busy backgrounds, creating iconic images with creamy, blurred backdrops. In cinema, shallow focus became a hallmark of the French New Wave and the work of directors like Jean-Luc Renoir and Orson Welles, who used it to guide the eye within complex compositions. The technique also became central to the visual language of music videos and advertising, where a sharp subject against a soft, dreamy background conveys a sense of intimacy or luxury.
Visual Characteristics
The defining feature of shallow focus is the rapid transition from sharp to soft within the image. The plane of focus, often called the focal plane, is typically positioned on the subject's eyes in portraiture or on a key object in a still life. The areas in front of and behind this plane exhibit a gradual blur, with the degree of blur increasing with distance from the focal plane. This blur is often rendered as "bokeh," a Japanese term for the aesthetic quality of the out-of-focus areas. Bokeh can appear as smooth, circular highlights when the lens has a wide aperture and rounded diaphragm blades, or as more distracting, geometric shapes with less expensive lenses. The effect is most pronounced with telephoto lenses and at close focusing distances, as both factors compress the depth of field. In video, shallow focus can be combined with rack focusing, where the plane of focus shifts from one subject to another, to guide the viewer's attention through a scene.
Origins in Art and Photography
The aesthetic of selective focus predates photography, appearing in painting where artists like Rembrandt used chiaroscuro to draw the eye to a face while allowing other details to fall into shadow. With the invention of the daguerreotype in 1839, early photographers discovered that the shallow depth of field inherent to large-format lenses could be used to create portraits with softly blurred backgrounds, a style that became popular in the 1850s. The Pictorialist movement of the late 19th century embraced soft focus as a way to elevate photography to an art form, often using lenses with spherical aberration to create a dreamy, painterly effect. In the 20th century, the development of faster films and lenses with wider maximum apertures, such as the f/1.2 and f/1.4 lenses introduced by manufacturers like Leica and Canon, gave photographers greater control over depth of field. The work of Henri Cartier-Bresson, though often associated with sharp street photography, frequently used shallow focus to isolate fleeting moments, while the portraits of Irving Penn and Arnold Newman used selective focus to emphasize the sitter's face against a textured background.
In Film and Cinematography
In cinema, shallow focus has been a tool for storytelling since the silent era. D.W. Griffith used selective focus in his 1916 film "Intolerance" to highlight characters within large sets. The 1940s saw the rise of deep focus, championed by cinematographer Gregg Toland in "Citizen Kane," which kept both foreground and background sharp, but shallow focus remained a staple for close-ups and emotional scenes. Directors like Alfred Hitchcock used it to create suspense, as in "Rear Window" (1954), where the protagonist's view of his neighbors is rendered with a telephoto lens that compresses space and blurs the courtyard. In the 1970s, the New Hollywood era embraced shallow focus for its naturalistic look, with cinematographers like Gordon Willis using it to create moody, low-key lighting in "The Godfather" (1972). Modern filmmakers, including Christopher Nolan and Denis Villeneuve, frequently employ shallow focus to create a sense of scale and intimacy, often using anamorphic lenses that produce distinctive oval bokeh. The technique is also central to the visual style of television dramas and commercials, where a sharp subject against a blurred background is a standard way to convey premium quality.
Notable Practitioners
Several photographers and cinematographers are renowned for their mastery of shallow focus. Photographer Alexei Efros, though primarily a computer vision researcher, has studied the aesthetics of depth of field in photography, contributing to an understanding of how blur affects perception. In the fine art world, photographers like Sally Mann and Rinko Kawauchi use shallow focus to create intimate, ethereal images. In cinema, cinematographer Emmanuel Lubezki, known for his work on "The Tree of Life" (2011) and "Birdman" (2014), often uses natural light and wide apertures to achieve a fluid, shallow-focus look. Roger Deakins, another celebrated cinematographer, has used shallow focus in films like "Blade Runner 2049" (2017) to create vast, atmospheric landscapes where the subject emerges from a sea of blur. The technique is also a signature of the work of director Wong Kar-wai and his cinematographer Christopher Doyle, whose films like "In the Mood for Love" (2000) feature shallow focus and slow motion to evoke longing and memory.
Shallow Focus in AI Image Generation
With the rise of Generative AI and Deep learning models, shallow focus has become a key element in the vocabulary of AI image generation. Models such as Stable Diffusion, DALL-E, and Midjourney are trained on vast datasets of photographs and artworks, and they have learned to associate certain textual cues with specific visual styles. When a user includes phrases like "shallow depth of field," "bokeh," "blurred background," or "f/1.8" in a Prompt, the model interprets these as instructions to render the subject sharply while blurring the background. The effectiveness of these cues depends on the model's training data and its ability to map language to visual features. In practice, prompt writers often combine shallow focus with other descriptors, such as "portrait," "macro," or "cinematic," to achieve a desired look. The technique is particularly useful for generating realistic portraits, product shots, and nature scenes, where isolating the subject enhances visual appeal.
Prompt Engineering for Shallow Focus
In the context of AI image and video generation, invoking shallow focus requires careful choice of keywords. Common terms include "shallow depth of field," "bokeh," "background blur," "selective focus," and "wide aperture." More specific prompts might specify the lens, such as "85mm f/1.4," or the desired bokeh shape, like "circular bokeh." For video generation, prompts may include "rack focus" or "focus pull" to indicate a shift in the plane of focus over time. Models like Stable Diffusion and Midjourney respond differently to these cues; some may require more explicit phrasing, while others infer shallow focus from context, such as a portrait prompt with a busy background. Prompt writers often iterate, adjusting the wording and adding negative prompts like "sharp background" to avoid unwanted sharpness. The use of Temperature Scaling and Top-P (Nucleus) Sampling can also affect the randomness of the output, but the primary control remains the textual description. As of 2025, newer models with improved Cross-Attention mechanisms are better at following complex prompts, allowing for more precise control over depth of field.
Relationship to Other Visual Styles
Shallow focus is closely related to other photographic techniques that manipulate depth of field. Bokeh is the aesthetic quality of the blur itself, and a prompt for "bokeh" often implies shallow focus. "Tilt-shift" photography uses lens movements to create a miniature effect, which combines shallow focus with a tilted focal plane. "Macro" photography inherently involves shallow focus due to the close working distance. In AI generation, shallow focus is often paired with "cinematic lighting" or "film grain" to evoke a movie still. It contrasts with "deep focus," where the entire scene is sharp, a style that is less common in AI prompts but can be specified with terms like "f/16" or "everything in focus." Understanding these relationships helps prompt writers combine styles effectively, such as using "shallow depth of field" with "golden hour" to create a warm, intimate portrait.
Challenges and Limitations in AI Models
While AI models can produce convincing shallow focus, they sometimes struggle with consistency. The blur may be applied unevenly, or the subject may be partially blurred, especially in complex scenes with multiple objects. Models may also overuse the effect, blurring the entire image except a tiny area, which can look unnatural. Another issue is the "bokeh" quality; AI-generated bokeh often lacks the smooth, organic look of real lenses, appearing as noisy or patterned blobs. Prompt writers can mitigate these issues by using more specific language, such as "soft circular bokeh" or "cream-colored background," and by adjusting the model's parameters. As of 2025, research in Neural network architectures, such as the U-Net used in diffusion models, continues to improve the realism of depth-of-field effects, but perfect fidelity remains an open challenge.
Future Directions
The integration of shallow focus into AI generation is likely to become more sophisticated as models evolve. Techniques like ControlNet and Inpainting allow users to specify depth maps, which can guide the model to apply blur in a physically plausible way. Video generation models, such as those being developed by OpenAI and Google DeepMind, are beginning to incorporate temporal consistency, enabling realistic focus pulls and camera movements. The ability to specify lens characteristics, such as focal length and aperture, in a prompt is also improving, giving users finer control. As Large language models become better at understanding spatial relationships, prompts may become more intuitive, allowing users to describe a scene in natural language and have the model infer the appropriate depth of field. The convergence of photography, cinematography, and AI is creating new possibilities for visual storytelling, where shallow focus remains a fundamental tool for directing attention and emotion.