Pixel 6: Better Portraits for All Skin Tones

Google Pixel 6: Revolutionizing Portrait Photography with AI

Smartphone manufacturers are increasingly focused on enhancing facial representation in photographs. Google’s recently unveiled Pixel 6 introduces a new collection of AI-driven tools designed to improve how people appear in images. Key among these are Face Unblur – which minimizes motion blur on faces – and Real Tone.

Addressing Challenges in Diverse Skin Tone Representation

The majority of photos captured with smartphones feature people. Historically, achieving optimal exposure for images with multiple subjects, particularly those with varying skin tones, has proven difficult. The Pixel 6 integrates advanced computational photography to ensure everyone in a photo looks their best.

The Pixel development team collaborated with a diverse group of imaging professionals and photographers to refine white balance, exposure, and underlying algorithms. This collaborative effort aims to deliver consistently flattering results for all skin tones.

Real Tone: A Commitment to Inclusive Imaging

Google views Real Tone not as a final solution, but as an ongoing mission to improve its camera systems. The company has dedicated significant resources to ensuring better representation of all individuals – especially people of color – in photographic captures.

Florian Koenigsberger, Advanced Photography product marketing manager for the Android team, shared a personal perspective: “Throughout my life, there’s been a challenge: capturing a single photograph where everyone appears well-represented.” He emphasized that the new camera represents a step forward in this journey.

Expert Collaboration and Diverse Perspectives

The camera team engaged with photographers, colorists, cinematographers, and directors of photography to gain a comprehensive understanding of the challenges associated with lighting and capturing diverse skin tones. Professionals like Ava Berkofsky, Joshua Kissi, and Kira Kelly contributed their expertise.

Koenigsberger explained, “We prioritized incorporating a wide range of perspectives, encompassing not only ethnicity and skin tones but also varied professional practices.” He noted that colorists provided particularly valuable insights, viewing image creation as a scientific process.

Rigorous Testing and Real-World Scenarios

The Google product team provided these imaging experts with Pixel 6 cameras and challenged them to photograph complex scenarios. These included mixed lighting conditions, backlighting, interior shots, and images featuring multiple skin tones.

“We needed to identify points of failure, particularly for underrepresented communities, and then determine the best course of action,” Koenigsberger stated. “The imaging professionals were candid, directly engaging with our engineers. The conversations were insightful, fostering both technical and cultural learning.”

Nuances in Skin Tone and Texture

Discussions encompassed subtleties like ashiness, textures, and variations in mid-tones, particularly for darker skin tones. The team focused on understanding these nuances to improve the camera’s performance.

Enhanced Facial Detection and Performance

The process begins with the camera’s facial detection algorithms. Once a face is identified, the system optimizes image rendering. Testing revealed that the Pixel 6 consistently outperformed competitors and previous Pixel models.

The precise mechanics of the feature remain unclear – whether it applies global edits or individually adjusts each face. A more detailed examination of the camera’s functionality is planned to assess its practical implementation.

A Core Feature with a Diverse Training Dataset

The camera team emphasized that the work undertaken has resulted in a 25-fold increase in the diversity of training datasets used to create the camera algorithms. Real Tone is integrated as a fundamental component of the camera algorithms and cannot be deactivated.