Chen Liang's "Mirrorless Camera for Makers" Is a Tiny Photography Powerhouse
Built around a LILYGO T-Display S3 Pro, this compact camera offers interchangeable lenses and the possibility of onboard machine learning.
Maker Chen Liang (陳亮) has built a "mirrorless camera for makers," and it's considerably more compact than anything you'd find in an off-the-shelf package — powered by a LILYGO T-Display S3 Pro microcontroller.
"The full name of [the] mirrorless camera is 'mirrorless interchangeable-lens camera,'" Liang explains by way of introduction to the project. "Previously, [an] interchangeable-lens camera is [a] DSLR (Digital Single-Lens Reflex) camera. Start from digital camera, the optical viewfinder can be replaced by a digital viewfinder or live view screen. Then the whole optical design are simplified and the mirror of DSLR also be eliminated, so the new design interchangeable-lens camera is called [a] mirrorless camera."
Rather than buying an off-the-shelf commercial camera, though, Liang opted to jump into the mirrorless world with something a little more pocket-friendly. The heart of the project is the LILYGO T-Display S3 Pro, a compact smartphone-like development board with a 2.33" color touchscreen display behind which is an Espressif ESP32-S3R8 dual-core microcontroller.
To this, Liang has added the builder's choice of an Omnivision OV2640. OV3660, or OV5640 camera sensor — or any other with a compatible interface and an M12 mount, which provides the interchangeable-lens part of the camera's definition. A 3D-printed replacement for the T-Display's housing provides both somewhere the mount the camera and its lens and a tripod mount point — "[I] highly recommend us[ing the] tripod for the stabilities of manual focus," Liang writes. "Any tiny camera tripod should be OK."
A demonstration program captures photos when the button connected to general-purpose input/output (GPIO) Pin 12 is pressed, storing it to a microSD card. The result is a truly tiny camera which offers surprisingly high-quality captures — though Liang has suggested a range of possible improvements, including modifying the software to offer focusing aids and adding Wi-Fi-based remote control, GPS-based location tagging, or even onboard machine learning models for speech-driven capture, face-targeting auto-focus, and facial recognition.
The full project is documented on Instructables, with links to the parts used and the camera source code.