Add camera support for Tab5

[Semyonic]

The M5Stack Tab5 has a built-in 12MP SC2356 camera (1600x1200) via MIPI-CSI interface, but there is currently no camera module available in UIFlow2 MicroPython firmware to access it.

Hardware

• Device: M5Stack Tab5 (ESP32-P4)
• Camera: SC2356 12MP via MIPI-CSI
• Firmware: UIFlow2 v2.4.1

Current Behavior

import camera  # ImportError: no module named 'camera'

No camera-related modules are available in the Tab5 UIFlow2 firmware.

Expected Behavior

A camera module should be available for Tab5, similar to CoreS3:

import camera

camera.init(pixformat=camera.RGB565, framesize=camera.QVGA)
img = camera.snapshot()
jpg = img.to_jpeg(quality=80)

Use Case

I'm building a voice assistant with Module LLM that uses VLM (Vision Language Model) for image recognition. The VLM requires camera input:

# This works on CoreS3 but not Tab5
llm_0.vlm_setup(model='internvl2.5-1B-364-ax630c', ...)
llm_0.vlm_inference_img(work_id, camera_jpg_data, 'vlm_inference')
llm_0.vlm_inference(work_id, 'Describe this image', 'vlm_inference')

Without camera support, Tab5 cannot use VLM features with Module LLM.

Evidence Camera Works

The M5Tab5-UserDemo app demonstrates the camera working via ESP-IDF, proving the hardware and driver are functional. The camera driver just needs to be exposed to MicroPython.

Technical Context

• ESP-IDF has SC2356 driver in esp-video-components
• ESP32-P4 MIPI-CSI is supported in ESP-IDF v5.4+
• MicroPython v1.27 added ESP32-P4 support
• Camera module exists for CoreS3 (DVP interface) but not Tab5 (MIPI-CSI)

This would enable Tab5 to fully utilize Module LLM's VLM capabilities, making it a powerful edge AI device for vision + language tasks.

[lbuque]

@Semyonic This feature is under development and has encountered some issues, such as the camera image not displaying smoothly on the screen.

We expect to support this feature in March. Please give us some time.