FLIR_Image_Capture_note.pdf

There are a few different capturing approaches available:

• Digital video CMOS 8-bit, CMOS 14-bit, BT656 8-bit, LVDS This output doesn’t go through storage. CMOS 14-bit output doesn’t have any camera processing, the camera sends the raw capturing data over CMOS. 8-bit data is processed.
• Snapshot capture This is described in Image Capture Note. Uses camera's flash memory. The image is stored in 8-bit BMP format, then can be retrieved. One snapshot can take several seconds (Gemini and GPT say that)
• Analog video NTSC/PAL
• CameraLink
• Serial? When capturing with Camera control GUI, the image is transferred over serial. Probably belongs to Camera Link

Camera also supports

Internal & External framesync. For now we are using Internal framesync.

Primary goal: take 100fps for 30s

Using digital video. Enabling CMOS, setting it to 14-bit, disabling CameraLink.

To implement the primary task, we don't need information from Image Capture Note. We also don't need to request every individual frame or go through the camera's storage. The frames should just keep coming!

Steps (will be updated): 0. Use internal frame sync. The DCMI signals are generated by the camera itself.

1. Enable CMOS, set it to 14-bit, disable CameraLink, analog DIGITAL_OUTPUT_MODE (0x12), call it several times
2. Set Window ROI + frame rate in 1 command COOLED_CORE_COMMAND
3. Start capturing DCMI on STM32. Frames will start coming by themselves!

Generally speaking, the frames can be stored in the camera's memory and then transferred to PAY,

Secondary goal: take a snapshot through flash

Useful for debugging, implementing storage on our side. Still useful to support it on the mission if things go wrong. It is always easier to request 1 frame from camera's memory then consume 3000.

The commands for that are described in the image capture note and will be added here.

Main point: requesting one frame requires to send different commands from requesting a video.