Hi, we are a team of computer engineering students and we are desingning a 4D photogrammetry capture rig built with Jetson Nano’s and 4K@30fps CSI cameras. (we are using the Nano version with 4GB of RAM and 2 CSI slots). Our current concern is regarding the capacity of the Nano to simultaneously record two 4K feeds. The camera model we chose is the e-CAM80_CUNX - Sony 4K Camera:
This camera outputs a feed in a raw NV12 format. Considering our system has no real-time requirement, what would be the most adequate way of storing/recording theses two 4k feed to a file? Note that the recording of each camera will be approximately 30 seconds (the complete system will have around 16 cameras, connected to 8 Jetson). Also, it is worth mentionning that we must use a lossless encoding (for pixel matching later on in the 3D reconstruction).
This being said, we are having difficulty to estimate the size of a frame (after it has been encoded), and thus we are having difficulty to evaluate the feasability of:
- Recording to an SD card (if the write speed allows it, what would be the required size for the card?)
- Recording to an external media over ethernet (we know the Jetson supports Gigabit ethernet, but we are concerned with the receiving-end media, which might get flooded with data if all 8 jetson’s shoot 1GB/s)
Are there other ways to store the frames that we have not thought of? Bear in mind that we have a limited budget. For instance, we thought of using NVRs, but these seem a bit over our budget.
Lastly, we know that respecting all these requirements will not be easy with a tight budget, and thus we have already thought of downgrading the resolution to 1440p or 1080p (which might be mandatory anyway considering the Nano cannot encode two 4K@30fps feeds at the same time, as we saw in the datasheet).
Many thanks in advance,
All of the team