Utilizing two Arducam Raspberry Pi HQ Cameras at once, how effectively change resolution and ensure it's following through in the pipeline

We’re very new to the Nvidia platform, and Gstreamer pipelines, We’re using two Uctronics Arducam B0279s in tandem to basically get a dual video of what we’re analyzing, but we seem to have a problem where in the video is being downscaled somehow. Essentially, we want to take these videos in tandem at 4k@30fps if possible, but 2k@40fps is also fine.

I tried here to attach the output when we ultimately run the code it looks like it’s being downscaled to 960x540, which is half of 1920x1080. It LOOKS LIKE it’s telling us that it’s runing in 4k@30fps, but that’s not what the output of the videos saved or the analysis code we are running separately are seeing. It looks like it’s being downscaled to both, and we’re getting very low resolution and in the saved videos very low FPS. What’s the best way to ensure we’re effectively controlling both of these factors?

Your command downscale to 960x540. Change it to what you want.

.... ! nvvidconv flip-method=0 ! video/x-raw, width=(int)960,height=(int)540, format=(string)BGRx ! .....

So, in that section, do we just upscale it to half of the 4k ones?

Sorry, the reason I guess I’m confused is because it does report that it’s 4032x3040, but if I modify that code and halve the outputs (i.e. 2016 and 1520) should that theoretically fix it?

Sorry, I’m having issues getting into our gitlab to provide snippets of the full code haha.

There’s are two part have the width/height parameter, first place to choose sensor output size second place configure the output size.

Thank you, Shane!

I was also wondering about a separate particular issue. When we take video through manual commands (or separate code we have), we get a camera error where I think I’m interpreting it correctly as the camera pipeline is getting “stuck” activated in the code. As a result, the only thing we’ve been able to figure out is to basically restart the Jetson and it’ll work again. I attached the error below. Is there a way to mitigate this by any chance?