Use a Raspi camera on P2?

Somebody asked this at the live forum last Wednesday, and maybe I didn't explain well enough...

The short answer is no. Raspi has a relatively new type of camera interface called MIPI. It's sorta like HDMI or more like LVDS, in a way, at the low level.

I don't think the P2 can decode a MIPI signal in real time, but would love to be proven wrong.

One easy way is to use something like a Raspi Zero to connect to the camera and then use the I/O pins of the Raspi Zero to connect to the P2.
I've demonstrated (in this forum somewhere) some code that lets the P2 give commands to the Raspi Zero.
I did this for graphics, but could also be done for the camera.

So, I think we are stuck with parallel interface cameras. This is fine, for now...
In the years to come the parallel interface may be phased out though...

Comments

  • could the P2 sample a composite signal?

    curious,

    Mike
  • Somebody (forgot name again) was working on that... had b&w going , was trying to get color iirc...

    That would be nice...
  • I think you're thinking of @SaucySoliton , they had color too from what I remember

    I wouldn't rule out being able to receive LVDS signals using the A > D comparator mode, with the D threshold set appropriately. Its unclear how fast the path through that comparator would be
  • What's nice about the composite cameras is that they are very low cost and can have long, simple cables between camera and controller.

    Only real downside is relatively low resolution. Still, could be very useful in many applications...
  • There are LVDS receiver chips. The real problems are:
    1. Keeping the data rate below ~300mbps. Same problem that we have with not being able to do much more than 480p on HDMI.
    2. Sampling the serial signal at the right time. If the camera is the clock master then we will have to sync the P2 to that clock.
    3. Not enough memory to store an image bigger than composite video anyway.


    P2 can sample composite video. We were even planning that when designing the scope filter. It won't match a dedicated video decoder asic.

    I really need to push out some composite video code. Just need to do vertical sync first. But the images I processed on the computer look better than the display from the P2. I test the P2 with a monochrome camera since I don't decode color there anyway. Maybe the chroma carrier acts to dither around the larger steps in the ADC output. Or the computer decoder just has better filters.
  • Does the P2 support USB high speed (480Mbit/s)? The cheap USB cameras are all terribly bad but I have some from Basler which are suitable for machine vision. They support USB3.0 and have really fast reaction times and good resolution. However, although my PCs claim to support USVB3.0 they can't catch up with the data rates of the cameras. But even at USB2.0 high speed the results are good.
  • ManAtWork wrote: »
    Does the P2 support USB high speed (480Mbit/s)? The cheap USB cameras are all terribly bad but I have some from Basler which are suitable for machine vision. They support USB3.0 and have really fast reaction times and good resolution. However, although my PCs claim to support USVB3.0 they can't catch up with the data rates of the cameras. But even at USB2.0 high speed the results are good.

    No.



    The imx219 (Pi camera) datasheet suggests an output PLL frequency range of 432-916MHz. There doesn't seem to be a divider for the MIPI transmitter. :disappointed:
Sign In or Register to comment.