These two imagers are very similar (same pixel size, same dynamic range, etc), the only difference is the number of vertical pixels and therefore the vertical FOV of the imagers. The CMV2000 has 1088 vertical pixels and is a 2/3” optical format lens while the CMV4000 has 2048 and is a 1” optical format. This means that the stock 2/3” format lenses on the MultiSense do not fully illuminate the CMV4000. Sample images for both imagers are below for reference (note that they are downsized from the native imager resolutions).
Since the CMV4000 has double the number of pixels -- the FPGA has double the work to do -- which means the max operating frame of the CMV4000 is 50% of the CMV2000. For example, here are few operating models of the SGM bitstream:
For a full list of operating modes and frame rates see:
Windowing of the image sensor (to crop out some VFOV to increase frame rate) is on our FPGA development roadmap but the release date is still to-be-determined. In the future, that will allow a CMV4000 to be cropped (e.g. masked) and have the same vertical FOV and frame rate as a CMV2000.
One thing to be aware of is that our auto-exposure and auto-white balance algorithms on-board the MultiSense run at the output frame rate, and since the CMV4000 has a frame rate of 1/2 of the CMV2000 those algorithms will take twice as much time to converge on a CMV4000 as a CMV2000. In relatively constant lighting conditions this does not matter, but if scene lighting is changing at high rate the CMV4000 will have a harder time keeping up.
But there is no different in pixel performance, dynamic range, noise floor, S/N, or anything else that would cause a change in performance. CMOSIS uses the exact same pixel design for the CMV2000 ad CMV4000 imagers -- the 4000 just has more of them.
Lastly, our API fully supports both imagers (mono and color variants of each) -- the only difference is frame rate, vertical FOV, and unit cost.