Would you share your system information?
YUV422 feature support is from Haswell Gen CPU/GPU or later.

It's an i8700K which I believe is CoffeeLake. Here's info from a successful Nvenc render.

2020/11/10 08:31:50 --H:\TestFootage\Panasonic S1-vlog\P1000045_TMP.mp4
2020/11/10 08:31:50 ----------------------------------------------------------------------------------
2020/11/10 08:31:50 --NVEncC (x64) 5.19 (r1694) by rigaya, Nov 1 2020 05:46:38 (VC 1916/Win/avx2)
2020/11/10 08:31:50 --OS Version Windows 10 x64 (19041)
2020/11/10 08:31:50 --CPU Intel Core i7-8700K @ 3.70GHz [TB: 4.39GHz] (6C/12T)
2020/11/10 08:31:50 --GPU #0: GeForce GTX 1050 Ti (768 cores, 1392 MHz)[PCIe3x16][456.38]
2020/11/10 08:31:50 --NVENC / CUDA NVENC API 11.0, CUDA 11.1, schedule mode: auto
2020/11/10 08:31:50 --Input Buffers CUDA, 20 frames
2020/11/10 08:31:50 --Input Info avsw: h264(yuv422(10bit))->nv12 [-], 3840x2160, 24000/1001 fps
2020/11/10 08:31:50 --AVSync forcecfr
2020/11/10 08:31:50 --Vpp Filters copyHtoD
2020/11/10 08:31:50 -- cspconv(nv12 -> yv12)
2020/11/10 08:31:50 -- resize(spline36): 3840x2160 -> 1920x1080
2020/11/10 08:31:50 -- cspconv(yv12 -> nv12)
2020/11/10 08:31:50 --Output Info H.264/AVC high @ Level auto
2020/11/10 08:31:50 -- 1920x1080p 1:1 23.976fps (24000/1001fps)
2020/11/10 08:31:50 -- avwriter: h264 => mp4

Is your Intel GPU enabled?

Yes. Qsvenc works just fine with 420.

As I were not able to reproduce the issue in my IceLake, KabyLake or Haswell environments (all without dGPU), I think it might be related to an environment with dGPU.

As there is no way to test this on my environments, I decided to avoid this issue by a workaround, switching the GPU based YUV422→YUV420 conversion which seems to be causing this issue to the CPU based one in QSVEnc 4.09. Please have a try.

Works great. Thanks a lot.

Thanks for confirming, nice to hear it works!