how your configuration looks like? I have a setup with a rpi 3 and a capture card and I have a slight dalay of 0.7-1 sec and i'm not able to remove it in any method.
I use RPi 3 too and had a delay until I reduced resolution to 720x480 and set scaling factor = 4. Now Hyperion task takes approximately 15% CPU (before it was 90%). I have USB Grabber.
I am using libreelec os on my raspberri pi 3b + and it still has hperion (not hperion.ng) for the external receiver capture grabber. in hperion.config.json / SMOOTHING CONFIG "time_ms" : 130,