Hey @ljw2k, hope you are well. I can't speak to the 2.8mm as I don't have that, only the 4mm (but I may try and pickup a 2.8 in the future). What I can speak to is the test and the cams in hand on this scene. As always clips are straight from cam, into editor for titles, frame synced for the side by sides then encoded and back out so what you see is straight from cam in both cases preserved in pro codecs along the way. With regards to settings, you see those on screen, therefore apart from the slight reduction in gain which is set to 44 (to also help reduce the noise I mentioned), contrast tweaked slightly (all in cam), those are the only user facing, light impacting setting other than exposure which is also called out. Therefore, accurate representation of both cams in same scene.
The light drops off into the yard the further you go so if we look at the illuminators (which obviously each manufacturer applies a level of tuning to for their cams), I think they do quite well in punching through the light in what comes close to pitch black. You also have to take into account that the bushes, fence line etc dont reflect light (in this case to brighten distanced objects) and are dark in nature unlike if there were slabs all the way out to the fence line as a path or even a solid slat fence, wall in the background, bright windows front and center etc. The 1/60 shots with illuminators only for example I don’t think look that bad based on those being the only source of light.
Remember that AGC / Auto Iris algorithms (another code base adjustment not user facing) also play a role here, are handled very differently in the implementation by both companies and will take into account the bright areas and adjust. This is more aggressive on the Hik if you look at the skin color + patio for example where those are very rarely blown out (therefore preserving detail) hence AGC / Auto Iris algorithms kicking in to adjust for the bright highlights which will darken the other areas of the image. In the case of illuminators those are not super bright therefore AGC / Auto Iris is more balanced (see more in the background). Whereas If you look at the shots with the constant additional light (bulb + spot) this also makes sense as without AGC / Auto Iris kicking in aggressively on the Hik, the areas of patios (which are always front and center to the cam and washed with light) would always be blown out.
Lastly, remember the 5442 (a great camera) struggles to get to the fence line in this scene in color and that’s with an aggressive AGC and on a 4MP sensor albeit with slightly (if we want to be completely transparent) smaller aperture (f1.6 vs f1.0). Therefore I think both 4K cams perform quite well actually. While the Dahua is more liberal in its algorithms (doesn’t reduce for light in FOV as much) therefore potentially allowing more adjustment to reduce highlight blowout etc in cam, Hik takes the more aggressive approach BUT could adjust / bump their AGC / Auto Iris algorithms slightly for night (especially for the 4mm, 6mm potentially) and gain further improvements should they wish to do so. The key would be balancing gain under the covers with additional noise that may be added while preserving the highlight details they do today.
Personally I think this shows an exciting time for these 4K units as they will only get better from here and this means the next 1-2 years (outside of chip shortages
) could see these improve further, with more offerings such as vari bullets and turrets as well as larger sensors (on the same page as you here @Parley).
HTH
The light drops off into the yard the further you go so if we look at the illuminators (which obviously each manufacturer applies a level of tuning to for their cams), I think they do quite well in punching through the light in what comes close to pitch black. You also have to take into account that the bushes, fence line etc dont reflect light (in this case to brighten distanced objects) and are dark in nature unlike if there were slabs all the way out to the fence line as a path or even a solid slat fence, wall in the background, bright windows front and center etc. The 1/60 shots with illuminators only for example I don’t think look that bad based on those being the only source of light.
Remember that AGC / Auto Iris algorithms (another code base adjustment not user facing) also play a role here, are handled very differently in the implementation by both companies and will take into account the bright areas and adjust. This is more aggressive on the Hik if you look at the skin color + patio for example where those are very rarely blown out (therefore preserving detail) hence AGC / Auto Iris algorithms kicking in to adjust for the bright highlights which will darken the other areas of the image. In the case of illuminators those are not super bright therefore AGC / Auto Iris is more balanced (see more in the background). Whereas If you look at the shots with the constant additional light (bulb + spot) this also makes sense as without AGC / Auto Iris kicking in aggressively on the Hik, the areas of patios (which are always front and center to the cam and washed with light) would always be blown out.
Lastly, remember the 5442 (a great camera) struggles to get to the fence line in this scene in color and that’s with an aggressive AGC and on a 4MP sensor albeit with slightly (if we want to be completely transparent) smaller aperture (f1.6 vs f1.0). Therefore I think both 4K cams perform quite well actually. While the Dahua is more liberal in its algorithms (doesn’t reduce for light in FOV as much) therefore potentially allowing more adjustment to reduce highlight blowout etc in cam, Hik takes the more aggressive approach BUT could adjust / bump their AGC / Auto Iris algorithms slightly for night (especially for the 4mm, 6mm potentially) and gain further improvements should they wish to do so. The key would be balancing gain under the covers with additional noise that may be added while preserving the highlight details they do today.
Personally I think this shows an exciting time for these 4K units as they will only get better from here and this means the next 1-2 years (outside of chip shortages
) could see these improve further, with more offerings such as vari bullets and turrets as well as larger sensors (on the same page as you here @Parley).HTH
