In this scenario, I would like to focus specifically on humanoids (read: human shaped blinking eyes, protected by eye-lashes and evolved in a waterless environment and fitting human eyesockets, from the outside at least) with the desired effect of them having superior eyesight to humans - by having a better night vision but without worsening overall eyesight.
In this case, let's also disregard technicalities related to evolution (like "Why would they have both night and day vision?", or "That would not be favoured since it would be costly."). Should the technicality be insane though ("It works but now they need 50,000 kcal a day to function.") I'd appreciate a heads up.
My own research focused on the two main differences between humans and animals with a superior night vision: the tapetum lucidum and the photoreceptor cells. There is a TL;DR summary of the main conclusions below.
The obvious and flashy (;)) option is naturally the tapetum lucidum. It's a retroreflector lying behind the retina that causes light to pass the retina twice (there and back), which makes for a stronger stimulus for the photoreceptors and enhances light sensitivity. Dark environments appear lighter, since the quantity of detected light increases.
As was however stated on Wikipedia and in this question, this enhancement comes unfortunately at the expense of visual acuity since it causes images to appear blurry.
Rods and Cones
In the same question, it was suggested that one increases the number of rods.
Now, Rod cells are more sensitive to light but lack the color vision of cone cells and their response time to stimuli [para. 2] is slower - that means less details and lower change in image detection. Cone cells do what rod cells can't, but are far less sensitive to light.
Increasing the number of rod cells does not sound like a bad idea, but contrary to the statement in that answer, humans actually have more rod cells (100 mil), than cone cells (7 mil), which is why we are actually able to see something in the darkness to begin with.
Tapetum lucidum has blurriness and loss of acuity. Should the increase in the number of rods have an ample effect on the effectivity of the night vision, it might be at the expense of the cone cells, which would, in turn lead to color-blindness and loss of acuity and movement perception.
And now I am stumped. Is there a way to minimize the drawbacks? Is there another way to achieve better night vision? Thank you in advance for possible answers.