Water flows down hill. How do we know that water was perfectly leveled out at the point of the line ups?
The hand held camera's slight up down motion, in line with the black line of the water in the foreground affects the scene significantly in the far background, even if it is a pixel.
Everything needs to be perfectly leveled and aligned, and this water device is insufficient.
Furthermore, on a mountain or large hill, how do you know that the true horizon hasn't disappeared into an atmospheric fog that you can't see, thousands of miles away from you, and is squished beyond imperceptibility? This is clearly what happens when you get to high altitudes like from an international flight. The horizon is very foggy. What makes you think that the same is not true at lower altitudes, but the disappearance is more squished into the horizon by perspective?
On YouTube, there are examples of flat earth proponents using this technique to demonstrate that the horizon DOES rise to eye-level. And in the comments section, flat earth skeptics harshly discount the demo/test using many of these same critiques.
I think the principle is sound. It comes down to the execution. Anyone making a claim either way (horizon rises to eye-level or it dips below horizontal), should be able to explain why it is so and how it could be verified. How do you know it's true, one way or the other? The "dip" below horizontal on a globe the size of the earth is so small, you can't tell with just your Mk 1 mod 0 eyeball. . Testers for "flatness" and "globeness" both face the same challenges before they can claim their pet conclusion has been verified.
1. Yes, I think a clear horizon with contrast is critical.
2. Everything needs to be steady and stable. Handheld is no good, IMO. Camera should be on a tripod. Water leveler kept still and given a few moments to stabilize and then don't jostle it.
3. Everything does need to be level, but mainly it's the camera/eye level that's the key. The water will find its own level, assuming no bubble dams or vacuums. Lining up the two water level indices along the sightline of the camera/eye is the crux of the demo.
I'm thinking using the water level itself to level the camera:
This demo isn't to measure the angle of declination. It's just to see if anyone with access to simple measuring tools can detect if any angle exists. It may be imperceptible close to MSL, but if in a flat earth the "horizon is always at eye level" then increasing elevation "h" should verify that claim since in globe geometry it won't.
If you're going to make the claim one way or the other, whether in support of flatness or curve, you should be able to back it up.