Hi Brian,
When you unwrap an object, you do it for a specific purpose. There is, as I tried to explain above, not the best way, AKA “one size fits all” approach. Nor is it always needed to begin with.
So, let’s investigate this case. A cast iron has either the noisy surface of a sand mold or is refined with machines. The rough mold aesthetic is the overall quality, and seams show up only where the mold was split. This split is typically used to take the hardwood dummy out and leave an empty space where the iron will flow into. This means we have a third surface quality somewhere: where the iron is flowing from the outside to the hollow space inside. This is the first area that will be worked on to achieve a tool part in this case. This can be done in many ways. It would be good to have that as a single island.
The simplest part here is the (negative) sandy surface. Which is normally done with a noise-shader in the bump, normal, or displacement channel. Typically, larger details will be done via displacement and smaller parts via bump, with the challenge to find a sweet-spot. Noise does not need any projection or UV if we talk about a 3D-shaders. Those need to know where a surface (including Phong) is and what value to provide.
The next area is the part where tools have been used to enable the functionality of this tool. Parts that move in a certain way to each other need to be smooth or even polished. (Visualizing cheap or pro-tools delivers the decision here). Complete, polished tools have no need for UV, they need a reflective surface, but scratches and dents need a projection or UV information.
If a grinding tool was used, it would vary in appearance. Here we might require UV if the tool was used continuously around the surface. Those need perhaps patches around UV seams, via projection or paint, stamping.
Another area is given for company logos and words or numbers that came with the mold or machined later into it. Those use a UV with certainty, but also here, a flat projection could do the trick. One might love to have the UV mesh baked into a layer, to be able to use Photoshop (or any other image app) to place those graphical or typo elements into place with ease. Setting up for each material and scale, rotate offset, or even tile seems more work and clutters the scene.
As a side note, the standard projections often help get what one needs, certainly in conjunction with polygon selections.
Again, BodyPaint3D is taking huge advantages from a proper UV, if not at all dependent on it.
In short, to just produce a UV without a plan might be wasted time. It is a translation of images. Those images would be a top view, side view, and so on, of a real object. Then the UV is adapted to it. While many qualities can’t be photographed (refraction or reflections), they need to be simulated with shaders or inside the material. The way an object is photographed to get the textures wanted sets the direction for the UVs.
To make all of that a little bit clearer, even I repeat myself here: If the UV would be an exact replicate of the model, what would be achieved? It would be 3D. It is a translation between 2D and 3D that makes its value, and there is the challenge. (Yes, there is UVW, but that is information for an algorithm, not for images/textures.)
Does this clear the use of UVs?
Cheers