-----------
Splinecage-phase solutions
-----------
This list concerns the inevitable artifacts that either only arise at the spline-cage phase, or can only be resolved at the spline-cage phase, and only semi-rigid and organic structures, and PatchSmooth script usage issues only. This has nothing to do with purely mechanical structures, spline-cage or not. It is pre-quiltwork, post low-poly, 100% spline issues, artifacts right after the Surface modifier and thus the UVW Unwrap. A vital moment.
Having UV seams and spatial separations are all good, but there are a lot of times when this isn't appropriate because, even though there may be sharp seams and creases that are desirable, the result of using the PatchSmooth script is more important, and many times this is at odds with the low-poly geometry being used. It's usually also because sharp creases are merely a feature of what is otherwise a necessarily contiguous object, importantly so. A simple cone with an infinitely sharp singularity at its peak tries to become a hot-air balloon because of all the convergences meeting from a sparse length of edges. PatchSmooth doesn't take the volume itself into account, only density versus lengths, so a sharp long cone is exactly what it doesn't work well with. It is technically bad topology, but simultaneously we can expect nothing else from a typical low-poly mesh. It's also why, unless 100% mathematically perfect, pre-tessellated topologies result in corrugations, and why there exists this '7th-gen Goldilocks-zone' when it comes to roughly what a spline-cage should be constructed from, a maximum of several thousand vertices at the very most. Corrugations occur when density too greatly exceeds logical and efficient structure.
This document is not purely about any preparations, neither purely about spline-cage articulation, although (while at the exact same phase) it is all related with plenty of overlap. This is only about some common immediate artifacts expected at the initial spline-cage phase, and their solutions...
-----------
- 'Unweld-seams'
This is unavoidable, the entire workflow relies on having these seams broken and parted right through into the spline-cage stage. It will always be different and always require manual bezier handle editing, though the following skills will definitely be involved. The more smooth and organic the structures, the better the result of the averaging bezier normals after welding, this is more of a problem from semi-rigid and near-mechanical structures. These unweld seams definitely require the most busywork, with no clear single solution.
- 'Conical-point'
Due to the nature of how the PatchSmooth script operates via pure topology, an intense convergence with long edges all coming to a singularity, such as that of a cone spike, results in a "hot-air balloon" effect. Take that peak vertex and convert it to "smooth" vertex-type. It's not 100% perfect, but perfection means a mechanical form and then shouldn't use this PatchSmooth script method anyway. Adding a stack or two back in the mesh phase of editing will definitely help, if its UVs allow.
- 'End-cap'
Fan-style end-caps have the consequence of suffering an unfortunate 'balloon-knot' effect unless nurtured as a separated object, which inevitably it would be. It requires some very tight perpendicular frilling by extruding of its outermost edges, and can be styled using extra frilling with petal structures across the related plane, to further strain each curved edge segment, which provides optional tension to the inner structure that will remain. Again, if perfection is required, it must be mechanical and forego the PatchSmooth script anyway.
- 'Arch-end'
At a concaved opposition to an end-cap, though not a cylinder and unless completely separated by UV seams as a mechanical structure, axial creases can become easily confused and a corrugated cucumber effect can occur. Instead of separating these into different objects, frilling support-feet with an artificial seam, like that of a UV seam, can help the right-angle curve stay clean and require only the wall-face to have its perpendicular axis scaled to zero.
- 'Crease-in'
We're going to use the word 'crease' as used in a transitional method, needing sharpening and going from soft to sharp. Use manual selection to freeze or hide nearby vertices and soft-selection on only vital points, using scale to spread apart along the dominant axes. This may also require a custom orthographic view from a camera aligned to the relevant offset axis, a tedious and delicate technique to be precise.
- 'Crease-out'
As paradoxical and needless and this one sounds, again we're talking about contradicting transitions, so the opposite may be required, needing softening and going from sharp to soft. Selection sets, soft-selection and Relax modifier is great, but you can also try to use a custom orthographic view and scale to spread apart the non-dominant remaining axis. You may consider this as like a semi-crease solution where other full softening techniques are overkill.
- 'Sharpen'
It will sometimes be necessary to attempt honing a lost crease, due to it being impossible to do things like frilling, stacking or extruding of anything, mainly because the softening is more important than the crease, or more convenient. A solution is to wait until the spline-cage phase, after the UV transplant, and break (unweld) the vertices along the edge and rotate the (now two) vertices underward along the appropriate (opposing) axis, and perhaps exaggerated. When welded again, the surface will become pinched together, and a further scale can be applied ala the crease solutions, to really hone the inherent sharpness.
.
.
.
...
After a lot of time toying with different fidget-cube monstrosities, I'm struggling to imagine what I may have missed here. There is almost no problem with organic smooth structures except for UV seams, and purely mechanical structures shouldn't even need a PatchSmooth script while also having several different other approaches compatible with spline-cages, something I already proved with a soda can years ago now. Problems arise when these two worlds collide, which can be few to many depending on whatever it is and how complicated it is, I stand by my initial claim that a low-poly mesh which is constructed well, follows these guidelines and having necessary preparations such as supports, extruding, frilling, can't possibly have too many issues nor any issues unable to be resolved after everything I've documented here and by now.
.
.
.
.
.
.
https://www.violae.net/wiki/tiki-index.php?page=Spline-cages
https://www.violae.net/wiki/tiki-index.php?page=Buttertub
https://www.violae.net/wiki/tiki-index.php?page=Spline-cages(advanced
https://www.violae.net/wiki/tiki-index.php?page=UVmesh-to-Splinecage(recipe)
.
.
.