{"id":173,"date":"2026-10-11T10:03:42","date_gmt":"2026-10-11T10:03:42","guid":{"rendered":"https:\/\/summergeometry.org\/sgi2026\/?p=173"},"modified":"2026-10-13T08:12:12","modified_gmt":"2026-10-13T08:12:12","slug":"piecewise-developable-surfaces","status":"publish","type":"post","link":"https:\/\/summergeometry.org\/sgi2026\/piecewise-developable-surfaces\/","title":{"rendered":"Piecewise Developable Surfaces"},"content":{"rendered":"\n<p class=\"wp-block-paragraph\">Fellows: Yufan Hu, <a href=\"https:\/\/www.linkedin.com\/in\/robiul-islam-hannan\" data-type=\"link\" data-id=\"https:\/\/www.linkedin.com\/in\/robiul-islam-hannan\">Robiul Islam Hannan<\/a>, Kabir Sahni<br>Mentor: Yusuf Sahillioglu<br>Volunteer: Vivien van Veldhuizen<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Introduction:<\/h4>\n\n\n\n<p class=\"has-text-align-left wp-block-paragraph\">Developable surfaces are surfaces that can be unfolded onto a plane without stretching, compression, or other geometric distortion. They are relevant to applications in geometric design, digital fabrication, manufacturing, architecture, and related areas where three-dimensional geometry must be represented or fabricated from planar material. In continuous setting developable surface is charecterized by Gaussian curvature. The Gaussian curvature is formulated as:  <\/p>\n\n\n\n<div class=\"wp-block-math\"><math display=\"block\"><semantics><mrow><mi>K<\/mi><mo form=\"prefix\" stretchy=\"false\">(<\/mo><mi>v<\/mi><mo form=\"postfix\" stretchy=\"false\">)<\/mo><mo>=<\/mo><mn>2<\/mn><mi>\u03c0<\/mi><mo>\u2212<\/mo><mrow><munder><mo>\u2211<\/mo><mrow><mrow><mi mathvariant=\"normal\">\u0394<\/mi><\/mrow><mo>\u220b<\/mo><mi>v<\/mi><\/mrow><\/munder><\/mrow><msub><mi>\u03b8<\/mi><mrow><mi mathvariant=\"normal\">\u0394<\/mi><\/mrow><\/msub><mo form=\"prefix\" stretchy=\"false\">(<\/mo><mi>v<\/mi><mo form=\"postfix\" stretchy=\"false\">)<\/mo><mspace width=\"1em\"><\/mspace><\/mrow><annotation encoding=\"application\/x-tex\">K(v) = 2\\pi &#8211; \\sum_{\\Delta \\ni v} \\theta_{\\Delta}(v) \\quad <\/annotation><\/semantics><\/math><\/div>\n\n\n\n<ul class=\"wp-block-list\">\n<li><math><semantics><mrow><mi>K<\/mi><mo>&gt;<\/mo><mn>0<\/mn><\/mrow><annotation encoding=\"application\/x-tex\"> K&gt;0<\/annotation><\/semantics><\/math> means surface is locally a bowl-like shape<\/li>\n\n\n\n<li><math><semantics><mrow><mi>K<\/mi><mo>=<\/mo><mn>0<\/mn><\/mrow><annotation encoding=\"application\/x-tex\"> K=0<\/annotation><\/semantics><\/math> &nbsp;means surface is locally flat in at least one direction&nbsp;<\/li>\n\n\n\n<li><math><semantics><mrow><mi>K<\/mi><mo>&lt;<\/mo><mn>0<\/mn><\/mrow><annotation encoding=\"application\/x-tex\"> K&lt;0 <\/annotation><\/semantics><\/math> means surface is locally a saddle-like shape<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">Our objective was to minimize non-flat curvature regions while preserving the geometric characteristics of the original surface. To address this problem, we formulated three vertex-based optimization techniques- normal one-ring, double-cone, and virtual-sphere<strong>&#8211;<\/strong> to relocate vertices through constrained directional searches while minimizing local angle-defect curvature. A complementary 3D-grid-based local search method explores possible vertex movements in the local normal and tangent directions. It then selects the movement that reduces the curvature while maintaining the surface geometry and preventing changes in triangle orientation.<\/p>\n\n\n\n\n\n<h4 class=\"wp-block-heading\">Algorithm 1: Vertex relocation methods<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Overview<\/strong><br>All three methods restrict vertex motion to one-dimensional directions. However, only the double-cone and virtual-sphere methods explicitly sample and search along multiple rays, while the normal-constrained method jointly optimizes scalar displacements along vertex normals.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Normal joint optimizer<\/strong><br>We use the <em>angle defect<\/em> to measure discrete <em>Gaussian curvature <\/em>at each interior vertex. A vertex is considered approximately developable when its absolute angle defect is below a prescribed threshold:<\/p>\n\n\n\n<div class=\"wp-block-math\"><math display=\"block\"><semantics><mrow><mi>|<\/mi><msub><mrow><mi mathvariant=\"normal\">\u03a9<\/mi><\/mrow><mi>i<\/mi><\/msub><mi>|<\/mi><mo>\u2264<\/mo><mi>\u03f5<\/mi><\/mrow><annotation encoding=\"application\/x-tex\">|\\Omega_i|\\leq\\epsilon<\/annotation><\/semantics><\/math><\/div>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Vertices satisfying <math><semantics><mrow><mi>\u03f5<\/mi><mo>&lt;<\/mo><mi>|<\/mi><msub><mrow><mi mathvariant=\"normal\">\u03a9<\/mi><\/mrow><mi>i<\/mi><\/msub><mi>|<\/mi><mo>\u2264<\/mo><mn>1<\/mn><\/mrow><annotation encoding=\"application\/x-tex\">\\epsilon &lt; |\\Omega_i| \\leq 1<\/annotation><\/semantics><\/math> are classified as <em>moderate-defect <\/em>vertices and selected as optimization targets.<\/li>\n\n\n\n<li>Vertices whose initial angle defects satisfy <math><semantics><mrow><mi>|<\/mi><msub><mrow><mi mathvariant=\"normal\">\u03a9<\/mi><\/mrow><mi>i<\/mi><\/msub><mi>|<\/mi><mo>&gt;<\/mo><mn>1<\/mn><\/mrow><annotation encoding=\"application\/x-tex\">|\\Omega_i|&gt;1<\/annotation><\/semantics><\/math> are treated as <em>singular vertices<\/em> and remain fixed, since the unavoidable curvature of the surface should be preserved somewhere.<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\"><br>Moving one vertex changes the corner angles and angle defects of its neighbors. Therefore, we jointly optimize each moderate-defect vertex together with its one-ring neighborhood. During one optimization step, each movable vertex is restricted to move along its current normal<\/p>\n\n\n\n<div class=\"wp-block-math\"><math display=\"block\"><semantics><mrow><msubsup><mi>\ud835\udc31<\/mi><mi>i<\/mi><mo lspace=\"0em\" rspace=\"0em\" class=\"tml-prime\">\u2032<\/mo><\/msubsup><mo>=<\/mo><msub><mi>\ud835\udc31<\/mi><mi>i<\/mi><\/msub><mo>+<\/mo><msub><mi>\u03b4<\/mi><mi>i<\/mi><\/msub><msub><mi>\ud835\udc27<\/mi><mi>i<\/mi><\/msub><\/mrow><annotation encoding=\"application\/x-tex\">\\mathbf{x}_i&#8217; =\n\\mathbf{x}_i+\\delta_i\\mathbf{n}_i<\/annotation><\/semantics><\/math><\/div>\n\n\n\n<div class=\"wp-block-group\"><div class=\"wp-block-group__inner-container is-layout-constrained wp-block-group-is-layout-constrained\">\n<p class=\"has-text-align-left wp-block-paragraph\">Here, <math><semantics><msub><mi>\ud835\udc27<\/mi><mi>i<\/mi><\/msub><annotation encoding=\"application\/x-tex\">\\mathbf{n}_i<\/annotation><\/semantics><\/math> is the fixed vertex normal during the current optimization step, and <math><semantics><msub><mi>\u03b4<\/mi><mi>i<\/mi><\/msub><annotation encoding=\"application\/x-tex\">\\delta_i<\/annotation><\/semantics><\/math> is the scalar displacement being optimized. This converts the vertex relocation problem from three variables per vertex to one variable per vertex, while still allowing neighboring vertices to move together.<br><br>The main objective penalizes only the part of the angle defect that exceeds the threshold:<\/p>\n<\/div><\/div>\n\n\n\n<div class=\"wp-block-math\"><math display=\"block\"><semantics><mrow><msub><mi>E<\/mi><mi>\u03ba<\/mi><\/msub><mo>=<\/mo><mfrac><mn>1<\/mn><mrow><mi>|<\/mi><mi class=\"mathcal\">\u210b<\/mi><mi>|<\/mi><\/mrow><\/mfrac><mrow><munder><mo>\u2211<\/mo><mrow><mi>i<\/mi><mo>\u2208<\/mo><mi class=\"mathcal\">\u210b<\/mi><\/mrow><\/munder><\/mrow><msup><mrow><mo fence=\"true\" form=\"prefix\">(<\/mo><mfrac><mrow><mrow><mi>max<\/mi><mo>\u2061<\/mo><\/mrow><mo form=\"prefix\" stretchy=\"false\">(<\/mo><mi>|<\/mi><msub><mrow><mi mathvariant=\"normal\">\u03a9<\/mi><\/mrow><mi>i<\/mi><\/msub><mi>|<\/mi><mo>\u2212<\/mo><mi>\u03f5<\/mi><mo separator=\"true\">,<\/mo><mn>0<\/mn><mo form=\"postfix\" stretchy=\"false\" lspace=\"0em\" rspace=\"0em\">)<\/mo><\/mrow><mi>\u03f5<\/mi><\/mfrac><mo fence=\"true\" form=\"postfix\">)<\/mo><\/mrow><mn>2<\/mn><\/msup><\/mrow><annotation encoding=\"application\/x-tex\">E_{\\kappa} =\n\\frac{1}{|\\mathcal{H}|}\n\\sum_{i\\in\\mathcal{H}}\n\\left(\n\\frac{\\max(|\\Omega_i|-\\epsilon,0)}{\\epsilon}\n\\right)^2<\/annotation><\/semantics><\/math><\/div>\n\n\n\n<div class=\"wp-block-group\"><div class=\"wp-block-group__inner-container is-layout-constrained wp-block-group-is-layout-constrained\">\n<p class=\"has-text-align-left wp-block-paragraph\">The monitored set <math><semantics><mi class=\"mathcal\">\u210b<\/mi><annotation encoding=\"application\/x-tex\">\\mathcal{H}<\/annotation><\/semantics><\/math> contains the <em>movable vertices<\/em> and their <em>neighbors<\/em>. A vertex contributes zero energy once it satisfies <math><semantics><mrow><mi>|<\/mi><msub><mrow><mi mathvariant=\"normal\">\u03a9<\/mi><\/mrow><mi>i<\/mi><\/msub><mi>|<\/mi><mo>\u2264<\/mo><mi>\u03f5<\/mi><\/mrow><annotation encoding=\"application\/x-tex\">|\\Omega_i|\\leq\\epsilon<\/annotation><\/semantics><\/math>. Evaluating the energy over the neighborhood is important because reducing curvature at one vertex may transfer curvature to nearby vertices.<br>The complete objective combines this curvature term with regularization terms:<\/p>\n\n\n\n<div class=\"wp-block-math\"><math display=\"block\"><semantics><mrow><mi>E<\/mi><mo>=<\/mo><msub><mi>E<\/mi><mi>\u03ba<\/mi><\/msub><mo>+<\/mo><msub><mi>\u03bb<\/mi><mn>0<\/mn><\/msub><msub><mi>E<\/mi><mn>0<\/mn><\/msub><mo>+<\/mo><msub><mi>\u03bb<\/mi><mi>s<\/mi><\/msub><msub><mi>E<\/mi><mi>s<\/mi><\/msub><mo>+<\/mo><msub><mi>\u03bb<\/mi><mi>p<\/mi><\/msub><msub><mi>E<\/mi><mi>p<\/mi><\/msub><mo>+<\/mo><msub><mi>\u03bb<\/mi><mi>l<\/mi><\/msub><msub><mi>E<\/mi><mi>l<\/mi><\/msub><mo>+<\/mo><msub><mi>\u03bb<\/mi><mi>e<\/mi><\/msub><msub><mi>E<\/mi><mi>e<\/mi><\/msub><\/mrow><annotation encoding=\"application\/x-tex\">E =\nE_{\\kappa}\n+\\lambda_0E_0\n+\\lambda_sE_s\n+\\lambda_pE_p\n+\\lambda_lE_l\n+\\lambda_eE_e<\/annotation><\/semantics><\/math><\/div>\n\n\n\n<p class=\"wp-block-paragraph\">The remaining energy terms control how the mesh is deformed:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><math><semantics><msub><mi>E<\/mi><mn>0<\/mn><\/msub><annotation encoding=\"application\/x-tex\">E_0<\/annotation><\/semantics><\/math> weakly encourages smaller angle defects<\/li>\n\n\n\n<li><math><semantics><msub><mi>E<\/mi><mi>s<\/mi><\/msub><annotation encoding=\"application\/x-tex\">E_s<\/annotation><\/semantics><\/math> preserves curvature around singular vertices<\/li>\n\n\n\n<li><math><semantics><msub><mi>E<\/mi><mi>p<\/mi><\/msub><annotation encoding=\"application\/x-tex\">E_p<\/annotation><\/semantics><\/math> limits vertex displacement<\/li>\n\n\n\n<li><math><semantics><msub><mi>E<\/mi><mi>l<\/mi><\/msub><annotation encoding=\"application\/x-tex\">E_l<\/annotation><\/semantics><\/math> encourages neighboring vertices to move smoothly, and <\/li>\n\n\n\n<li><math><semantics><msub><mi>E<\/mi><mi>e<\/mi><\/msub><annotation encoding=\"application\/x-tex\">E_e<\/annotation><\/semantics><\/math> penalizes relative edge-length changes.<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">Together, these terms prevent the optimizer from reducing angle defect by introducing excessive displacement, irregular local motion, or severe geometric distortion.<\/p>\n<\/div><\/div>\n\n\n\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"492\" src=\"https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/07\/normal_joint_optimizer_steps-1-1024x492.png\" alt=\"\" class=\"wp-image-281\" srcset=\"https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/07\/normal_joint_optimizer_steps-1-1024x492.png 1024w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/07\/normal_joint_optimizer_steps-1-300x144.png 300w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/07\/normal_joint_optimizer_steps-1-768x369.png 768w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/07\/normal_joint_optimizer_steps-1-1200x576.png 1200w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/07\/normal_joint_optimizer_steps-1.png 1500w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">Figure1: Normal-constrained joint optimization.<\/figcaption><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Double-cone optimizer<\/strong><br>The double-cone optimizer processes moderate-defect vertices sequentially, starting with the vertex having the largest absolute angle defect. At each target vertex, it samples movement directions from two cones centered on the positive and negative vertex-normal directions.<br>Using the local orthonormal frame (<math><semantics><mrow><msub><mi>\ud835\udc27<\/mi><mi>i<\/mi><\/msub><mo separator=\"true\">,<\/mo><msub><mi>\ud835\udc2d<\/mi><mrow><mi>i<\/mi><mn>1<\/mn><\/mrow><\/msub><mo separator=\"true\">,<\/mo><msub><mi>\ud835\udc2d<\/mi><mrow><mi>i<\/mi><mn>2<\/mn><\/mrow><\/msub><\/mrow><annotation encoding=\"application\/x-tex\">\\mathbf{n}_i,\\mathbf{t}_{i1},\\mathbf{t}_{i2}<\/annotation><\/semantics><\/math>), an outward ray is defined as<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><math><semantics><mrow><msubsup><mi>\ud835\udc2b<\/mi><mi>j<\/mi><mo lspace=\"0em\" rspace=\"0em\">+<\/mo><\/msubsup><mo>=<\/mo><msub><mi>c<\/mi><mi>j<\/mi><\/msub><msub><mi>\ud835\udc27<\/mi><mi>i<\/mi><\/msub><mo>+<\/mo><msqrt><mrow><mn>1<\/mn><mo>\u2212<\/mo><msubsup><mi>c<\/mi><mi>j<\/mi><mn>2<\/mn><\/msubsup><\/mrow><\/msqrt><mrow><mo fence=\"true\" form=\"prefix\">(<\/mo><mrow><mi>cos<\/mi><mo>\u2061<\/mo><mspace width=\"0.1667em\"><\/mspace><\/mrow><msub><mi>\u03d5<\/mi><mi>j<\/mi><\/msub><mspace width=\"0.1667em\"><\/mspace><msub><mi>\ud835\udc2d<\/mi><mrow><mi>i<\/mi><mn>1<\/mn><\/mrow><\/msub><mo>+<\/mo><mrow><mi>sin<\/mi><mo>\u2061<\/mo><mspace width=\"0.1667em\"><\/mspace><\/mrow><msub><mi>\u03d5<\/mi><mi>j<\/mi><\/msub><mspace width=\"0.1667em\"><\/mspace><msub><mi>\ud835\udc2d<\/mi><mrow><mi>i<\/mi><mn>2<\/mn><\/mrow><\/msub><mo fence=\"true\" form=\"postfix\">)<\/mo><\/mrow><mo separator=\"true\">,<\/mo><mspace width=\"2em\"><\/mspace><msubsup><mi>\ud835\udc2b<\/mi><mi>j<\/mi><mo lspace=\"0em\" rspace=\"0em\">\u2212<\/mo><\/msubsup><mo>=<\/mo><mo form=\"prefix\" stretchy=\"false\">\u2212<\/mo><msubsup><mi>\ud835\udc2b<\/mi><mi>j<\/mi><mo lspace=\"0em\" rspace=\"0em\">+<\/mo><\/msubsup><mi>.<\/mi><\/mrow><annotation encoding=\"application\/x-tex\">\\mathbf{r}_j^{+} = c_j\\mathbf{n}_i+ \\sqrt{1-c_j^2} \\left( \\cos\\phi_j\\,\\mathbf{t}_{i1} + \\sin\\phi_j\\,\\mathbf{t}_{i2} \\right), \\qquad \\mathbf{r}_j^{-}=-\\mathbf{r}_j^{+}.<\/annotation><\/semantics><\/math><\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><br>We sample 30 rays using a cone half-angle of (<math><semantics><msup><mn>60<\/mn><mo>\u2218<\/mo><\/msup><annotation encoding=\"application\/x-tex\">60^\\circ<\/annotation><\/semantics><\/math>). The golden-angle sequence distributes the rays around the vertex normal.<br>Along each ray, candidate positions are parameterized by<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><br><math><semantics><mrow><msub><mi>\ud835\udc31<\/mi><mi>i<\/mi><\/msub><mo form=\"prefix\" stretchy=\"false\">(<\/mo><mi>t<\/mi><mo form=\"postfix\" stretchy=\"false\">)<\/mo><mo>=<\/mo><msubsup><mi>\ud835\udc31<\/mi><mi>i<\/mi><mi>k<\/mi><\/msubsup><mo>+<\/mo><mi>t<\/mi><msub><mi>\ud835\udc2b<\/mi><mi>j<\/mi><\/msub><mo separator=\"true\">,<\/mo><mspace width=\"2em\"><\/mspace><mn>0<\/mn><mo>\u2264<\/mo><mi>t<\/mi><mo>\u2264<\/mo><msub><mi>L<\/mi><mi>i<\/mi><\/msub><mi>.<\/mi><\/mrow><annotation encoding=\"application\/x-tex\">\\mathbf{x}_i(t)=\\mathbf{x}_i^k+t\\mathbf{r}_j, \\qquad 0\\leq t\\leq L_i.<\/annotation><\/semantics><\/math><\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><br>The search distance <math><semantics><msub><mi>L<\/mi><mi>i<\/mi><\/msub><annotation encoding=\"application\/x-tex\">L_i<\/annotation><\/semantics><\/math> limits both the current movement and the total displacement from the original mesh. We first evaluate five coarse intervals along each ray. If two consecutive samples contain a sign change in the angle defect, binary search is used to locate a near-zero-defect position.<br>A candidate is accepted only if it reduces the curvature objective and passes the geometric safety checks, including triangle orientation, area preservation, edge distortion, and cumulative displacement limits.<br>The cone construction is used only to generate search directions; the method does not compute the Shape Diameter Function.<\/p>\n\n\n\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"492\" src=\"https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/07\/image-10-1024x492.png\" alt=\"\" class=\"wp-image-283\" srcset=\"https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/07\/image-10-1024x492.png 1024w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/07\/image-10-300x144.png 300w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/07\/image-10-768x369.png 768w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/07\/image-10-1200x576.png 1200w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/07\/image-10.png 1500w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">Figure2: Double-cone ray search.<\/figcaption><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Virtual-sphere optimizer<\/strong><br>The virtual-sphere optimizer uses the same search procedure, binary search, objective function, sequential updates, and geometric safety checks as the double-cone method. The only difference is how the ray directions are generated.<br>Instead of restricting the directions to two cones around the vertex normal, we use a Fibonacci sphere to distribute (m) rays over the entire unit sphere:<br><math><semantics><mrow><msub><mi>z<\/mi><mi>j<\/mi><\/msub><mo>=<\/mo><mn>1<\/mn><mo>\u2212<\/mo><mfrac><mrow><mn>2<\/mn><mo form=\"prefix\" stretchy=\"false\">(<\/mo><mi>j<\/mi><mo>+<\/mo><mn>1<\/mn><mi>\/<\/mi><mn>2<\/mn><mo form=\"postfix\" stretchy=\"false\" lspace=\"0em\" rspace=\"0em\">)<\/mo><\/mrow><mi>m<\/mi><\/mfrac><mo separator=\"true\">,<\/mo><mspace width=\"2em\"><\/mspace><msub><mi>\u03d5<\/mi><mi>j<\/mi><\/msub><mo>=<\/mo><mi>j<\/mi><mi>\u03c0<\/mi><mo form=\"prefix\" stretchy=\"false\">(<\/mo><mn>3<\/mn><mo>\u2212<\/mo><msqrt><mn>5<\/mn><\/msqrt><mo form=\"postfix\" stretchy=\"false\">)<\/mo><msub><mi>\ud835\udc2b<\/mi><mi>j<\/mi><\/msub><mo>=<\/mo><mrow><mo fence=\"true\" form=\"prefix\">(<\/mo><msqrt><mrow><mn>1<\/mn><mo>\u2212<\/mo><msubsup><mi>z<\/mi><mi>j<\/mi><mn>2<\/mn><\/msubsup><\/mrow><\/msqrt><mrow><mi>cos<\/mi><mo>\u2061<\/mo><mspace width=\"0.1667em\"><\/mspace><\/mrow><msub><mi>\u03d5<\/mi><mi>j<\/mi><\/msub><mo separator=\"true\">,<\/mo><mspace width=\"0.1667em\"><\/mspace><msqrt><mrow><mn>1<\/mn><mo>\u2212<\/mo><msubsup><mi>z<\/mi><mi>j<\/mi><mn>2<\/mn><\/msubsup><\/mrow><\/msqrt><mrow><mi>sin<\/mi><mo>\u2061<\/mo><mspace width=\"0.1667em\"><\/mspace><\/mrow><msub><mi>\u03d5<\/mi><mi>j<\/mi><\/msub><mo separator=\"true\">,<\/mo><mspace width=\"0.1667em\"><\/mspace><msub><mi>z<\/mi><mi>j<\/mi><\/msub><mo fence=\"true\" form=\"postfix\">)<\/mo><\/mrow><\/mrow><annotation encoding=\"application\/x-tex\">z_j=1-\\frac{2(j+1\/2)}{m}, \\qquad \\phi_j=j\\pi(3-\\sqrt{5}) \\mathbf{r}_j= \\left( \\sqrt{1-z_j^2}\\cos\\phi_j,\\, \\sqrt{1-z_j^2}\\sin\\phi_j,\\, z_j \\right)<\/annotation><\/semantics><\/math><br>This allows the vertex to move in unrestricted normal, tangential, and oblique directions. Our current experiment uses (m=30) rays.<\/p>\n\n\n\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"492\" src=\"https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/07\/image-11-1024x492.png\" alt=\"\" class=\"wp-image-286\" srcset=\"https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/07\/image-11-1024x492.png 1024w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/07\/image-11-300x144.png 300w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/07\/image-11-768x369.png 768w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/07\/image-11-1200x576.png 1200w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/07\/image-11.png 1500w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">Figure3: Virtual-sphere ray search.<\/figcaption><\/figure>\n\n\n\n<div class=\"wp-block-group\"><div class=\"wp-block-group__inner-container is-layout-constrained wp-block-group-is-layout-constrained\">\n<h4 class=\"wp-block-heading\">Algorithm 2: Grid-based method<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Overview<\/strong><br>This method proposes a local derivative-free optimization approach for reducing the non-flat curvatures of the triangle mesh while preserving the shape. We first describe the mathematical background of the algorithm and later explain how it reduces non-flat curvatures.<\/p>\n<\/div><\/div>\n\n\n\n<h6 class=\"wp-block-heading\"><strong><strong>3D-grid local search with orientation protection<\/strong><\/strong><\/h6>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Step1<strong>:<\/strong>Gaussian curvature approximation<\/strong><br>The algorithm uses the <em>Gauss-Bonnet standard discretization formula<\/em> for triangulated surfaces to estimate discrete Gaussian curvature at each vertex.<\/p>\n\n\n\n<div class=\"wp-block-math\"><math display=\"block\"><semantics><mrow><mi>K<\/mi><mo form=\"prefix\" stretchy=\"false\">(<\/mo><mi>v<\/mi><mn>1<\/mn><mo form=\"postfix\" stretchy=\"false\">)<\/mo><mo>=<\/mo><mn>2<\/mn><mi>\u03c0<\/mi><mo>\u2212<\/mo><mrow><munder><mo>\u2211<\/mo><mrow><mrow><mi mathvariant=\"normal\">\u0394<\/mi><\/mrow><mo>\u220b<\/mo><mi>v<\/mi><\/mrow><\/munder><\/mrow><msub><mi>\u03b8<\/mi><mrow><mi mathvariant=\"normal\">\u0394<\/mi><\/mrow><\/msub><mo form=\"prefix\" stretchy=\"false\">(<\/mo><mi>v<\/mi><mo form=\"postfix\" stretchy=\"false\">)<\/mo><mspace width=\"1em\"><\/mspace><\/mrow><annotation encoding=\"application\/x-tex\">K(v1) = 2\\pi &#8211; \\sum_{\\Delta \\ni v} \\theta_{\\Delta}(v) \\quad <\/annotation><\/semantics><\/math><\/div>\n\n\n\n<div class=\"wp-block-math\"><math display=\"block\"><semantics><mrow><mi>K<\/mi><mo form=\"prefix\" stretchy=\"false\">(<\/mo><mi>v<\/mi><mn>2<\/mn><mo form=\"postfix\" stretchy=\"false\">)<\/mo><mo>=<\/mo><mi>\u03c0<\/mi><mo>\u2212<\/mo><mrow><munder><mo>\u2211<\/mo><mrow><mrow><mi mathvariant=\"normal\">\u0394<\/mi><\/mrow><mo>\u220b<\/mo><mi>v<\/mi><\/mrow><\/munder><\/mrow><msub><mi>\u03b8<\/mi><mrow><mi mathvariant=\"normal\">\u0394<\/mi><\/mrow><\/msub><mo form=\"prefix\" stretchy=\"false\">(<\/mo><mi>v<\/mi><mo form=\"postfix\" stretchy=\"false\">)<\/mo><mspace width=\"1em\"><\/mspace><\/mrow><annotation encoding=\"application\/x-tex\">K(v2) = \\pi &#8211; \\sum_{\\Delta \\ni v} \\theta_{\\Delta}(v) \\quad <\/annotation><\/semantics><\/math><\/div>\n\n\n\n<p class=\"wp-block-paragraph\">Here,<math><semantics><mrow><mi>v<\/mi><mn>1<\/mn><\/mrow><annotation encoding=\"application\/x-tex\">v1<\/annotation><\/semantics><\/math> and <math><semantics><mrow><mi>v<\/mi><mn>2<\/mn><\/mrow><annotation encoding=\"application\/x-tex\">v2<\/annotation><\/semantics><\/math> denote the sets of interior and boundary vertices. <math><semantics><mrow><msub><mi>\u03b8<\/mi><mrow><mi mathvariant=\"normal\">\u0394<\/mi><\/mrow><\/msub><mo form=\"prefix\" stretchy=\"false\">(<\/mo><mi>v<\/mi><mo form=\"postfix\" stretchy=\"false\">)<\/mo><mspace width=\"1em\"><\/mspace><\/mrow><annotation encoding=\"application\/x-tex\">\\theta_{\\Delta}(v) \\quad<\/annotation><\/semantics><\/math>represents the interior angle of triangle \u0394\u200b at vertex <math><semantics><mi>v<\/mi><annotation encoding=\"application\/x-tex\">v<\/annotation><\/semantics><\/math>. <math><semantics><mrow><msub><mi>\u03b8<\/mi><mrow><mi mathvariant=\"normal\">\u0394<\/mi><\/mrow><\/msub><mo form=\"prefix\" stretchy=\"false\">(<\/mo><mi>v<\/mi><mo form=\"postfix\" stretchy=\"false\">)<\/mo><mspace width=\"1em\"><\/mspace><\/mrow><annotation encoding=\"application\/x-tex\">\\theta_{\\Delta}(v) \\quad<\/annotation><\/semantics><\/math> is calculated using the following formula:<\/p>\n\n\n\n<div class=\"wp-block-math\"><math display=\"block\"><semantics><mrow><mi>\u03b8<\/mi><mo>=<\/mo><mi>arctan2<\/mi><mo>\u2061<\/mo><mo form=\"prefix\" stretchy=\"false\">(<\/mo><mi>\u2016<\/mi><mi>u<\/mi><mo>\u00d7<\/mo><mi>v<\/mi><mi>\u2016<\/mi><mo separator=\"true\">,<\/mo><mi>u<\/mi><mo>\u22c5<\/mo><mi>v<\/mi><mo form=\"postfix\" stretchy=\"false\">)<\/mo><\/mrow><annotation encoding=\"application\/x-tex\">\\theta = \\operatorname{arctan2}(\\Vert u \\times v \\Vert, u \\cdot v)<\/annotation><\/semantics><\/math><\/div>\n\n\n\n<p class=\"wp-block-paragraph\"><math xmlns=\"http:\/\/www.w3.org\/1998\/Math\/MathML\"><semantics><mrow><mi>u<\/mi><mo>\u22c5<\/mo><mi>v<\/mi><\/mrow><annotation encoding=\"application\/x-tex\">u \\cdot v<\/annotation><\/semantics><\/math> is the cross product proportional to the <math xmlns=\"http:\/\/www.w3.org\/1998\/Math\/MathML\"><semantics><mrow><mi>cos<\/mi><mo>\u2061<\/mo><mi>\u03b8<\/mi><\/mrow><annotation encoding=\"application\/x-tex\">\\cos\\theta<\/annotation><\/semantics><\/math> and <math xmlns=\"http:\/\/www.w3.org\/1998\/Math\/MathML\"><semantics><mrow><mi mathvariant=\"normal\">\u2225<\/mi><mi>u<\/mi><mo>\u00d7<\/mo><mi>v<\/mi><mi mathvariant=\"normal\">\u2225<\/mi><\/mrow><annotation encoding=\"application\/x-tex\">\\|u \\times v\\|<\/annotation><\/semantics><\/math> is proportional to the <math xmlns=\"http:\/\/www.w3.org\/1998\/Math\/MathML\"><semantics><mrow><mi>sin<\/mi><mo>\u2061<\/mo><mi>\u03b8<\/mi><\/mrow><annotation encoding=\"application\/x-tex\">\\sin\\theta<\/annotation><\/semantics><\/math> .<br><\/p>\n\n\n\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"554\" src=\"https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/08\/step1-1024x554.png\" alt=\"\" class=\"wp-image-1017\" srcset=\"https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/08\/step1-1024x554.png 1024w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/08\/step1-300x162.png 300w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/08\/step1-767x415.png 767w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/08\/step1-1536x831.png 1536w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/08\/step1-1200x650.png 1200w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/08\/step1-2048x1109.png 2048w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/08\/step1-1980x1072.png 1980w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">Figure4: Gaussian curvature approximation.<\/figcaption><\/figure>\n\n\n\n<p class=\"has-text-align-left wp-block-paragraph\"><strong>Step2<strong>:<\/strong><\/strong><br><strong>Local Coordinate Frame Construction:<\/strong><\/p>\n\n\n\n<p class=\"has-text-align-left wp-block-paragraph\">                                                          <math><semantics><mrow><mover><mi>u<\/mi><mo stretchy=\"false\" class=\"tml-xshift\">^<\/mo><\/mover><mo>=<\/mo><mfrac><mrow><mover><mi>n<\/mi><mo stretchy=\"false\" class=\"tml-xshift\">^<\/mo><\/mover><mo>\u00d7<\/mo><mi>e<\/mi><\/mrow><mrow><mi>|<\/mi><mover><mi>n<\/mi><mo stretchy=\"false\" class=\"tml-xshift\">^<\/mo><\/mover><mo>\u00d7<\/mo><mi>e<\/mi><mi>|<\/mi><\/mrow><\/mfrac><\/mrow><annotation encoding=\"application\/x-tex\">\\hat{u} = \\frac{\\hat{n} \\times e}{|\\hat{n} \\times e|}<\/annotation><\/semantics><\/math><br><br>                                                          <math><semantics><mrow><mover><mi>v<\/mi><mo stretchy=\"false\" class=\"tml-xshift\">^<\/mo><\/mover><mo>=<\/mo><mover><mi>n<\/mi><mo stretchy=\"false\" class=\"tml-xshift\">^<\/mo><\/mover><mo>\u00d7<\/mo><mover><mi>u<\/mi><mo stretchy=\"false\" class=\"tml-xshift\">^<\/mo><\/mover><\/mrow><annotation encoding=\"application\/x-tex\">\\hat{v} = \\hat{n} \\times \\hat{u}<\/annotation><\/semantics><\/math>       <\/p>\n\n\n\n<p class=\"has-text-align-left wp-block-paragraph\">To parameterize the local search space around a vertex, the algorithm constructs a customized local coordinate frame <math><semantics><mrow><mo form=\"prefix\" stretchy=\"false\">(<\/mo><mover><mi>n<\/mi><mo stretchy=\"false\" class=\"tml-xshift\">^<\/mo><\/mover><mo separator=\"true\">,<\/mo><mover><mi>u<\/mi><mo stretchy=\"false\" class=\"tml-xshift\">^<\/mo><\/mover><mo separator=\"true\">,<\/mo><mover><mi>v<\/mi><mo stretchy=\"false\" class=\"tml-xshift\">^<\/mo><\/mover><mo form=\"postfix\" stretchy=\"false\">)<\/mo><\/mrow><annotation encoding=\"application\/x-tex\">(\\hat{n}, \\hat{u}, \\hat{v})<\/annotation><\/semantics><\/math> using the vertex unit normal vector <math><semantics><mover><mi>n<\/mi><mo stretchy=\"false\" class=\"tml-xshift\">^<\/mo><\/mover><annotation encoding=\"application\/x-tex\">\\hat{n}<\/annotation><\/semantics><\/math> and two orthogonal tangent vectors <math><semantics><mrow><mover><mi>u<\/mi><mo stretchy=\"false\" class=\"tml-xshift\">^<\/mo><\/mover><mo separator=\"true\">,<\/mo><mover><mi>v<\/mi><mo stretchy=\"false\" class=\"tml-xshift\">^<\/mo><\/mover><\/mrow><annotation encoding=\"application\/x-tex\">\\hat{u}, \\hat{v}<\/annotation><\/semantics><\/math>. <math><semantics><mi>e<\/mi><annotation encoding=\"application\/x-tex\">e<\/annotation><\/semantics><\/math> is an arbitrary reference axis not parallel to <math><semantics><mover><mi>n<\/mi><mo stretchy=\"false\" class=\"tml-xshift\">^<\/mo><\/mover><annotation encoding=\"application\/x-tex\">\\hat{n}<\/annotation><\/semantics><\/math>                                        <math><semantics><annotation encoding=\"application\/x-tex\">  <\/annotation><\/semantics><\/math>                                                                                                                   <\/p>\n\n\n\n<p class=\"has-text-align-left wp-block-paragraph\"><strong>3D Grid<\/strong> <strong>Vector:<\/strong><br><br><math><semantics><mrow><mi>\u03b4<\/mi><mo form=\"prefix\" stretchy=\"false\">(<\/mo><msub><mi>t<\/mi><mi>n<\/mi><\/msub><mo separator=\"true\">,<\/mo><msub><mi>t<\/mi><mi>u<\/mi><\/msub><mo separator=\"true\">,<\/mo><msub><mi>t<\/mi><mi>v<\/mi><\/msub><mo form=\"postfix\" stretchy=\"false\">)<\/mo><mo>=<\/mo><msub><mi>t<\/mi><mi>n<\/mi><\/msub><mover><mi>n<\/mi><mo stretchy=\"false\" class=\"tml-xshift\">^<\/mo><\/mover><mo>+<\/mo><msub><mi>t<\/mi><mi>u<\/mi><\/msub><mover><mi>u<\/mi><mo stretchy=\"false\" class=\"tml-xshift\">^<\/mo><\/mover><mo>+<\/mo><msub><mi>t<\/mi><mi>v<\/mi><\/msub><mover><mi>v<\/mi><mo stretchy=\"false\" class=\"tml-xshift\">^<\/mo><\/mover><mo separator=\"true\">,<\/mo><mspace width=\"1em\"><\/mspace><msub><mi>t<\/mi><mi>n<\/mi><\/msub><mo>\u2208<\/mo><mi>\u03ba<\/mi><mi>S<\/mi><mo separator=\"true\">,<\/mo><mspace width=\"1em\"><\/mspace><msub><mi>t<\/mi><mi>u<\/mi><\/msub><mo separator=\"true\">,<\/mo><msub><mi>t<\/mi><mi>v<\/mi><\/msub><mo>\u2208<\/mo><mi>S<\/mi><mo separator=\"true\">,<\/mo><mspace width=\"1em\"><\/mspace><mi>S<\/mi><mo>=<\/mo><mrow><mo lspace=\"0em\" rspace=\"0em\">\u2212<\/mo><mi>\u03b7<\/mi><mo separator=\"true\">,<\/mo><mn>0<\/mn><mo separator=\"true\">,<\/mo><mi>\u03b7<\/mi><\/mrow><\/mrow><annotation encoding=\"application\/x-tex\">\\delta(t_n, t_u, t_v) = t_n \\hat{n} + t_u \\hat{u} + t_v \\hat{v}, \\quad t_n \\in \\kappa S, \\quad t_u, t_v \\in S, \\quad S = {-\\eta, 0, \\eta}<\/annotation><\/semantics><\/math><br><br>We introduce a 3D sampling grid using <math><semantics><mrow><mi>\u03b4<\/mi><mo form=\"prefix\" stretchy=\"false\">(<\/mo><msub><mi>t<\/mi><mi>n<\/mi><\/msub><mo separator=\"true\">,<\/mo><msub><mi>t<\/mi><mi>u<\/mi><\/msub><mo separator=\"true\">,<\/mo><msub><mi>t<\/mi><mi>v<\/mi><\/msub><mo form=\"postfix\" stretchy=\"false\">)<\/mo><mo>=<\/mo><msub><mi>t<\/mi><mi>n<\/mi><\/msub><mover><mi>n<\/mi><mo stretchy=\"false\" class=\"tml-xshift\">^<\/mo><\/mover><mo>+<\/mo><msub><mi>t<\/mi><mi>u<\/mi><\/msub><mover><mi>u<\/mi><mo stretchy=\"false\" class=\"tml-xshift\">^<\/mo><\/mover><mo>+<\/mo><msub><mi>t<\/mi><mi>v<\/mi><\/msub><mover><mi>v<\/mi><mo stretchy=\"false\" class=\"tml-xshift\">^<\/mo><\/mover><\/mrow><annotation encoding=\"application\/x-tex\">\\delta(t_n, t_u, t_v) = t_n \\hat{n} + t_u \\hat{u} + t_v \\hat{v}<\/annotation><\/semantics><\/math>the equation by choosing a linear combination of the basis vectors. Here, the step offsets are sampled from a uniform range bounded by a maximum displacement step <math><semantics><mi>\u03b7<\/mi><annotation encoding=\"application\/x-tex\">\\eta<\/annotation><\/semantics><\/math>. Crucially, the normal displacement <math><semantics><msub><mi>t<\/mi><mi>n<\/mi><\/msub><annotation encoding=\"application\/x-tex\">t_n<\/annotation><\/semantics><\/math> is scaled down by a damping factor <math><semantics><mi>\u03ba<\/mi><annotation encoding=\"application\/x-tex\">\\kappa<\/annotation><\/semantics><\/math> which helps preserve mesh structure and fine geometric details.<\/p>\n\n\n\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"477\" src=\"https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/08\/step2-2-1024x477.png\" alt=\"\" class=\"wp-image-1020\" srcset=\"https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/08\/step2-2-1024x477.png 1024w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/08\/step2-2-300x140.png 300w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/08\/step2-2-768x358.png 768w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/08\/step2-2-1536x716.png 1536w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/08\/step2-2-1200x560.png 1200w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/08\/step2-2-2048x955.png 2048w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/08\/step2-2-1980x923.png 1980w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">Figure5: Local coordinate frame and 3D grid sampling.<\/figcaption><\/figure>\n\n\n\n<p class=\"has-text-align-left wp-block-paragraph\"><br><br><strong>Step4: Triangle normal calculation<\/strong> <strong>and orientation protection <\/strong><br><br><math><semantics><mrow><msubsup><mi>\ud835\udc27<\/mi><mrow><mi mathvariant=\"normal\">\u0394<\/mi><\/mrow><mo lspace=\"0em\" rspace=\"0em\" class=\"tml-prime\">\u2032<\/mo><\/msubsup><mo>=<\/mo><mfrac><mrow><mo form=\"prefix\" stretchy=\"false\" lspace=\"0em\" rspace=\"0em\">(<\/mo><msub><mi>p<\/mi><mn>1<\/mn><\/msub><mo>\u2212<\/mo><msub><mi>p<\/mi><mn>0<\/mn><\/msub><mo form=\"postfix\" stretchy=\"false\">)<\/mo><mo>\u00d7<\/mo><mo form=\"prefix\" stretchy=\"false\">(<\/mo><msub><mi>p<\/mi><mn>2<\/mn><\/msub><mo>\u2212<\/mo><msub><mi>p<\/mi><mn>0<\/mn><\/msub><mo form=\"postfix\" stretchy=\"false\" lspace=\"0em\" rspace=\"0em\">)<\/mo><\/mrow><mrow><mi>\u2016<\/mi><mo form=\"prefix\" stretchy=\"false\">(<\/mo><msub><mi>p<\/mi><mn>1<\/mn><\/msub><mo>\u2212<\/mo><msub><mi>p<\/mi><mn>0<\/mn><\/msub><mo form=\"postfix\" stretchy=\"false\">)<\/mo><mo>\u00d7<\/mo><mo form=\"prefix\" stretchy=\"false\">(<\/mo><msub><mi>p<\/mi><mn>2<\/mn><\/msub><mo>\u2212<\/mo><msub><mi>p<\/mi><mn>0<\/mn><\/msub><mo form=\"postfix\" stretchy=\"false\">)<\/mo><mi>\u2016<\/mi><\/mrow><\/mfrac><mo separator=\"true\">,<\/mo><mspace width=\"1em\"><\/mspace><msubsup><mi>\ud835\udc27<\/mi><mrow><mi mathvariant=\"normal\">\u0394<\/mi><\/mrow><mo lspace=\"0em\" rspace=\"0em\" class=\"tml-prime\">\u2032<\/mo><\/msubsup><mo>\u22c5<\/mo><msubsup><mi>\ud835\udc27<\/mi><mrow><mi mathvariant=\"normal\">\u0394<\/mi><\/mrow><mn>0<\/mn><\/msubsup><mo>\u2265<\/mo><mn>0.2<\/mn><mspace width=\"1em\"><\/mspace><mi>\u2200<\/mi><mrow><mi mathvariant=\"normal\">\u0394<\/mi><\/mrow><mo>\u220b<\/mo><mi>v<\/mi><\/mrow><annotation encoding=\"application\/x-tex\">\\mathbf{n}&#8217;_{\\Delta} = \\frac{(p_1 &#8211; p_0) \\times (p_2 &#8211; p_0)}{\\Vert (p_1 &#8211; p_0) \\times (p_2 &#8211; p_0) \\Vert}, \\quad \\mathbf{n}&#8217;_{\\Delta} \\cdot \\mathbf{n}^0_{\\Delta} \\ge 0.2 \\quad \\forall \\Delta \\ni v<\/annotation><\/semantics><\/math><br><br>The algorithm enforces a strict face-inversion constraint of all triangles <math><semantics><mrow><mi mathvariant=\"normal\">\u0394<\/mi><\/mrow><annotation encoding=\"application\/x-tex\">\\Delta<\/annotation><\/semantics><\/math> incident to the moving vertex <math><semantics><mi>v<\/mi><annotation encoding=\"application\/x-tex\">v<\/annotation><\/semantics><\/math> to ensure that the vertex displacement does not compromise the topological integrity. For any candidate vertex position, the updated unit normal vector <math><semantics><msubsup><mi>\ud835\udc27<\/mi><mrow><mi mathvariant=\"normal\">\u0394<\/mi><\/mrow><mo lspace=\"0em\" rspace=\"0em\" class=\"tml-prime\">\u2032<\/mo><\/msubsup><annotation encoding=\"application\/x-tex\">\\mathbf{n}&#8217;_\\Delta<\/annotation><\/semantics><\/math> is calculated via<math><semantics><mrow><mo form=\"prefix\" stretchy=\"false\">(<\/mo><msub><mi>p<\/mi><mn>1<\/mn><\/msub><mo>\u2212<\/mo><msub><mi>p<\/mi><mn>0<\/mn><\/msub><mo form=\"postfix\" stretchy=\"false\">)<\/mo><mo>\u00d7<\/mo><mo form=\"prefix\" stretchy=\"false\">(<\/mo><msub><mi>p<\/mi><mn>2<\/mn><\/msub><mo>\u2212<\/mo><msub><mi>p<\/mi><mn>0<\/mn><\/msub><mo form=\"postfix\" stretchy=\"false\">)<\/mo><\/mrow><annotation encoding=\"application\/x-tex\">(p_1 &#8211; p_0) \\times (p_2 &#8211; p_0)<\/annotation><\/semantics><\/math>. The algorithm then evaluates the dot product between this candidate normal and the face&#8217;s original, unmodified normal <br><math><semantics><msubsup><mi>\ud835\udc27<\/mi><mrow><mi mathvariant=\"normal\">\u0394<\/mi><\/mrow><mn>0<\/mn><\/msubsup><annotation encoding=\"application\/x-tex\">\\mathbf{n}^0_\\Delta<\/annotation><\/semantics><\/math>. By requiring <math><semantics><mrow><msubsup><mi>\ud835\udc27<\/mi><mrow><mi mathvariant=\"normal\">\u0394<\/mi><\/mrow><mo lspace=\"0em\" rspace=\"0em\" class=\"tml-prime\">\u2032<\/mo><\/msubsup><mo>\u22c5<\/mo><msubsup><mi>\ud835\udc27<\/mi><mrow><mi mathvariant=\"normal\">\u0394<\/mi><\/mrow><mn>0<\/mn><\/msubsup><mo>\u2265<\/mo><mn>0.2<\/mn><\/mrow><annotation encoding=\"application\/x-tex\">\\mathbf{n}&#8217;_\\Delta \\cdot \\mathbf{n}^0_\\Delta \\ge 0.2<\/annotation><\/semantics><\/math>, the algorithm imposes a lower strict bound that rejects any vertex update causing an adjacent triangle distortion. This function prevents mesh flipping, severe distortion, and self-intersection during local search.<\/p>\n\n\n\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"439\" src=\"https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/08\/step3-1024x439.png\" alt=\"\" class=\"wp-image-1021\" srcset=\"https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/08\/step3-1024x439.png 1024w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/08\/step3-767x329.png 767w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/08\/step3-300x129.png 300w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/08\/step3-1536x659.png 1536w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/08\/step3-1200x515.png 1200w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/08\/step3-2048x879.png 2048w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/08\/step3-1980x850.png 1980w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">Figure6: Orientation protection.<\/figcaption><\/figure>\n\n\n\n<p class=\"has-text-align-left wp-block-paragraph\"><br><br><strong>Step5: <\/strong><\/p>\n\n\n\n<p class=\"has-text-align-left wp-block-paragraph\"><strong>Energy objective function: <\/strong><br><br><math><semantics><mrow><mi>E<\/mi><mo form=\"prefix\" stretchy=\"false\">(<\/mo><msub><mi>p<\/mi><mi>v<\/mi><\/msub><mo form=\"postfix\" stretchy=\"false\">)<\/mo><mo>=<\/mo><mi>|<\/mi><mi>K<\/mi><mo form=\"prefix\" stretchy=\"false\">(<\/mo><mi>v<\/mi><mo form=\"postfix\" stretchy=\"false\">)<\/mo><mi>|<\/mi><mo>+<\/mo><msub><mo>\u2211<\/mo><mrow><mi>u<\/mi><mo>\u2208<\/mo><mi>N<\/mi><mo form=\"prefix\" stretchy=\"false\">(<\/mo><mi>v<\/mi><mo form=\"postfix\" stretchy=\"false\" lspace=\"0em\" rspace=\"0em\">)<\/mo><\/mrow><\/msub><mrow><mo fence=\"true\" form=\"prefix\">[<\/mo><mi>|<\/mi><mi>K<\/mi><mo form=\"prefix\" stretchy=\"false\">(<\/mo><mi>u<\/mi><mo form=\"postfix\" stretchy=\"false\">)<\/mo><mi>|<\/mi><mo>+<\/mo><mi>W<\/mi><mo>\u22c5<\/mo><mn>1<\/mn><mo fence=\"false\" symmetric=\"true\" minsize=\"1.2em\" maxsize=\"1.2em\">[<\/mo><mi>|<\/mi><mi>K<\/mi><mo form=\"prefix\" stretchy=\"false\">(<\/mo><mi>u<\/mi><mo form=\"postfix\" stretchy=\"false\">)<\/mo><mi>|<\/mi><mo>&gt;<\/mo><mi>\u03b5<\/mi><mo fence=\"false\" symmetric=\"true\" minsize=\"1.2em\" maxsize=\"1.2em\">]<\/mo><mo fence=\"true\" form=\"postfix\">]<\/mo><\/mrow><mo>+<\/mo><mi>\u03bb<\/mi><mi>\u2016<\/mi><msub><mi>p<\/mi><mi>v<\/mi><\/msub><mo>\u2212<\/mo><msubsup><mi>p<\/mi><mi>v<\/mi><mrow><mo form=\"prefix\" stretchy=\"false\" lspace=\"0em\" rspace=\"0em\">(<\/mo><mn>0<\/mn><mo form=\"postfix\" stretchy=\"false\" lspace=\"0em\" rspace=\"0em\">)<\/mo><\/mrow><\/msubsup><mi>\u2016<\/mi><\/mrow><annotation encoding=\"application\/x-tex\">E(p_v) = |K(v)| + \\sum_{u \\in N(v)} \\left[ |K(u)| + W \\cdot 1\\big[ |K(u)| &gt; \\varepsilon \\big] \\right] + \\lambda \\| p_v &#8211; p_v^{(0)} \\|<\/annotation><\/semantics><\/math><br><br><math><semantics><mrow><mi>|<\/mi><mi>K<\/mi><mo form=\"prefix\" stretchy=\"false\">(<\/mo><mi>v<\/mi><mo form=\"postfix\" stretchy=\"false\">)<\/mo><mi>|<\/mi><\/mrow><annotation encoding=\"application\/x-tex\">|K(v)|<\/annotation><\/semantics><\/math> minimizes the local Gaussian curvature defect at vertex <math><semantics><mi>v<\/mi><annotation encoding=\"application\/x-tex\">v<\/annotation><\/semantics><\/math> to smooth irregularities. <math><semantics><mrow><mo form=\"prefix\" stretchy=\"false\">(<\/mo><msub><mo>\u2211<\/mo><mrow><mi>u<\/mi><mo>\u2208<\/mo><mi>N<\/mi><mo form=\"prefix\" stretchy=\"false\">(<\/mo><mi>v<\/mi><mo form=\"postfix\" stretchy=\"false\" lspace=\"0em\" rspace=\"0em\">)<\/mo><\/mrow><\/msub><mo form=\"postfix\" stretchy=\"false\">)<\/mo><\/mrow><annotation encoding=\"application\/x-tex\">(\\sum_{u \\in N(v)})<\/annotation><\/semantics><\/math> evaluates Gaussian curvature changes on all 1-ring neighbour vertices <math><semantics><mrow><mi>u<\/mi><mo>\u2208<\/mo><mi>N<\/mi><mo form=\"prefix\" stretchy=\"false\">(<\/mo><mi>v<\/mi><mo form=\"postfix\" stretchy=\"false\">)<\/mo><\/mrow><annotation encoding=\"application\/x-tex\">u \\in N(v)<\/annotation><\/semantics><\/math>. <math><semantics><mrow><mo form=\"prefix\" stretchy=\"false\">[<\/mo><mi>|<\/mi><mi>K<\/mi><mo form=\"prefix\" stretchy=\"false\">(<\/mo><mi>u<\/mi><mo form=\"postfix\" stretchy=\"false\">)<\/mo><mi>|<\/mi><mo>&gt;<\/mo><mi>\u03b5<\/mi><mo form=\"postfix\" stretchy=\"false\">]<\/mo><\/mrow><annotation encoding=\"application\/x-tex\">\\mathbb[|K(u)| &gt; \\varepsilon]<\/annotation><\/semantics><\/math> this indicator term applies a penalty <math><semantics><mi>w<\/mi><annotation encoding=\"application\/x-tex\">w<\/annotation><\/semantics><\/math> if a move creates curvature spikes above threshold <math><semantics><mi>\u03b5<\/mi><annotation encoding=\"application\/x-tex\">\\varepsilon<\/annotation><\/semantics><\/math>. The regularization term <math><semantics><mi>\u03bb<\/mi><annotation encoding=\"application\/x-tex\">\\lambda<\/annotation><\/semantics><\/math> penalizes to prevent excessive surface drift.<br><br><strong>Optimization:<\/strong><br><br>                              <math><semantics><mrow><msubsup><mi>p<\/mi><mi>v<\/mi><mtext>opt<\/mtext><\/msubsup><mo>=<\/mo><mrow><munder><mi>arg\u2009min<\/mi><mo>\u2061<\/mo><mspace width=\"0.1667em\"><\/mspace><mrow><mi>p<\/mi><mo>\u2208<\/mo><mo form=\"prefix\" stretchy=\"false\">{<\/mo><msubsup><mi>p<\/mi><mi>v<\/mi><mtext>orig<\/mtext><\/msubsup><mo form=\"postfix\" stretchy=\"false\">}<\/mo><mo>\u222a<\/mo><mo form=\"prefix\" stretchy=\"false\">{<\/mo><msubsup><mi>p<\/mi><mi>v<\/mi><mtext>orig<\/mtext><\/msubsup><mo>+<\/mo><msub><mi>\u03b4<\/mi><mi>i<\/mi><\/msub><mo form=\"postfix\" stretchy=\"false\" lspace=\"0em\" rspace=\"0em\">}<\/mo><\/mrow><\/munder><\/mrow><mi>E<\/mi><mo form=\"prefix\" stretchy=\"false\">(<\/mo><mi>p<\/mi><mo form=\"postfix\" stretchy=\"false\">)<\/mo><\/mrow><annotation encoding=\"application\/x-tex\">p_v^{\\text{opt}} = \\underset{p \\in \\{p_v^{\\text{orig}}\\} \\cup \\{p_v^{\\text{orig}} + \\delta_i\\}}{\\operatorname{arg\\,min}} E(p)<\/annotation><\/semantics><\/math><br><br>The optimization evaluates candidate positions <math><semantics><mi>p<\/mi><annotation encoding=\"application\/x-tex\">p<\/annotation><\/semantics><\/math> from the discrete 3D sampling grid set {<math><semantics><mrow><msubsup><mi>p<\/mi><mi>v<\/mi><mtext>orig<\/mtext><\/msubsup><mo>+<\/mo><msub><mi>\u03b4<\/mi><mi>i<\/mi><\/msub><\/mrow><annotation encoding=\"application\/x-tex\">{p_v^{\\text{orig}} + \\delta_i}<\/annotation><\/semantics><\/math>} alongside the original position <math><semantics><msubsup><mi>p<\/mi><mi>v<\/mi><mtext>orig<\/mtext><\/msubsup><annotation encoding=\"application\/x-tex\">p_v^{\\text{orig}}<\/annotation><\/semantics><\/math>. It greedily selects the candidate position <math><semantics><msubsup><mi>p<\/mi><mi>v<\/mi><mtext>opt<\/mtext><\/msubsup><annotation encoding=\"application\/x-tex\">p_v^{\\text{opt}}<\/annotation><\/semantics><\/math> that minimizes the energy objective <math><semantics><mrow><mi>E<\/mi><mo form=\"prefix\" stretchy=\"false\">(<\/mo><mi>p<\/mi><mo form=\"postfix\" stretchy=\"false\">)<\/mo><\/mrow><annotation encoding=\"application\/x-tex\">E(p)<\/annotation><\/semantics><\/math>, ensuring the orientation protection constraint.<\/p>\n\n\n\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"418\" src=\"https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/08\/step4-1024x418.png\" alt=\"\" class=\"wp-image-1022\" srcset=\"https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/08\/step4-1024x418.png 1024w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/08\/step4-300x122.png 300w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/08\/step4-767x313.png 767w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/08\/step4-1536x627.png 1536w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/08\/step4-1200x490.png 1200w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/08\/step4-2048x836.png 2048w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/08\/step4-1980x809.png 1980w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\"><strong>Figure7: Energy minimization and optimization<\/strong>.<\/figcaption><\/figure>\n\n\n\n\n\n<h2 class=\"wp-block-heading\">Curvature Reduction and Geometry Preservation<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">We compare the four operators from two perspectives: curvature reduction and geometry preservation. Curvature-related metrics evaluate how effectively each operator reduces non-flat regions, while geometry-related metrics measure the corresponding deformation and mesh-quality loss.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Table 1: Curvature-state changes.<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">The transition counts are selected to distinguish corrected curvature violations from newly introduced ones. The final moderate-gray count measures the remaining non-flat vertices, while threshold energy describes the severity of these violations.. <\/p>\n\n\n\n<figure class=\"wp-block-table alignwide is-style-regular\"><table class=\"has-fixed-layout\"><thead><tr><th>Family<\/th><th>Operator<\/th><th>Gray \u2192 Yellow \u2191<\/th><th>Yellow \u2192 Gray \u2193<\/th><th>Net reduction \u2191<\/th><th>Final moderate gray \u2193<\/th><th>Threshold energy \u2193<\/th><\/tr><\/thead><tbody><tr><td>Vertex-based<\/td><td>Normal one-ring<\/td><td><strong>432<\/strong><\/td><td>55<\/td><td><strong>377<\/strong><\/td><td><strong>156<\/strong><\/td><td><strong>0.676<\/strong><\/td><\/tr><tr><td>Vertex-based<\/td><td>Double cone<\/td><td>369<\/td><td>38<\/td><td>331<\/td><td>202<\/td><td>3.659<\/td><\/tr><tr><td>Vertex-based<\/td><td>Virtual sphere<\/td><td>349<\/td><td><strong>29<\/strong><\/td><td>320<\/td><td>213<\/td><td>3.782<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n<figure class=\"wp-block-image alignwide size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"576\" src=\"https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/07\/screenshot_000001-1-1024x576.png\" alt=\"\" class=\"wp-image-302\" srcset=\"https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/07\/screenshot_000001-1-1024x576.png 1024w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/07\/screenshot_000001-1-300x169.png 300w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/07\/screenshot_000001-1-768x432.png 768w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/07\/screenshot_000001-1-1536x865.png 1536w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/07\/screenshot_000001-1-2048x1153.png 2048w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/07\/screenshot_000001-1-1200x675.png 1200w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/07\/screenshot_000001-1-1980x1114.png 1980w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">Figure8: From left to right: the original mesh, normal one-ring operator, double-cone operator, virtual-sphere operator, and 3D-grid operator.<\/figcaption><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Table2: Curvature state transition  in Algorithm 2<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">We evaluated the proposed 3D grid based algorithm on five models namely Cheburashka, Fandisk, Stanford-bunny, Homer, Suzanne. The optimal threshold epsilon was set to 0.1 and the penalty was fixed at 50. The table summrizes the before and after curvature counts with the algorithm run time. <\/p>\n\n\n\n<figure class=\"wp-block-table alignwide\"><table><thead><tr><th class=\"has-text-align-center\" data-align=\"center\">Model<\/th><th>Epsilon<\/th><th>Penalty<\/th><th class=\"has-text-align-center\" data-align=\"center\">Flat Vertices<br>(Before)<\/th><th class=\"has-text-align-center\" data-align=\"center\">Non-Flat Vertices<br>(Before)<\/th><th class=\"has-text-align-center\" data-align=\"center\">Flat Vertices<br>(After)<\/th><th class=\"has-text-align-center\" data-align=\"center\">Non-Flat Vertices<br>(After)<\/th><th class=\"has-text-align-center\" data-align=\"center\">Flat Vertex<br>Gain<\/th><th class=\"has-text-align-center\" data-align=\"center\">Run time<\/th><\/tr><\/thead><tbody><tr><td class=\"has-text-align-center\" data-align=\"center\">Cheburashka<\/td><td>0.1<\/td><td>50<\/td><td class=\"has-text-align-center\" data-align=\"center\">6293<\/td><td class=\"has-text-align-center\" data-align=\"center\">376<\/td><td class=\"has-text-align-center\" data-align=\"center\">6411<\/td><td class=\"has-text-align-center\" data-align=\"center\">258<\/td><td class=\"has-text-align-center\" data-align=\"center\">+118<\/td><td class=\"has-text-align-center\" data-align=\"center\">308.198s<\/td><\/tr><tr><td class=\"has-text-align-center\" data-align=\"center\">Fandisk<\/td><td>0.1<\/td><td>50<\/td><td class=\"has-text-align-center\" data-align=\"center\">6427<\/td><td class=\"has-text-align-center\" data-align=\"center\">48<\/td><td class=\"has-text-align-center\" data-align=\"center\">6429<\/td><td class=\"has-text-align-center\" data-align=\"center\">46<\/td><td class=\"has-text-align-center\" data-align=\"center\">+2<\/td><td class=\"has-text-align-center\" data-align=\"center\">58.310s<\/td><\/tr><tr><td class=\"has-text-align-center\" data-align=\"center\">Stanford-bunny<\/td><td>0.1<\/td><td>50<\/td><td class=\"has-text-align-center\" data-align=\"center\">34071<\/td><td class=\"has-text-align-center\" data-align=\"center\">540<\/td><td class=\"has-text-align-center\" data-align=\"center\">34444<\/td><td class=\"has-text-align-center\" data-align=\"center\">167<\/td><td class=\"has-text-align-center\" data-align=\"center\">+373<\/td><td class=\"has-text-align-center\" data-align=\"center\">410.793s<\/td><\/tr><tr><td class=\"has-text-align-center\" data-align=\"center\">Homer<\/td><td>0.1<\/td><td>50<\/td><td class=\"has-text-align-center\" data-align=\"center\">5313<\/td><td class=\"has-text-align-center\" data-align=\"center\">689<\/td><td class=\"has-text-align-center\" data-align=\"center\">5526<\/td><td class=\"has-text-align-center\" data-align=\"center\">476<\/td><td class=\"has-text-align-center\" data-align=\"center\">+213<\/td><td class=\"has-text-align-center\" data-align=\"center\">535.254s<\/td><\/tr><tr><td class=\"has-text-align-center\" data-align=\"center\">Suzanne<\/td><td>0.1<\/td><td>50<\/td><td class=\"has-text-align-center\" data-align=\"center\">115<\/td><td class=\"has-text-align-center\" data-align=\"center\">350<\/td><td class=\"has-text-align-center\" data-align=\"center\">193<\/td><td class=\"has-text-align-center\" data-align=\"center\">272<\/td><td class=\"has-text-align-center\" data-align=\"center\">+78<\/td><td class=\"has-text-align-center\" data-align=\"center\">317.417s<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n<figure class=\"wp-block-image alignwide size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"544\" src=\"https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/10\/cheburaska-1-1024x544.png\" alt=\"\" class=\"wp-image-1389\" srcset=\"https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/10\/cheburaska-1-1024x544.png 1024w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/10\/cheburaska-1-300x160.png 300w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/10\/cheburaska-1-767x408.png 767w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/10\/cheburaska-1-1200x638.png 1200w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/10\/cheburaska-1-1536x817.png 1536w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/10\/cheburaska-1.png 1907w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">Figure9: Visualization of the Cheburashka before(left) and after(right) applying the algorithm. Red, yellow, and blue color point out non-flat, flat, and regions converted non-flat to flat.<\/figcaption><\/figure>\n\n\n\n<figure class=\"wp-block-image alignwide size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"535\" src=\"https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/10\/fandisk-2-1024x535.png\" alt=\"\" class=\"wp-image-1391\" srcset=\"https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/10\/fandisk-2-1024x535.png 1024w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/10\/fandisk-2-300x157.png 300w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/10\/fandisk-2-768x401.png 768w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/10\/fandisk-2-1200x627.png 1200w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/10\/fandisk-2-1536x802.png 1536w, https:\/\/summergeometry.org\/sgi2026\/wp-content\/uploads\/2026\/10\/fandisk-2.png 1907w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">Figure10: Visualization of the Fandisk before(right) and after(left) applying the algorithm. Red, yellow, and blue color point out non-flat, flat, and regions converted non-flat to flat.<\/figcaption><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Table 3: Geometry preservation.<\/strong><br>Normalized RMS displacement measures the overall surface change relative to the model size, while maximum edge change captures the worst local edge-length distortion. Minimum area ratio measures the strongest triangle compression, with values closer to one indicating better area preservation. Minimum triangle quality ranges from zero to one, where values near zero indicate thin or nearly degenerate triangles. Maximum normal change measures the largest local surface-orientation deviation, and the flipped-face count detects invalid face inversions.<\/p>\n\n\n\n<figure class=\"wp-block-table alignwide\"><table class=\"has-fixed-layout\"><thead><tr><th>Family<\/th><th class=\"has-text-align-center\" data-align=\"center\">Operator<\/th><th>Normalized RMS \u2193<\/th><th>Max edge change \u2193<\/th><th>Min area ratio \u2191<\/th><th>Min quality \u2191<\/th><th>Max normal change \u2193<\/th><th>Flipped faces<\/th><\/tr><\/thead><tbody><tr><td>Vertex-based<\/td><td class=\"has-text-align-center\" data-align=\"center\">Normal one-ring<\/td><td>0.00569%<\/td><td>24.71%<\/td><td>0.304<\/td><td>0.00841<\/td><td><strong>28.29\u00b0<\/strong><\/td><td>0<\/td><\/tr><tr><td>Vertex-based<\/td><td class=\"has-text-align-center\" data-align=\"center\">Double cone<\/td><td><strong>0.00373%<\/strong><\/td><td><strong>19.14%<\/strong><\/td><td>0.313<\/td><td><strong>0.01390<\/strong><\/td><td>29.19\u00b0<\/td><td>0<\/td><\/tr><tr><td>Vertex-based<\/td><td class=\"has-text-align-center\" data-align=\"center\">Virtual sphere<\/td><td>0.00407%<\/td><td>23.97%<\/td><td>0.594<\/td><td><strong>0.01390<\/strong><\/td><td>73.81\u00b0<\/td><td>0<\/td><\/tr><tr><td>Grid-based<\/td><td class=\"has-text-align-center\" data-align=\"center\">3D grid<\/td><td>0.01315%<\/td><td>21.99%<\/td><td>0.351<\/td><td>0.01392<\/td><td>30.00\u00b0<\/td><td>0<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Overall result:<\/strong><br>The results indicate no operator dominates across all criteria. The double cone achieved the highest normalized RMS and maximum edge change, with a competitive minimum quality. The virtual sphere preserves triangle area best, but at the cost of a much larger maximum normal change. The normal one ring yields the smallest normal deviation but the lowest minimum quality and the largest edge change. The 3D grid is competitive in quality and normal change but has the highest RMS displacement. All the operators produced zero flipped faces.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Future directions:<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Extending these algorithm for real time use will likely require cost reduction strategies such as principal component analysis (PCA), proper orthogonal decomposition (POD).<\/li>\n\n\n\n<li>The evaluation covers geometric fidelity and mesh validity only. A systematic study of memory scaling and computational cost reduction strategies is a natural direction for future work.<\/li>\n\n\n\n<li>Testing on more complex models is needed to check the generalizability of the algorithms.<\/li>\n<\/ul>\n\n\n\n<h2 class=\"wp-block-heading\">Reference:<\/h2>\n\n\n\n<ul class=\"wp-block-list\">\n<li>R. H. Byrd, P. Lu, J. Nocedal, and C. Zhu, <em>A Limited Memory Algorithm for Bound Constrained Optimization,<\/em> pp. 1190&#8211;1208, 1995. <a href=\"https:\/\/doi.org\/10.1137\/0916069\">https:\/\/doi.org\/10.1137\/0916069<\/a><\/li>\n\n\n\n<li>L. Shapira, A. Shamir, and D. Cohen-Or, <em>Consistent Mesh Partitioning and Skeletonisation using the Shape Diameter Function<\/em>, vol. 24, no. 4, pp. 249&#8211;259, 2008. <a href=\"https:\/\/doi.org\/10.1007\/s00371-007-0197-5\">https:\/\/doi.org\/10.1007\/s00371-007-0197-5<\/a><\/li>\n\n\n\n<li>B. Keinert, M. Innmann, M. S\\&#8221;anger, and M. Stamminger, <em>Spherical Fibonacci Mapping<\/em>, vol. 34, no. 6, article 193, 2015. <a href=\"https:\/\/doi.org\/10.1145\/2816795.2818131\">https:\/\/doi.org\/10.1145\/2816795.2818131<\/a><\/li>\n\n\n\n<li>D. Zint, R. Grosso, and F. Lunz, \u201cDiscrete mesh optimization on surface and volume meshes,\u201d in <em>Proc. 28th Int. Meshing Roundtable (IMR)<\/em>, 2019, pp. 159\u2013174. <\/li>\n\n\n\n<li>A. Nealen, T. Igarashi, O. Sorkine, and M. Alexa, \u201cLaplacian mesh optimization,\u201d in <em>Proc. 4th Int. Conf. Computer Graphics and Interactive Techniques in Australasia and Southeast Asia (GRAPHITE \u201806)<\/em>, 2006, pp. 381\u2013389.<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\"><\/p>\n","protected":false},"excerpt":{"rendered":"<p>Fellows: Yufan Hu, Robiul Islam Hannan, Kabir SahniMentor: Yusuf SahilliogluVolunteer: Vivien van Veldhuizen Introduction: Developable surfaces are surfaces that can be unfolded onto a plane without stretching, compression, or other geometric distortion. They are relevant to applications in geometric design, digital fabrication, manufacturing, architecture, and related areas where three-dimensional geometry must be represented or fabricated [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[110],"tags":[],"ppma_author":[18,104],"class_list":["post-173","post","type-post","status-publish","format-standard","hentry","category-research"],"authors":[{"term_id":18,"user_id":0,"is_guest":1,"slug":"cap-robiulislamhannan019","display_name":"robiulislamhannan019","avatar_url":"https:\/\/secure.gravatar.com\/avatar\/?s=96&d=mm&r=g","author_category":"","first_name":"","last_name":"","user_url":"","job_title":"","description":""},{"term_id":104,"user_id":0,"is_guest":1,"slug":"cap-yhu1","display_name":"yhu1","avatar_url":"https:\/\/secure.gravatar.com\/avatar\/?s=96&d=mm&r=g","author_category":"","first_name":"","last_name":"","user_url":"","job_title":"","description":""}],"_links":{"self":[{"href":"https:\/\/summergeometry.org\/sgi2026\/wp-json\/wp\/v2\/posts\/173","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/summergeometry.org\/sgi2026\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/summergeometry.org\/sgi2026\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/summergeometry.org\/sgi2026\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/summergeometry.org\/sgi2026\/wp-json\/wp\/v2\/comments?post=173"}],"version-history":[{"count":9,"href":"https:\/\/summergeometry.org\/sgi2026\/wp-json\/wp\/v2\/posts\/173\/revisions"}],"predecessor-version":[{"id":1427,"href":"https:\/\/summergeometry.org\/sgi2026\/wp-json\/wp\/v2\/posts\/173\/revisions\/1427"}],"wp:attachment":[{"href":"https:\/\/summergeometry.org\/sgi2026\/wp-json\/wp\/v2\/media?parent=173"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/summergeometry.org\/sgi2026\/wp-json\/wp\/v2\/categories?post=173"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/summergeometry.org\/sgi2026\/wp-json\/wp\/v2\/tags?post=173"},{"taxonomy":"author","embeddable":true,"href":"https:\/\/summergeometry.org\/sgi2026\/wp-json\/wp\/v2\/ppma_author?post=173"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}