makepad

History

vjroger 093cc2e31a material_lab: a row of spheres above the cubes, and the scene projected at the widget's aspect -- a cube shows a light only on the faces it turns toward it (the rect light reached only a cube's thin top face until e0956b4d3 moved it), while a sphere has every normal, so a light's falloff, the environment's reflection, a rim or a highlight always shows on it. Every scene now puts a sphere with its column's material above each cube (y 1.35..2.35; resident geometry for the world scenes, a GLB model for the hook row, one smooth mesh of shared vertices wound as each cube is), the rect scene gives the sphere its own light at the cube light's place (each out of the other row's reach), and the camera aims at y 1.15 to frame both rows. The 800x300 pass is drawn stretched over the widget, which on Windows is 800x271 under the in-window title bar, so shapes were 10% wide; the projection now takes the widget's aspect and spheres are round. The pixel tests split each column at the background gap between the rows (found in the Unlit column) and assert every check on the cube and on the sphere, and place columns by pixels per metre, the same across as down now. Checked: cargo test -p makepad-example-material-lab on D3D11 (3 passed; rect column 0 means 130.9 cube and 88.5 sphere, column 1 1.7 and 0.8; IBL metal spheres 53), cargo check --tests for x86_64-apple-darwin		2026-10-03 15:01:17 +02:00
..
pixels.rs	material_lab: a row of spheres above the cubes, and the scene projected at the widget's aspect -- a cube shows a light only on the faces it turns toward it (the rect light reached only a cube's thin top face until e0956b4d3 moved it), while a sphere has every normal, so a light's falloff, the environment's reflection, a rim or a highlight always shows on it. Every scene now puts a sphere with its column's material above each cube (y 1.35..2.35; resident geometry for the world scenes, a GLB model for the hook row, one smooth mesh of shared vertices wound as each cube is), the rect scene gives the sphere its own light at the cube light's place (each out of the other row's reach), and the camera aims at y 1.15 to frame both rows. The 800x300 pass is drawn stretched over the widget, which on Windows is 800x271 under the in-window title bar, so shapes were 10% wide; the projection now takes the widget's aspect and spheres are round. The pixel tests split each column at the background gap between the rows (found in the Unlit column) and assert every check on the cube and on the sphere, and place columns by pixels per metre, the same across as down now. Checked: cargo test -p makepad-example-material-lab on D3D11 (3 passed; rect column 0 means 130.9 cube and 88.5 sphere, column 1 1.7 and 0.8; IBL metal spheres 53), cargo check --tests for x86_64-apple-darwin	2026-10-03 15:01:17 +02:00