pardes.git/shaders/post/crt.glsl, branch release-0.24

Rules and chrome scale with the display's density; dark pages get visible separators and focus tints

2026-10-01T03:12:16Z

Density: dp() in gui.zig draws every size given in pixels in logical pixels (SDL's window display scale, rounded, at least one), as acme scales Border, ButtonBorder and Scrollgap: the rules (rule_px), the tag rule, the grip's ring, its dirty gap and mode bars, the rail's gap to the text and its minimum, the thumb's inset, the rim, a bar cursor's minimum width, underline and strikethrough thickness (the atlas is dropped on a scale change), the rule-grab margin, and the Crt's scanline and mask periods (pardesScale in the post prefix). Cell- and tagline-measured sizes stay: the rail's width, Lift's blur and offsets, every motion; iResolution, fragCoord and iMouse stay in the display's pixels, as ghostty's. PARDES_TEST_SCALE forces a density in a harness (a test window is 1x otherwise, whatever display it lands on) and sizes the test font with it. Colours: a separator that stands off its page by less than 1.5:1 is lifted toward the text to 1.6:1 (helix, atelier, paper; a theme with no page, dark, is judged against #121212), and empty_col follows it. forge's own rule already stands at 3.83:1; what hid it was density: 2 device pixels on a 3x screen, now 6 (dp). A focused tag's tint closer than 1.5:1 to the plain tag on a dark page (1.25:1 on a light one) is pushed further the way it leans, as far as its text keeps its own contrast or 4.5: forge 1.23 -> 1.51 (#2d3440 -> #374253), and likewise dusk, orchard, ink, lagoon, clay, harvest, spectrum, solarium, forge_soft; daybreak, atelier, paper to 1.25. dark and helix gain a tint of their own (dark: #484566 under #c8c8c8, 1.54:1; helix: ramp 237 under 250, 1.65:1). Tag caret: past a full row's end, the G3 glide box sits on the tag's right edge, where the layer's caret and the pixel shell's are. Tests: every native theme's separator >= 1.5 and every dark one has a tint >= 1.25 with readable text; forge keeps its own border; the tag caret glide box at a full row; at 2x the spine, tag rule, ring, rail start and gap, and bar cursor double. Legacy-ThemeFile test strips helix's new pair. Goldens: snapshot goldens change in colour rows only, the tints (orchard #283a2f -> #314738 at boot, forge, helix and dark in theme/themesel). GUI goldens: the harness's single Del of the renamed wide pane is refused since ysqnkyoy (a rename counted as unsaved; fixed on the 9P line in wkzlsqvy), so scenes 09-17 were recorded with that pane open and the file was stale since the merge (08 and 18 also moved with ykvnltvs's full-row tag wrap). The goldens here were recorded with that pane closed (as wkzlsqvy will close it); against the parent's binary under the same conditions they differ only in the tints: 01-13 forge's (#2d3440 -> #374253, and the glyph edges on it), 14 ink's; 15-18 are byte-identical. Until wkzlsqvy is merged, a run here fails 09-17. Each commit reproduces its own recorded goldens on this machine (checked at kvtmqkyn). A test window is 1x unless PARDES_TEST_SCALE says otherwise, whatever display it opens on. Prebuilt shaders refreshed from their sources. Captures: .scratch/render/dark-borders/{before,after,compare,scale2,scale3}.

Crt, Ripple and Glitch take a level, 0 to 3, tunable at run time

2026-10-01T03:12:15Z

Coordinator's request: `Crt 0..3` (likewise Ripple, Glitch), 0 off, `on` the default level 2 (the polished rewrite), `off` and a bare word as before. The level is one uniform (pardesLevel, beside pardesOpaque in the prefix's own block); each pass picks its constants by it. Level 3 is at least as strong as the old effects, measured as pixels changed by more than 8 in a channel against a plain frame: Crt 18.3% (old 17.4%), Ripple 11.1% (9.5%), Glitch 9.3% a frame (8.1%); Glitch's level raises how often it bursts as well as how hard. Config and ctl report `Crt 2` / `Crt off`. Shared files touched: ninep/ctl.zig (the report line). Not touched: pardes.zig, Messages.zig, mouse.zig, gui.zig, detached/*.

Post chain: Shadertoy passes compiled by glslc; Crt, Ripple and Glitch rewritten as Shadertoy files

2026-10-01T03:12:15Z

Stage 9 of docs/render-pipeline.md. src/gui/Post.zig runs the chain over the finished frame: each pass reads the one before, ping-pong textures between, the last writes the window; an empty chain is the direct path. The prefix (shaders/post/prefix.glsl) is ghostty's uniform block, name for name and offset for offset (tested against a copy of its Uniforms, the selection colours by their GLSL names), at SDL GPU's bindings, y down. A user's file (Shader ) is compiled by glslc with compileGlsl's own flags on a thread of its own, taken in at the next loop step; errors name the file's own lines; a failed compile keeps the last good pipeline; glslc missing is said once. The bundled passes compile with the build, through the same prefix. Crt, Ripple and Glitch are Shadertoy files, rewritten: no barrel (input is identity, tested), everything in linear light and dithered, keyed to iTime. scene_effects, crt.zig and crt.frag.glsl are gone; the core keeps no clock for the chain. ShaderAnimation off|on|always drives redraw level A: the chain alone over the retained frame (34 us CPU a redraw, measured). Shared files touched: pardes.zig (nextWake loses the scene line, disableSceneEffects empties the chain, two tests), builtins.zig (one switch arm), ninep/ctl.zig (two switch arms), macos.zig (flags from the chain), gui.zig. Not touched: Messages.zig, mouse.zig, detached/*, host_io, tty.