/********************************************************************** * * Sokol 3d cube demo * * Copyright (c) 2021 Dario Deledda. All rights reserved. * Use of this source code is governed by an MIT license * that can be found in the LICENSE file. * * HOW TO COMPILE SHADERS: * - download the sokol shader convertor tool from https://github.com/floooh/sokol-tools-bin/archive/pre-feb2021-api-changes.tar.gz * ( also look at https://github.com/floooh/sokol-tools/blob/master/docs/sokol-shdc.md ) * - compile the .glsl shader with: * linux : sokol-shdc --input rt_glsl.glsl --output rt_glsl.h --slang glsl330 * windows: sokol-shdc.exe --input rt_glsl.glsl --output rt_glsl.h --slang glsl330 * * --slang parameter can be: * - glsl330: desktop GL * - glsl100: GLES2 / WebGL * - glsl300es: GLES3 / WebGL2 * - hlsl4: D3D11 * - hlsl5: D3D11 * - metal_macos: Metal on macOS * - metal_ios: Metal on iOS device * - metal_sim: Metal on iOS simulator * - wgpu: WebGPU * * you can have multiple platforms at the same time passing parameters like this: --slang glsl330:hlsl5:metal_macos * for further infos have a look at the sokol shader tool docs. * * TODO: * - frame counter **********************************************************************/ import gg import gg.m4 import gx // import math import sokol.sapp import sokol.gfx import sokol.sgl import time // GLSL Include and functions #flag -I @VROOT/. #include "rt_glsl.h" #Please use sokol-shdc to generate the necessary rt_glsl.h file from rt_glsl.glsl (see the instructions at the top of this file) fn C.rt_shader_desc(gfx.Backend) &C.sg_shader_desc const ( win_width = 800 win_height = 800 bg_color = gx.white ) struct App { mut: gg &gg.Context texture C.sg_image init_flag bool frame_count int mouse_x int = -1 mouse_y int = -1 // glsl cube_pip_glsl C.sg_pipeline cube_bind C.sg_bindings // time ticks i64 } /****************************************************************************** * Texture functions ******************************************************************************/ fn create_texture(w int, h int, buf &byte) C.sg_image { sz := w * h * 4 mut img_desc := C.sg_image_desc{ width: w height: h num_mipmaps: 0 min_filter: .linear mag_filter: .linear // usage: .dynamic wrap_u: .clamp_to_edge wrap_v: .clamp_to_edge label: &byte(0) d3d11_texture: 0 } // comment if .dynamic is enabled img_desc.data.subimage[0][0] = C.sg_range{ ptr: buf size: size_t(sz) } sg_img := C.sg_make_image(&img_desc) return sg_img } fn destroy_texture(sg_img C.sg_image) { C.sg_destroy_image(sg_img) } // Use only if usage: .dynamic is enabled fn update_text_texture(sg_img C.sg_image, w int, h int, buf &byte) { sz := w * h * 4 mut tmp_sbc := C.sg_image_data{} tmp_sbc.subimage[0][0] = C.sg_range{ ptr: buf size: size_t(sz) } C.sg_update_image(sg_img, &tmp_sbc) } /****************************************************************************** * Draw functions ****************************************************************************** Cube vertex buffer with packed vertex formats for color and texture coords. Note that a vertex format which must be portable across all backends must only use the normalized integer formats (BYTE4N, UBYTE4N, SHORT2N, SHORT4N), which can be converted to floating point formats in the vertex shader inputs. The reason is that D3D11 cannot convert from non-normalized formats to floating point inputs (only to integer inputs), and WebGL2 / GLES2 don't support integer vertex shader inputs. */ struct Vertex_t { x f32 y f32 z f32 color u32 u f32 v f32 // u u16 // for compatibility with D3D11 // v u16 // for compatibility with D3D11 } fn init_cube_glsl(mut app App) { // cube vertex buffer // d := u16(32767) // for compatibility with D3D11, 32767 stand for 1 d := f32(1.0) c := u32(0xFFFFFF_FF) // color RGBA8 vertices := [ // Face 0 Vertex_t{-1.0, -1.0, -1.0, c, 0, 0}, Vertex_t{ 1.0, -1.0, -1.0, c, d, 0}, Vertex_t{ 1.0, 1.0, -1.0, c, d, d}, Vertex_t{-1.0, 1.0, -1.0, c, 0, d}, // Face 1 Vertex_t{-1.0, -1.0, 1.0, c, 0, 0}, Vertex_t{ 1.0, -1.0, 1.0, c, d, 0}, Vertex_t{ 1.0, 1.0, 1.0, c, d, d}, Vertex_t{-1.0, 1.0, 1.0, c, 0, d}, // Face 2 Vertex_t{-1.0, -1.0, -1.0, c, 0, 0}, Vertex_t{-1.0, 1.0, -1.0, c, d, 0}, Vertex_t{-1.0, 1.0, 1.0, c, d, d}, Vertex_t{-1.0, -1.0, 1.0, c, 0, d}, // Face 3 Vertex_t{ 1.0, -1.0, -1.0, c, 0, 0}, Vertex_t{ 1.0, 1.0, -1.0, c, d, 0}, Vertex_t{ 1.0, 1.0, 1.0, c, d, d}, Vertex_t{ 1.0, -1.0, 1.0, c, 0, d}, // Face 4 Vertex_t{-1.0, -1.0, -1.0, c, 0, 0}, Vertex_t{-1.0, -1.0, 1.0, c, d, 0}, Vertex_t{ 1.0, -1.0, 1.0, c, d, d}, Vertex_t{ 1.0, -1.0, -1.0, c, 0, d}, // Face 5 Vertex_t{-1.0, 1.0, -1.0, c, 0, 0}, Vertex_t{-1.0, 1.0, 1.0, c, d, 0}, Vertex_t{ 1.0, 1.0, 1.0, c, d, d}, Vertex_t{ 1.0, 1.0, -1.0, c, 0, d}, ] mut vert_buffer_desc := C.sg_buffer_desc{} unsafe { C.memset(&vert_buffer_desc, 0, sizeof(vert_buffer_desc)) } vert_buffer_desc.size = size_t(vertices.len * int(sizeof(Vertex_t))) vert_buffer_desc.data = C.sg_range{ ptr: vertices.data size: size_t(vertices.len * int(sizeof(Vertex_t))) } vert_buffer_desc.@type = .vertexbuffer vert_buffer_desc.label = 'cube-vertices'.str vbuf := gfx.make_buffer(&vert_buffer_desc) // create an index buffer for the cube indices := [ u16(0), 1, 2, 0, 2, 3, 6, 5, 4, 7, 6, 4, 8, 9, 10, 8, 10, 11, 14, 13, 12, 15, 14, 12, 16, 17, 18, 16, 18, 19, 22, 21, 20, 23, 22, 20, ] mut index_buffer_desc := C.sg_buffer_desc{} unsafe {C.memset(&index_buffer_desc, 0, sizeof(index_buffer_desc))} index_buffer_desc.size = size_t(indices.len * int(sizeof(u16))) index_buffer_desc.data = C.sg_range{ ptr: indices.data size: size_t(indices.len * int(sizeof(u16))) } index_buffer_desc.@type = .indexbuffer index_buffer_desc.label = "cube-indices".str ibuf := gfx.make_buffer(&index_buffer_desc) // create shader shader := gfx.make_shader(C.rt_shader_desc(C.sg_query_backend())) mut pipdesc := C.sg_pipeline_desc{} unsafe { C.memset(&pipdesc, 0, sizeof(pipdesc)) } pipdesc.layout.buffers[0].stride = int(sizeof(Vertex_t)) // the constants [C.ATTR_vs_pos, C.ATTR_vs_color0, C.ATTR_vs_texcoord0] are generated by sokol-shdc pipdesc.layout.attrs[C.ATTR_vs_pos ].format = .float3 // x,y,z as f32 pipdesc.layout.attrs[C.ATTR_vs_color0 ].format = .ubyte4n // color as u32 pipdesc.layout.attrs[C.ATTR_vs_texcoord0].format = .float2 // u,v as f32 // pipdesc.layout.attrs[C.ATTR_vs_texcoord0].format = .short2n // u,v as u16 pipdesc.shader = shader pipdesc.index_type = .uint16 pipdesc.depth = C.sg_depth_state{ write_enabled: true compare: gfx.CompareFunc(C.SG_COMPAREFUNC_LESS_EQUAL) } pipdesc.cull_mode = .back pipdesc.label = 'glsl_shader pipeline'.str app.cube_bind.vertex_buffers[0] = vbuf app.cube_bind.index_buffer = ibuf app.cube_bind.fs_images[C.SLOT_tex] = app.texture app.cube_pip_glsl = gfx.make_pipeline(&pipdesc) println('GLSL init DONE!') } [inline] fn vec4(x f32, y f32, z f32, w f32) m4.Vec4 { return m4.Vec4{e:[x, y, z, w]!} } fn calc_tr_matrices(w f32, h f32, rx f32, ry f32, in_scale f32) m4.Mat4 { proj := m4.perspective(60, w/h, 0.01, 10.0) view := m4.look_at(vec4(f32(0.0) ,0 , 6, 0), vec4(f32(0), 0, 0, 0), vec4(f32(0), 1, 0, 0)) view_proj := view * proj rxm := m4.rotate(m4.rad(rx), vec4(f32(1), 0, 0, 0)) rym := m4.rotate(m4.rad(ry), vec4(f32(0), 1, 0, 0)) model := rym * rxm scale_m := m4.scale(vec4(in_scale, in_scale, in_scale, 1)) res := (scale_m * model) * view_proj return res } fn draw_cube_glsl(app App) { if app.init_flag == false { return } ws := gg.window_size_real_pixels() ratio := f32(ws.width) / ws.height dw := f32(ws.width / 2) dh := f32(ws.height / 2) // use the following commented lines to rotate the 3d glsl cube // rot := [f32(app.mouse_y), f32(app.mouse_x)] // calc_tr_matrices(dw, dh, rot[0], rot[1] ,2.3) tr_matrix := calc_tr_matrices(dw, dh, 0, 0, 2.3) gfx.apply_viewport(0, 0, ws.width, ws.height, true) // apply the pipline and bindings gfx.apply_pipeline(app.cube_pip_glsl) gfx.apply_bindings(app.cube_bind) // Uniforms // *** vertex shadeer uniforms *** // passing the view matrix as uniform // res is a 4x4 matrix of f32 thus: 4*16 byte of size vs_uniforms_range := C.sg_range{ ptr: &tr_matrix size: size_t(4 * 16) } gfx.apply_uniforms(C.SG_SHADERSTAGE_VS, C.SLOT_vs_params, &vs_uniforms_range) // *** fragment shader uniforms *** time_ticks := f32(time.ticks() - app.ticks) / 1000 mut tmp_fs_params := [ f32(ws.width), ws.height * ratio, // x,y resolution to pass to FS app.mouse_x, // mouse x ws.height - app.mouse_y * 2, // mouse y scaled time_ticks, // time as f32 app.frame_count, // frame count 0, 0 // padding bytes , see "fs_params" struct paddings in rt_glsl.h ]! fs_uniforms_range := C.sg_range{ ptr: &tmp_fs_params size: size_t(sizeof(tmp_fs_params)) } gfx.apply_uniforms(C.SG_SHADERSTAGE_FS, C.SLOT_fs_params, &fs_uniforms_range) // 3 vertices for triangle * 2 triangles per face * 6 faces = 36 vertices to draw gfx.draw(0, (3 * 2) * 6, 1) gfx.end_pass() gfx.commit() } fn frame(mut app App) { ws := gg.window_size_real_pixels() // clear mut color_action := C.sg_color_attachment_action{ action: gfx.Action(C.SG_ACTION_CLEAR) value: C.sg_color{ r: 0.0 g: 0.0 b: 0.0 a: 1.0 } } mut pass_action := C.sg_pass_action{} pass_action.colors[0] = color_action gfx.begin_default_pass(&pass_action, ws.width, ws.height) // glsl cube draw_cube_glsl(app) app.frame_count++ } /****************************************************************************** * Init / Cleanup ******************************************************************************/ fn my_init(mut app App) { // set max vertices, // for a large number of the same type of object it is better use the instances!! desc := sapp.create_desc() gfx.setup(&desc) sgl_desc := C.sgl_desc_t{ max_vertices: 50 * 65536 } sgl.setup(&sgl_desc) // create chessboard texture 256*256 RGBA w := 256 h := 256 sz := w * h * 4 tmp_txt := unsafe { malloc(sz) } mut i := 0 for i < sz { unsafe { y := (i >> 0x8) >> 5 // 8 cell x := (i & 0xFF) >> 5 // 8 cell // upper left corner if x == 0 && y == 0 { tmp_txt[i + 0] = byte(0xFF) tmp_txt[i + 1] = byte(0) tmp_txt[i + 2] = byte(0) tmp_txt[i + 3] = byte(0xFF) } // low right corner else if x == 7 && y == 7 { tmp_txt[i + 0] = byte(0) tmp_txt[i + 1] = byte(0xFF) tmp_txt[i + 2] = byte(0) tmp_txt[i + 3] = byte(0xFF) } else { col := if ((x + y) & 1) == 1 { 0xFF } else { 128 } tmp_txt[i + 0] = byte(col) // red tmp_txt[i + 1] = byte(col) // green tmp_txt[i + 2] = byte(col) // blue tmp_txt[i + 3] = byte(0xFF) // alpha } i += 4 } } unsafe { app.texture = create_texture(w, h, tmp_txt) free(tmp_txt) } // glsl init_cube_glsl(mut app) app.init_flag = true } fn cleanup(mut app App) { gfx.shutdown() } /****************************************************************************** * events handling ******************************************************************************/ fn my_event_manager(mut ev gg.Event, mut app App) { if ev.typ == .mouse_move { app.mouse_x = int(ev.mouse_x) app.mouse_y = int(ev.mouse_y) } if ev.typ == .touches_began || ev.typ == .touches_moved { if ev.num_touches > 0 { touch_point := ev.touches[0] app.mouse_x = int(touch_point.pos_x) app.mouse_y = int(touch_point.pos_y) } } } /****************************************************************************** * Main ******************************************************************************/ [console] // is needed for easier diagnostics on windows fn main() { // App init mut app := &App{ gg: 0 } app.gg = gg.new_context( width: win_width height: win_height use_ortho: true // This is needed for 2D drawing create_window: true window_title: '3D Ray Marching Cube' user_data: app bg_color: bg_color frame_fn: frame init_fn: my_init cleanup_fn: cleanup event_fn: my_event_manager ) app.ticks = time.ticks() app.gg.run() }