fix(OpenGL): use sRGB texture format for gamma correct interpolation

otherwise images will be too dark when scaled

pkg/opengl/Texture.zig

@@ -70,6 +70,9 @@ pub const InternalFormat = enum(c_int) {
     rgb = c.GL_RGB,
     rgba = c.GL_RGBA,
 
+    srgb = c.GL_SRGB,
+    srgba = c.GL_SRGB_ALPHA,
+
     // There are so many more that I haven't filled in.
     _,
 };

src/renderer/opengl/image.zig

@@ -368,8 +368,8 @@ pub const Image = union(enum) {
             internal: gl.Texture.InternalFormat,
             format: gl.Texture.Format,
         } = switch (self.*) {
-            .pending_rgb, .replace_rgb => .{ .internal = .rgb, .format = .rgb },
+            .pending_rgb, .replace_rgb => .{ .internal = .srgb, .format = .rgb },
-            .pending_rgba, .replace_rgba => .{ .internal = .rgba, .format = .rgba },
+            .pending_rgba, .replace_rgba => .{ .internal = .srgba, .format = .rgba },
             else => unreachable,
         };
 

src/renderer/shaders/image.f.glsl

@@ -6,7 +6,24 @@ layout(location = 0) out vec4 out_FragColor;
 
 uniform sampler2D image;
 
+// Converts a color from linear to sRGB gamma encoding.
+vec4 unlinearize(vec4 linear) {
+    bvec3 cutoff = lessThan(linear.rgb, vec3(0.0031308));
+    vec3 higher = pow(linear.rgb, vec3(1.0/2.4)) * vec3(1.055) - vec3(0.055);
+    vec3 lower = linear.rgb * vec3(12.92);
+
+    return vec4(mix(higher, lower, cutoff), linear.a);
+}
+
 void main() {
     vec4 color = texture(image, tex_coord);
+
+    // Our texture is stored with an sRGB internal format,
+    // which means that the values are linearized when we
+    // sample the texture, but for now we actually want to
+    // output the color with gamma compression, so we do
+    // that.
+    color = unlinearize(color);
+
     out_FragColor = vec4(color.rgb * color.a, color.a);
 }