Linear, Gamma and sRGB Color Spaces

July 01, 2018 Dark/Light

This post is based on a reply of mine to an ImGui issue (and discussion) about sRGB and linear color spaces.

Gamma color space

The actual color of a pixel, outputted on a monitor, does not linearly depend on the applied voltage signal for that pixel. For CRT monitors, the actual color is approximately proportional to the applied voltage raised to the power of a so-called gamma value, which depends on the monitor. This gamma value typically lies between 2.2 and 2.5 for CRT monitors. (Hence, the ubiquitous appearances of the magical constant 2.2 in rendering applications.)

\[L_\mathrm{actual} \sim V^\gamma\]

To bypass this gamma value in your computations, you need to gamma correct the computed colors of each pixel before presenting. By raising the computed color to a power of the reciprocal of that same gamma value, the computed color becomes proportional to the actual color.

\[V \sim L_\mathrm{computed}^{1/\gamma} \\ L_\mathrm{actual} \sim L_\mathrm{computed}\]

So in general gamma correction is a technique that adapts the computed colors to the transfer function of the monitor used for outputting. Both CRT and non-CRT monitors each have their own transfer function. This means that the final rendering pass should adapt the computed colors, based on the used monitor, to obtain correct actual colors.

These linear-to-gamma color space conversions seem like unnecessary overhead from a monitor construction point of view. It is physically perfectly possible to construct a CRT monitor with a gamma value of exactly one, ensuring that the computed color is already proportional to the actual color, and eliminating the need for gamma correction.

From a perceptual point of view, removing the need for gamma correction and using a monitor where computed colors are proportional to the corresponding actual colors is actually a bad idea. Typically, actual colors are represented with 8 (or 10) bits for each of the red, green and blue channel. This quantization only supports 256 (or 1024) different colors. Here, a 0 value represents completely black and a 255 (or a 1023) value represents completely white. But what about the intermediate values? If a linear encoding is used (i.e. a gamma value of 1), the majority of values would be perceptually very close to white and a very small minority would be perceptually close to black. By using a gamma encoding (e.g., a gamma value of 2.2), the distribution is perceptually more linear (i.e. equidistant intervals between black and white).

sRGB color space

The sRGB color space has similar goals, assigning a perceptual linear range of (computed) color values to the available 256 different (actual) color values. A rough approximation of transforming linear to sRGB color values consists of raising to a power of 2.2. A more accurate approximation would distinguish between a more linear correspondence near black and a gamma (2.4) encoding near white. And finally, you can just use the exact transformation between linear and sRGB color space as well. The more accurate, the more expensive the computation will be.

The sRGB color space is obviously used for sRGB monitors which are primarily used for content creation (e.g., textures, etc.). Images with a R8G8B8A8 format are in most cases represented in sRGB color space and definitely not in linear color space (as a user, you obviously do not need to guess, but just need to know which color space is used for the encoding). RGB color pickers typically operate in sRGB color space as well.

Color operations

Colors represented in linear color space can be mutually added and multiplied. Colors represented in a gamma color space cannot be mutually added, but can be mutually multiplied. Colors represented in sRGB color space can neither be mutually added nor multiplied. Note that I explicitly differentiate between a gamma and sRGB color space, since not everyone will use the most cheapest linear-to-sRGB approximation.

Colors represented in a gamma or sRGB color space cannot be used as vertex attributes during the barycentric interpolation for fragment generation. The color attribute of a vertex needs to be expressed in linear space before passing to the rasterizer stage. The pixel/fragment shader will thus always obtain a fragment color expressed in linear space.

The above applies to vertex attributes; so single color coefficients. But what about textures? For textures there is not a choice. Most textures will contain colors expressed in sRGB color space, and thus these textures should be explicitly encoded as sRGB colors (e.g. DXGI_FORMAT_..._SRGB). By using an explicit sRGB texture format, the hardware will perform the sRGB-to-linear conversion when appropriate. So with regard to texture sampling/filtering:

hardware sRGB-to-linear conversion -> filtering != filtering -> user defined sRGB-to-linear conversion

Similarly for blending (i.e. all blending except opaque blending). By using an explicit sRGB texture format, the hardware will perform the linear-to-sRGB conversion when appropriate.

blending -> hardware linear-to-sRGB conversion != user defined linear-to-sRGB conversion -> blending

Alternatively, a render target with a half-precision floating-point format can and should be used, if linear color values need to be stored instead. This, however, is not the responsibility of ImGui. ImGui should output linear colors, and the user of ImGui should provide an appropriate render target (i.e. R8G8B8A8_SRGB or R16G16B16A16). Most games let the user use a custom gamma encoding to adjust brightness instead of using the default sRGB encoding. So in these uses cases, ImGui will output to a render target with a half-precision floating-point format in a first pass. The final pass will apply the custom gamma encoding and output without blending to a R8G8B8A8 render target.

Use cases

ImGui

The changes required to ImGui:

Application to the D3D11 demo (similar for other graphics APIs):

main.cpp

// Perceptually a better choice than the current DXGI_FORMAT_R8G8B8A8_UNORM:
sd.BufferDesc.Format = DXGI_FORMAT_R8G8B8A8_UNORM_SRGB;

imgui_impl_dx11.cpp

static const char* vertexShader =
     "cbuffer vertexBuffer : register(b0) \
     {\
     float4x4 ProjectionMatrix; \
     };\
     struct VS_INPUT\
     {\
     float2 pos : POSITION;\
     float4 col : COLOR0;\
     float2 uv  : TEXCOORD0;\
     };\
     \
     struct PS_INPUT\
     {\
     float4 pos : SV_POSITION;\
     float4 col : COLOR0;\
     float2 uv  : TEXCOORD0;\
     };\
     \
     PS_INPUT main(VS_INPUT input)\
     {\
     PS_INPUT output;\
     output.pos = mul( ProjectionMatrix, float4(input.pos.xy, 0.f, 1.f));\
     output.col = SRGBtoLinear(input.col);\
     output.uv  = input.uv;\
     return output;\
     }";

MAGE

MAGE v0 uses the following conventions: