Hi,
My question is quite simple, It's to be sure of the good drive.
Each tone mapping effect can be enabled or disabled, but tonemapping as a tonemapping should always be enabled ?
This mean a RGBA16 (or 32) target should always be used to render the scene ?
Thanks to say your point of view
If you disable tone-mapping completely (so you're just displaying a 16-bit image to the screen, or copying it to an 8-bit image), then what you're actually doing is using a linear tone-mapper that remaps 0.0f to black and 1.0f to white.
For most real HDR images, this kind of tone-mapper will not produce good results 
Ok, so better to always have it enabled and don't have an option to disable it.
Filmic Tonemapping still the best ?
struct PS_INPUT
{
float4 Position : SV_POSITION;
float2 TexCoord : TEXCOORD0;
};
Texture2D DiffuseMap : register( t0 );
SamplerState LinearSampler : register( s0 );
float3 HableTonemap( in float3 x )
{
float A = 0.22f; // Shoulder Strength.
float B = 0.30f; // Linear Strength.
float C = 0.10f; // Linear Angle.
float D = 0.20f; // Toe Strength.
float E = 0.01f; // Toe Numerator.
float F = 0.30f; // Toe Denominator.
return ((x*(A*x+C*B)+D*E)/(x*(A*x+B)+D*F)) - E/F;
}
float4 main( in PS_INPUT Input ) : SV_TARGET
{
// Get the diffuse data.
float3 DiffuseData = DiffuseMap.Sample( LinearSampler, Input.TexCoord ).rgb;
// Compute the current color.
float ExposureBias = 2.0f;
float3 CurrentColor = HableTonemap( ExposureBias * DiffuseData );
// Compute the white scale.
float3 WhitePoint = float3( 11.2f, 11.2f, 11.2f );
float3 WhiteScale = 1.0f / HableTonemap( WhitePoint );
// Return the final color.
return float4( CurrentColor * WhiteScale, 1.0f );
}
