texbeml - ps

Apply a fake bump environment-map transform with luminance correction. This is accomplished by modifying the texture address data of the destination register, using address perturbation data (du,dv), a 2D bump environment matrix, and luminance.


texbeml dst, src



  • dst is the destination register.
  • src is a source register.


Pixel shader versions 1_1 1_2 1_3 1_4 2_0 2_x 2_sw 3_0 3_sw
texbeml x x x


The red and green color data in the src register is interpreted as the perturbation data (du,dv). The blue color data in the src register is interpreted as the luminance data.

This instruction transforms the red and green components in the source register using the 2D bump environment mapping matrix. The result is added to the texture coordinate set corresponding to the destination register number. A luminance correction is applied using the luminance value and the bias texture stage values. The result is used to sample the current texture stage.

This can be used for a variety of techniques based on address perturbation such as fake per-pixel environment mapping.

This operation always interprets du and dv as signed quantities. For versions 1_0 and 1_1, the Source Register Signed Scaling input modifier (_bx2) is not permitted on the input argument.

This instruction produces defined results when input textures contain mixed format data. For more information about surface formats, see D3DFORMAT.

// When using this instruction, texture registers must follow 
//   the following sequence:
// The texture assigned to stage tn contains the (du,dv) data
// The texture assigned to stage t(m) is sampled
tex     t(n)                    
texbeml t(m),  t(n)      where m > n

This example shows the calculations done within the instruction.

// 1. New values for texture addresses (u',v') are calculated
// 2. Sample the texture using (u',v')
// 3. Luminance correction is applied

u' = TextureCoordinates(stage m)u +

D3DTSS_BUMPENVMAT00(stage m)*t(n)R +

D3DTSS_BUMPENVMAT10(stage m)*t(n)G

v' = TextureCoordinates(stage m)v +

D3DTSS_BUMPENVMAT01(stage m)*t(n)R +

D3DTSS_BUMPENVMAT11(stage m)*t(n)G

t(m)RGBA = TextureSample(stage m) using (u',v') as coordinates

t(m)RGBA = t(m)RGBA *

[(t(n)B * D3DTSS_BUMPENVLSCALE(stage m)) +


Register data that has been read by a texbem or texbeml instruction cannot be read later, except by another texbem or texbeml.

// This example demonstrates the validation error caused by 
//   t0 being reread
tex t0
texbem t1, t0
add r0, t1, t0

(Instruction Error) (Statement 4) Register data that has been read by 
texbem or texbeml instruction cannot be read by other instructions


Here is an example shader with the texture maps identified and the texture stages identified.

tex t0              ; Define t0 to get a 2-tuple DuDv
texbeml t1, t0      ; Compute (u',v')
                    ; Apply luminance correction                    
                    ; Sample t1 using (u',v')
mov r0, t1          ; Output result

This example requires the following textures in the following texture stages.

  • Stage 0 is assigned a bump map with (du, dv) perturbation data.
  • Stage 1 is assigned a texture map with color data.
  • texbeml sets the matrix data on the texture stage that is sampled. This is different from the functionality of the fixed function pipeline where the perturbation data and the matrices occupy the same texture stage.

Pixel Shader Instructions