Tuning the Affine Feature

This section describes how to tune the Affine feature to control the affine submodules of LTM within a DC9400.

Before you begin

Make sure:
  • The details view shows the settings of the Affine feature for LTM.

    To tune this feature, you need to choose DC9400 > LTM > Affine in the Features view. For more details, see Tuning a DC9400 Feature.

    Figure 1. Details View When Affine under LTM is Selected

  • A VTunerServer is accessible.

    For information about how to set up a connection to a VTunerServer, see Setting Up a Connection to a VTunerServer.

About this task

This feature is used to tune the following affine submodules of LTM:
  • Luma adjustment (GuidanceMap), which is an optional global tone mapping submodule based on a 1D LUT before LTM
  • Affine grid filtering, based on the affine grid parameters specified in Step 3 and previous frame affine grids
  • Affine grid slicing, which uses a high-resolution luma value and the parameters specified in Step 4 to interpolate the affine transformation coefficients a and b for the current pixel where the interpolation is trilinear interpolation from the eight closest control points in the affine grid
  • Local affine transform, which performs pixel affine transformation ( y = a x + b ) by using the affine coefficients output from affine grid interpolation and uses interpolation to calculate the minimum luma output based on the minimum output scale LUT

Procedure

  1. Select the Enable check box to enable the affine submodules of LTM.
  2. To tune the luma adjustment (GuidanceMap) submodule, use one of the following methods to specify the guide map curve with 128 points in the Guide Map Curve table:
    • Double-click and edit a cell in the table to update a single point.
    • Click Edit and update the points in the prompted Edit Table dialog.

      The points must be specified in the same order as the table columns and separated by a comma (,).

    • Click Export to export the points to a .csv file, edit the file, and then import the updated points from the .csv file.

      The order of the points in the exported .csv file is the same as that in the table.

  3. To tune the affine grid filtering submodule, perform the following steps:
    1. In the Smooth Weight spin box, enter the temporal smooth weight for affine grid.

      The value must be in the range from 0 to 1024, inclusive.

    2. In the Affine Slope table, specify the affine slope with 2048 (=16×16×8) points by referring to Step 2.
    3. In the Affine Bias table, specify the affine bias with 2048 (=16×16×8) points by referring to Step 2.
  4. To tune the affine grid slicing submodule, perform the following steps:
    1. In the Affine Grid Slice Base table, specify the slice base by referring to Step 2.
    2. In the Affine Grid Width spin box, enter the width of the affine grid.

      The value must be in the range from 1 to 16, inclusive.

    3. In the Affine Grid Height spin box, enter the height of the affine grid.

      The value must be in the range from 1 to 16, inclusive.

    4. In the Affine Grid Depth spin box, enter the depth of the affine grid depth.

      The value must be in the range from 4 to 8, inclusive.

  5. To tune the local affine transform submodule, perform the following steps:
    1. In the Slope Bit spin box, enter the number of fractional bits.
    2. In the Bias Bit spin box, enter the number of bits that the bias is left shifted.
    3. In the Affine Scale Coef table, specify the scaling coefficients used to calculate the minimum output from the input luma by referring to Step 2.

Results

  • The tuning results are printed in the Console view in real time.
  • The image displayed on the VTunerServer side is updated.