1 / 70

Bilinear Accelerated Filter Approximation

Bilinear Accelerated Filter Approximation. Josiah Manson and Scott Schaefer Texas A&M University. Motivation. Trilinear Interpolation. Motivation. Optimized Combination of 2 Bilinear Samples. Mipmap Basis. Mipmap Basis. Mipmap Basis. Mipmap Basis. Mipmap Basis. Filter Approximation.

Download Presentation

Bilinear Accelerated Filter Approximation

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Bilinear Accelerated Filter Approximation Josiah Manson and Scott Schaefer Texas A&M University

  2. Motivation Trilinear Interpolation

  3. Motivation Optimized Combination of 2 Bilinear Samples

  4. Mipmap Basis

  5. Mipmap Basis

  6. Mipmap Basis

  7. Mipmap Basis

  8. Mipmap Basis

  9. Filter Approximation Filter to Approximate

  10. Filter Approximation Bilinear Interpolation

  11. Filter Approximation Best fit using 4 samples

  12. Cache Coherence Cardinality-Constrained Texture Filtering

  13. Cache Coherence Bilinear Accelerated Filter Approximation

  14. Freedom in Bilinear Samples 5 degrees of freedom + mipmap levels

  15. Optimization

  16. Optimization Filter to approximate

  17. Optimization Filter to approximate Bilinear combination of mipmap filters

  18. Optimization Filter to approximate Bilinear combination of mipmap filters Bilinear parameters Optimize

  19. Optimization Filter to approximate Bilinear combination of mipmap filters Bilinear parameters Relative weights of bilinear probes Optimize

  20. Optimization Filter to approximate Bilinear combination of mipmap filters Bilinear parameters Relative weights of bilinear probes Points in image Optimize

  21. Optimization Filter to approximate Bilinear combination of mipmap filters Bilinear parameters Relative weights of bilinear probes Points in image Scale and translation of input filter Optimize

  22. Discretization of Domain

  23. Discretization of Domain 4x4

  24. Discretization of Domain 4x4x2

  25. Coupled vs. Decoupled

  26. Coupled vs. Decoupled

  27. Coupled vs. Decoupled

  28. Lanczos 2 Error w.r.t. Discretization

  29. Lanczos 2 Error w.r.t. Discretization

  30. Lanczos 2 Error w.r.t. Discretization 4x4x2

  31. Error w.r.t. Filter Type at 4x4x2

  32. Symmetry of Domain 32 subdomains

  33. Symmetry of Domain 16 subdomains

  34. Symmetry of Domain 8 subdomains

  35. Symmetry of Domain 6 subdomains

  36. Table Sizes CCTF: Coupled: Decoupled: 6/8 * 8*(4*16 + 4+4+2) 6/8 * (5*4*16 + 2*2) 6/8 * (5*4*16 + 2*(4+4+2)) = 444 bytes = 243 bytes = 255 bytes 6 subdomains, 8 bits per byte Number of parameters and fetches Bits per linear fit of parameter Integer choice of texel/level

  37. Uniform Scaling: Lanczos 2 Trilinear Interpolation

  38. Uniform Scaling: Lanczos 2 Coupled Bilinear

  39. Uniform Scaling: Lanczos 2 Decoupled Bilinear

  40. Uniform Scaling: Lanczos 2 CCTF

  41. 3D Rotation: Lanczos 2 Coupled Bilinear

  42. 3D Rotation: Lanczos 2 Decoupled Bilinear

  43. 3D Rotation: Lanczos 2 CCTF

  44. Speed in FPS

  45. Conclusion • Improved quality over trilinear interpolation • Almost same quality as CCTF • 2x faster than CCTF • Decoupling samples removes constraints • Small GPU lookup table

  46. Lanczos 2 Error w.r.t. Discretization

  47. Uniform Scaling: Lanczos 2 Exact Evaluation

  48. 2D Translation: Lanczos 2 Exact Evaluation

  49. 3D Rotation: Lanczos 2 Trilinear Interpolation

More Related