Tone mapping

1. Tone mapping Digital Image Synthesis Yung-Yu Chuang 11/08/2005 with slides by Fredo Durand, and Alexei Efros

2. Tone mapping • How can we display it? Linear scaling?, thresholding? 10-6 106 dynamic range Real world radiance 10-6 106 Display intensity Pixel value 0 to 255 CRT has 300:1 dynamic range

3. Global operator (Reinhart et al)

4. Global operator results

5. What does the eye sees? The eye has a huge dynamic range Do we see a true radiance map?

6. Eye is not a photometer! • "Every light is a shade, compared to the higher lights, till you come to the sun; and every shade is a light, compared to the deeper shades, till you come to the night." — John Ruskin, 1879

7. Compressing dynamic range range range

8. Fast Bilateral Filteringfor the Display ofHigh-Dynamic-Range Images Frédo Durand & Julie Dorsey Laboratory for Computer Science Massachusetts Institute of Technology

9. Recover response curve HDR valuefor each pixel High-dynamic-range (HDR) images • CG Images • Multiple exposure photo [Debevec & Malik 1997] • HDR sensors

10. A typical photo • Sun is overexposed • Foreground is underexposed

11. Gamma compression • X -> Xg • Colors are washed-out Input Gamma

12. Gamma compression on intensity • Colors are OK, but details (intensity high-frequency) are blurred Intensity Gamma on intensity Color

13. Chiu et al. 1993 • Reduce contrast of low-frequencies • Keep high frequencies Low-freq. Reduce low frequency High-freq. Color

14. The halo nightmare • For strong edges • Because they contain high frequency Low-freq. Reduce low frequency High-freq. Color

15. Our approach • Do not blur across edges • Non-linear filtering Large-scale Output Detail Color

16. Edge-preserving filtering • Blur, but not across edges • Anisotropic diffusion [Perona & Malik 90] • Blurring as heat flow • LCIS [Tumblin & Turk] • Bilateral filtering [Tomasi & Manduci, 98] Input Gaussian blur Edge-preserving

17. Comparison with our approach • We use only 2 scales • Can be seen as illumination and reflectance • Different edge-preserving filter from LCIS Large-scale Detail Output Compressed

18. Start with Gaussian filtering • Here, input is a step function + noise output input

19. Start with Gaussian filtering • Spatial Gaussian f output input

20. Start with Gaussian filtering • Output is blurred output input

21. Gaussian filter as weighted average • Weight of x depends on distance to x output input

22. The problem of edges • Here, “pollutes” our estimate J(x) • It is too different output input

23. Principle of Bilateral filtering • [Tomasi and Manduchi 1998] • Penalty g on the intensity difference output input

24. Bilateral filtering • [Tomasi and Manduchi 1998] • Spatial Gaussian f output input

25. Bilateral filtering • [Tomasi and Manduchi 1998] • Spatial Gaussian f • Gaussian g on the intensity difference output input

26. Normalization factor • [Tomasi and Manduchi 1998] • k(x)= output input

27. Bilateral filtering is non-linear • [Tomasi and Manduchi 1998] • The weights are different for each output pixel output input

28. Contrast reduction Input HDR image Contrast too high!

29. Contrast reduction Input HDR image Intensity Color

30. Contrast reduction Input HDR image Large scale Intensity FastBilateral Filter Color

31. Contrastreduction Input HDR image Large scale Intensity FastBilateral Filter Detail Color

32. Contrast reduction Input HDR image Scale in log domain Large scale Large scale Intensity Reducecontrast FastBilateral Filter Detail Color

33. Contrast reduction Input HDR image Large scale Largescale Intensity Reducecontrast FastBilateral Filter Detail Detail Preserve! Color

34. Contrast reduction Input HDR image Output Large scale Largescale Intensity Reducecontrast FastBilateral Filter Detail Detail Preserve! Color Color

35. Informal comparison Bilateral[Durand et al.] Photographic[Reinhard et al.] Gradient domain[Fattal et al.]

36. Informal comparison Bilateral[Durand et al.] Photographic[Reinhard et al.] Gradient domain[Fattal et al.]