Poisson Image Editing & Terrain Synthesis

1. Poisson Image Editing & Terrain Synthesis Howard Zhou Jie Sun howardz@cc.gatech.edusun@cc.gatech.edu 2003 . 4.29

2. Table of Contents • Introduction / motivation • Poisson Image Editing • Terrain Synthesis (Texture based methods) • Future work • Conclusion

3. Table of Contents • Introduction / motivation • Poisson Image Editing • Terrain Synthesis (Texture based methods) • Future work • Conclusion

4. Introduction / motivation • Poisson Image Editing • Seamless • Texture based terrain synthesis • Current method based on fractals • Very limited control • Terrain style adjusted by parameter tuning • What if the user draws a rough sketch and supply a height map and says: “I want this to be like this”

5. Poisson Image Editing • Review

6. Our implmentation • Matlab • Sparse matrix PDF solver • Use conjugate gradient solver supplied by Matlab • Can be faster if …

7. Seamless insertion

8. Inserting objects with holes

9. Inserting transparent objects

10. Texture flattening Result directly related to Edge detection result

11. Local illumination changes alpha = 0.05 beta = 0.2 alpha = 0.05 beta = 0.4

12. Seamless tiling • Good when seam is not significant • Often needs to increase the contrast of the result • but don’t an automatic way, maybe use histogram of the original image

13. Seamless tiling Good when the seam is not significant

14. Seamless tiling Show some more

15. Seamless tiling

16. Seamless tiling

17. Seamless tiling Contrast can be globally fixed But how?

18. Seamless tiling Seams not good Cannot be fixed

19. Table of Contents • Introduction / motivation • Poisson Image Editing • Terrain Synthesis (Texture based methods) • Future work • Conclusion

20. Previous approachTexture based terrain synthesis • Current method based on fractals • Very limited control • Terrain style adjusted by parameter tuning • What if the user draws a rough sketch and supply a height map and says: “I want this to be like this”

21. Texture based terrain synthesis • Image analogy • Texture synthesis on laplacian + piecewise seamless tiling • Graph cut / seamless tiling • Separating the details

22. Data: height map

23. Display height map

24. Image analogy A. Hertzmann, C. Jacobs, N. Oliver, B. Curless, D. Salesin.SIGGRAPH 2001

25. Texture by number

26. Texture by number

27. How do we get (A) automatically • Blurring (filtering) • Texture flattening using edge detection result or contour

28. Image analogy + (texture flattening + blurring)

29. Laplacian Synthesis • Regard laplacian as a particular texture • Texture synthesis • Integrate

30. Results

31. Terrain

32. Terrain

33. Problems & possible solutions • Depend on the boundary conditions • Use the boundary attached to the Laplacian • There is only one unique solution of this linear system • Lost the power of Poisson editing • Should use a non-conservative gradient field

34. Graph cut + seamless tiling

35. Laplacian removing boundary(since the boundary is known)

36. Image smoothing edge (1 D)

37. Using Poisson Solver

38. Terrain Analysis • The detail of the terrain differs at different altitude • Terrain = f ( altitude ) • Altitude = g ( style )

39. Example: Terrain map

40. Low Frequency - Altitude

41. High Frequency – as a function of Altitude

42. Proposed Algorithm • Use “Copy & Paste” methods to generate an altitude map • Add high frequency probabilistically as indexed by the altitude map • Graph cuts/Image Quilting to make it seamless

43. Table of Contents • Introduction (motivation) • Re-illumination • Changing viewpoint • Future work • Conclusion

44. Future Work • Other texture methods (Graph cut, stocastic?) • Stylized map generation from real map • Real map from stylized map

45. Map vs. terrain

46. Conclusion • Implemented poisson image editing • Tried several texture based terrain synthesis methods • Lots to be done!

47. Questions ?