Presentation is loading. Please wait.

Presentation is loading. Please wait.

Image Editing Work Pixel-based Methods  Adobe Photoshop, GIMP, SuperGoo.

Similar presentations


Presentation on theme: "Image Editing Work Pixel-based Methods  Adobe Photoshop, GIMP, SuperGoo."— Presentation transcript:

1

2 Image Editing Work Pixel-based Methods  Adobe Photoshop, GIMP, SuperGoo

3 Image Editing Work Pixel-based Methods  Adobe Photoshop, GIMP, SuperGoo Pixel-based Methods  Adobe Photoshop, GIMP, SuperGoo Image-based Methods  Wolberg ’90-’00 Image Warping

4 Image Editing Work Pixel-based Methods  Adobe Photoshop, GIMP, SuperGoo Pixel-based Methods  Adobe Photoshop, GIMP, SuperGoo Image-based Methods  Wolberg ’90-’00 Image Warping Pixel-based Methods  Adobe Photoshop, GIMP, SuperGoo Image-based Methods  Wolberg ’90-’00 Image Warping  Beier & Neely ’92 Field Morphing

5 Image Editing Work Pixel-based Methods  Adobe Photoshop, GIMP, SuperGoo Image-based Methods  Wolberg ’90-’00 Image Warping  Beier & Neely ’92 Field Morphing  Lee, et. al ’94 Thin-plate Splines

6 Image Editing Work Pixel-based Methods  Adobe Photoshop, GIMP, SuperGoo Image-based Methods  Wolberg ’90-’00 Image Warping  Beier & Neely ’92 Field Morphing  Lee, et. al ’94 Thin-plate Splines Region-of-Interest Methods  Image-based methods could be applied to regions-of-interest

7 Image Editing Work Pixel-based Methods  Adobe Photoshop, GIMP, SuperGoo Image-based Methods  Wolberg ’90-’00 Image Warping  Beier & Neely ’92 Field Morphing  Lee, et. al ’94 Thin-plate Splines Region-of-Interest Methods  Image-based methods could be applied to regions-of-interest Pixel-based Methods  Adobe Photoshop, GIMP, SuperGoo Image-based Methods  Wolberg ’90-’00 Image Warping  Beier & Neely ’92 Field Morphing  Lee, et. al ’94 Thin-plate Splines Region-of-Interest Methods  Image-based methods could be applied to regions-of-interest  Elder & Goldberg ’98, ’01

8 Image Editing Work Pixel-based Methods  Adobe Photoshop, GIMP, SuperGoo Image-based Methods  Wolberg ’90-’00 Image Warping  Beier & Neely ’92 Field Morphing  Lee, et. al ’94 Thin-plate Splines Region-of-Interest Methods  Image-based methods could be applied to regions-of-interest Pixel-based Methods  Adobe Photoshop, GIMP, SuperGoo Image-based Methods  Wolberg ’90-’00 Image Warping  Beier & Neely ’92 Field Morphing  Lee, et. al ’94 Thin-plate Splines Region-of-Interest Methods  Image-based methods could be applied to regions-of-interest  Elder & Goldberg ’98, ’01

9 Texture Synthesis Work Efros & Leung ’99 Non-parametric Sampling Wei & Levoy 2000 Faster with TSVQ Harrison 2000 Resynthesizer plug-in Praun, et al. 2000 Lapped Textures Liang, et al. ‘01 Real-Time – Patch-Based Sampling Efros & Freeman ‘01 Image Quilting

10 Where does Object-Based Image Editing fit in? Image-based Methods Pixel-based Methods

11 Where does Object-Based Image Editing fit in? Object-Based Image Editing Image Warping, Morphing, Spline warping SuperGoo, clone tool, nudge tool, etc.

12 Object-Based Image Editing

13 What we’re going to cover: 1.Object selection & representation 2.Object editing operations 3.Rendering 4.Background filling and texture painting 5.Applications

14 Object Selection 1.Segment image into catchment basins (TRAPs) (Tobogganed Regions of Accumulated Plateaus) using a Toboggan-based watershed algorithm. 2.Manually “tag” TRAPs to select object Selected Object TRAPs

15 Object Representation Selected 5 TRAPs Object Fit Triangular Mesh to selected Object - for efficient OpenGL Rendering

16 Object Representation Selected 5 TRAPs Object Fit Triangular Mesh to selected Object - for efficient OpenGL Rendering

17 Object Representation 1. Recursive Divide & Conquer Polygonalization of Object Boundary

18 Object Representation 2. Insert Nodes (corners and basepoints)

19 Object Representation 3. Final Triangulation – using Delaunay algorithm

20 Object Representation

21 Object Editing Operations 1.Object selection & representation 2.Object editing operations 3.Rendering 4.Background filling and texture painting 5.Applications

22 Object Editing Operations 2.Object editing operations a.Direct Editing b.Indirect Editing

23 Object Editing Operations 2.Object editing operations a.Direct Editing b.Indirect Editing

24 Stretch 1. After the object has been selected Object

25 Stretch 1. After the object has been selected 2. And vertices (x,y) identified Object (x,y)

26 Stretch 1. After the object has been selected 2. And vertices (x,y) identified (x,y) Object 3. An anchor/pivot point is specified - x +x

27 Stretch 1. After the object has been selected 2. And vertices (x,y) identified 3. An anchor/pivot point is specified 4. And the user clicks on or near the object to stretch it with respect to the anchor point.

28 Stretch Object vertices (x,y), are stretched x Reference Lines

29 Stretch x Reference Lines show uniform stretch Object vertices (x,y), are stretched to (x ’,y ’ ), where (  x,  y) captures cursor movement b x, b y are object dimensions x'

30 We now introduce an attenuation function, A[i] Simple Linear Stretch i A[i]A[i] x x' With the same result … for constant A[i]

31 But if we vary A[i] Nonlinear Stretch i A[i]A[i] x'x' We get nonlinear stretch

32 But if we vary A[i] Less MovementMore Movement x'x' Nonlinear Stretch i A[i]A[i] Less Stretch x'x'

33 But if we vary A[i] Less MovementMore Movement Nonlinear Stretch i A[i]A[i] x'x' More Stretch

34 And if we vary A[i] even more Less MovementMore Movement x'x' Extreme Nonlinear Stretch Nonlinear Stretch i A[i]A[i] x'x'

35 Simple rotation matrix, with attenuated  i A[i]A[i] Rotate

36 Rotate Rotational Bend Simple rotation matrix, with attenuated  i A[i]A[i]

37 Bend-Stretching Calculate the bend based on stretch values User Actions Computation

38 Bend-Stretching Figure 10: Stretching and Bending Arm. (a) Arm bent and stretched simultaneously using Al[i] and Ar[i] (see Figure 9) with boundaries (green), axis of rotation (red). (b) Arm bent more to show occlusion.

39 Indirect Editing 2.Object editing operations a.Direct Editing b.Indirect Editing

40 Curve Deformers - Defaults Length Thickness Rotation Cursor Point Anchor Point

41 Length Flattening Length ~ flat Thickness Rotation Cursor Point Anchor Point

42 Bend-Stretching Length Leveling Thickness Rotation Level-off Cursor Point Anchor Point

43 Rendering 1.Object selection & representation 2.Object editing operations 3.Rendering 4.Background filling and texture painting 5.Applications

44 Antialiasing using OpenGL Variables: Layer Height Layer Width Layer Height w h

45 Antialiasing using OpenGL Variables: Layer Height Layer Width Layer Height Number of Layers Layer Width Object Boundary

46 Background Filling 1.Object selection & representation 2.Object editing operations 3.Rendering 4.Background filling and texture painting 5.Applications

47 Background Filling 2 Methods 1.Scale Down Filling (2a-3a) 2.Gridded (2b-3b) Sampling 1.Automatic 2.User Selects

48 Texture Preservation When an object stretches, if its texture is high in detail, the texture becomes overly smoothed, looking unnatural. To fix this, we keep object TRAP sizes constant, warping only their basepoint positions.

49 Applications 1.Object selection & representation 2.Object editing operations 3.Rendering 4.Background filling and texture painting 5.Applications

50 Video

51 Contributions & Conclusions 1. New object selection tool 2. Trap-based triangulation 3. Automatic background filling 4. Attenuation functions A[i] 5. Curve deformers 6. OpenGL antialiasing 7. Texture painting 8. Texture preservation 9. Image editing at the object level

52 Future Work 1. Structured background filling 2. Intelligent selection tools (Intelligent Scissors, Paint) 3. Predefined/Advanced/Compounded Attenuation functions 4. Subpixel edge model for decontamination of object fringe 5. Use multiple anchor points, Edit multiple objects simultaneously


Download ppt "Image Editing Work Pixel-based Methods  Adobe Photoshop, GIMP, SuperGoo."

Similar presentations


Ads by Google