Billboard Clouds: A New Approach for Extreme Model Simplification in Realtime Rendering - , Study notes of Computer Science

Billboard clouds, a new approach for extreme simplification in realtime rendering. The authors present an optimization approach to build a billboard cloud for a textured polygonal model and a given geometric error threshold. The technique is demonstrated to work on a large class of models, including smooth manifolds and composite objects.

Typology: Study notes

Pre 2010

Uploaded on 08/31/2009

koofers-user-el2
koofers-user-el2 🇺🇸

10 documents

1 / 2

Toggle sidebar

This page cannot be seen from the preview

Don't miss anything!

bg1
MIT Laboratory for Computer Science, March 2003
417
Billboard Clouds for Extreme Model Simplification
Xavier D´
ecoret, Fr´
edo Durand, Franc¸ois X. Sillion, Julie Dorsey
Introduction: We introduce billboard clouds a new approach for extreme simplification in the context of real-
time rendering. 3D models are simplified onto a set of planes with texture and transparency maps. We present an
optimization approach to build a billboard cloud for a textured polygonal model and a given geometric error threshold.
We demonstrate our technique on a large class of models, including smooth manifolds and composite objects.
(a) (b) (c) (d)
Figure 1: Example of a billboard cloud: (a) Original model (5,138 polygons) (b) false-color rendering using one color per billboard
to show the faces that were grouped on it (c) View of the (automatically generated) 32 textured billboards (d) the billboards side by
side.
Approach:We iteratively replace large sets of faces by a textured plane. The texture is generated by projecting the
associated set of faces on the plane. The faces and the plane are chosen so that the distance between vertices and their
projection is bounded (error threshold), and that the projected area of faces is maximized. To approximate the whole
model with a minimum number of planes, we uses a greedy approach that iteratively chooses planes that approximate
great quantities of faces, which we call dense planes. Density is evaluated in plane space, using a discretization and
adaptative refinement for efficiency. A good surface approximation is ensured by favoring planes that are “nearly
tangent” to the model. This method does not require connectivity information, but still avoids cracks by projecting
primitives onto multiple planes when needed. To avoid large textures with a lot of empty space, which can occur when
distant parts of the model are simplified on the same plane, our algorithm takes into account the compacity of the
generated textures when selecting set of faces to be simplified. For extreme simplification, our approach combines the
strengths of mesh decimation and image-based impostors. Billboard clouds is a new representation that can be used
not only for rendering but also quick shadow display or fast ray-object intersection as shown on Figure 2.
(a) (b) (c) (d)
Figure 2: Billboard clouds encode an object shape and appearance through a few quads and transparent textures.
This representation can be used for fast rendering of complex objects (a), for shadow computations (b) or ray-object
intersection (c). Storing additional information such as normal map in the textures also allow real-time relighting for
improved realism (d).
Progress:The simplification of complex object with billboard clouds gives good results. The shape and details can be
captured with a surprisingly low number of billboards. But this low number of billboards can not be straightforwardly
compared to the number of faces in a mesh simplification. The rendering of a billboard cloud is penalized mainly
by the fill rate : when 100 interleaved billboards are rendered, lots of pixels are drawn many times (the worst case
being parallel billboards viewed from front). Therefore, a billboard clouds is useful when its projected screen size is
pf2

Partial preview of the text

Download Billboard Clouds: A New Approach for Extreme Model Simplification in Realtime Rendering - and more Study notes Computer Science in PDF only on Docsity!

MIT Laboratory for Computer Science, March 2003 417

Billboard Clouds for Extreme Model Simplification

Xavier D´ecoret, Fr´edo Durand, Franc¸ois X. Sillion, Julie Dorsey

Introduction: We introduce billboard clouds – a new approach for extreme simplification in the context of real- time rendering. 3D models are simplified onto a set of planes with texture and transparency maps. We present an optimization approach to build a billboard cloud for a textured polygonal model and a given geometric error threshold. We demonstrate our technique on a large class of models, including smooth manifolds and composite objects.

(a) (b) (c) (d)

Figure 1: Example of a billboard cloud: (a) Original model (5,138 polygons) (b) false-color rendering using one color per billboard to show the faces that were grouped on it (c) View of the (automatically generated) 32 textured billboards (d) the billboards side by side.

Approach: We iteratively replace large sets of faces by a textured plane. The texture is generated by projecting the associated set of faces on the plane. The faces and the plane are chosen so that the distance between vertices and their projection is bounded (error threshold), and that the projected area of faces is maximized. To approximate the whole model with a minimum number of planes, we uses a greedy approach that iteratively chooses planes that approximate great quantities of faces, which we call dense planes. Density is evaluated in plane space, using a discretization and adaptative refinement for efficiency. A good surface approximation is ensured by favoring planes that are “nearly tangent” to the model. This method does not require connectivity information, but still avoids cracks by projecting primitives onto multiple planes when needed. To avoid large textures with a lot of empty space, which can occur when distant parts of the model are simplified on the same plane, our algorithm takes into account the compacity of the generated textures when selecting set of faces to be simplified. For extreme simplification, our approach combines the strengths of mesh decimation and image-based impostors. Billboard clouds is a new representation that can be used not only for rendering but also quick shadow display or fast ray-object intersection as shown on Figure 2.

(a) (b) (c) (d)

Figure 2: Billboard clouds encode an object shape and appearance through a few quads and transparent textures. This representation can be used for fast rendering of complex objects (a), for shadow computations (b) or ray-object intersection (c). Storing additional information such as normal map in the textures also allow real-time relighting for improved realism (d).

Progress: The simplification of complex object with billboard clouds gives good results. The shape and details can be captured with a surprisingly low number of billboards. But this low number of billboards can not be straightforwardly compared to the number of faces in a mesh simplification. The rendering of a billboard cloud is penalized mainly by the fill rate : when 100 interleaved billboards are rendered, lots of pixels are drawn many times (the worst case being parallel billboards viewed from front). Therefore, a billboard clouds is useful when its projected screen size is

418 MIT Laboratory for Computer Science, March 2003

not too large. When an object is far away, using a billboard cloud simplification gives higher quality results than a geometrically simplified version of the object[1] of same rendering cost.

Future: For billboard clouds to be interesting when viewed from closer range, we have to diminish the number of billboards. Unfortunately, if we want the billboard cloud to be viewable from any angle, this number of billboards cannot be lowered too much. The extrem case of a single billboard would obviously look “flat” from many viewpoints. Therefore we plan to investigate object based view-dependent billboards clouds. The space around an object is divided into regions as indicated on Figure 3. A billboard cloud is constructed for each of these regions or viewcells. By limit- ing the view-angle within a region, we expect to get a low billboard count for each cloud. Moreover, the distance range delimited by each viewcell also allows to optimize the resolution of billboards’ textures. Therefore, the extra storage caused by multiple view-dependent clouds versus a single view-independent one should not be too high compared to the benefits. Preliminary results seems promising but there is still a lot to investigate, and notably the transitions.

(object based viewcells) (close range) (middle range) (far range)

Figure 3: Examples of view-dependent billboards. A billboard cloud is build for each viewcell. Depending of the distance to the object, the algorithm detects that more and more billboards are required to represent the shape and appearance of a tree model ( 000 polygons). Note also that the further the billboard cloud is, the lower the resolution of the texture is.

Research Support: This research is supported by funding from the NTT Corporation.

References:

[1] M. Garland and P. Heckbert, “Surface simplification using quadric error metrics,” Proceedings of SIGGRAPH 97 ,

[2] X. D´ecoret, Preprocessing of large databases for interactive visualisation , Ph.D. thesis, Universit´e Joseph Four- rier, Grenoble, october 2002.

[3] X. D´ecoret, F. Sillion, F. Durand, and J. Dorsey, “Billboard clouds,” Tech. Rep. RR-4485, INRIA, Grenoble, June