/[gzz]/manuscripts/AniFont/anifont.tex
ViewVC logotype

Diff of /manuscripts/AniFont/anifont.tex

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 1.22 by tjl, Fri Oct 24 13:20:35 2003 UTC revision 1.23 by tjl, Sat Oct 25 09:43:42 2003 UTC
# Line 85  in an elegant fashion by probing the har Line 85  in an elegant fashion by probing the har
85    
86  \begin{figure*}  \begin{figure*}
87  \begin{tabular}{p{1.5cm}|p{\colwidth}p{\colwidth}p{\colwidth}p{\colwidth}p{\colwidth}}  \begin{tabular}{p{1.5cm}|p{\colwidth}p{\colwidth}p{\colwidth}p{\colwidth}p{\colwidth}}
88  Coordinate mapping&Nearest neighbour&Trilinear&Trilinear+ Anisotropic&FSAA 4xSS&Custom vertex-based 4xSS \\  Coordinate mapping&Nearest neighbour&Trilinear&Trilinear+ Anisotropic&FSAA 4xSS&Vertex-based 4xSS \\
89  \hline\\  \hline\\
90  Iso&  Iso&
91  \snapshot{snapsps/aniso-gffx-tbl-iso-nearest.ps}   &  \snapshot{snapsps/aniso-gffx-tbl-iso-nearest.ps}   &
92  \snapshot{snapsps/aniso-gffx-tbl-iso-trilinear.ps}   &  \snapshot{snapsps/aniso-gffx-tbl-iso-trilinear.ps}   &
93  \snapshot{snapsps/aniso-gffx-tbl-iso-aniso.ps}   &  \snapshot{snapsps/aniso-gffx-tbl-iso-aniso.ps}   &
94  \snapshot{snapsps/aniso-gf4go-tbl-iso-fsaa.ps}   &  \snapshot{snapsps/aniso-gf4go-tbl-iso-fsaa.ps}   &
95  \snapshot{snapsps/aniso-gffx-tbl-iso-super4.ps}   \\[1ex]  \snapshot{snapsps/aniso-gffx-tbl-iso-super4.ps}   \\
96    \\
97  %  %
98  Aniso&  Aniso&
99  \snapshot{snapsps/aniso-gffx-tbl-aniso-nearest.ps}   &  \snapshot{snapsps/aniso-gffx-tbl-aniso-nearest.ps}   &
# Line 169  sampling at too low a resolution for muc Line 170  sampling at too low a resolution for muc
170    
171  \section{Why does stretch and squish improve image quality?}  \section{Why does stretch and squish improve image quality?}
172    
 - quality of trilinear filtering result depends strongly on subpixel position  
173    
174  \begin{figure}  \begin{figure}[t!]
175  \centering  \centering
176  \begin{tabular}{c|c}  \begin{tabular}{c|c}
177  Trilinear  & Stretch-squish \\  Trilinear  & Stretch-squish 2x\\
178  \hline\\  \hline\\
179  \snapshot{snapsps/aniso-gffx-ortho-trilinear-0.ps}&  \snapshot{snapsps/aniso-gffx-ortho-trilinear-0.ps}&
180  \snapshot{snapsps/aniso-gffx-ortho-stretchsquish-0.ps}\\  \snapshot{snapsps/aniso-gffx-ortho-stretchsquish-0.ps}\\
# Line 184  Trilinear  & Stretch-squish \\ Line 184  Trilinear  & Stretch-squish \\
184  \snapshot{snapsps/aniso-gffx-ortho-stretchsquish-2.ps}\\  \snapshot{snapsps/aniso-gffx-ortho-stretchsquish-2.ps}\\
185  \end{tabular}  \end{tabular}
186  \caption{  \caption{
187  \label{figstretchsquishsamples}  \label{figstretchsquishwhyworks}
188  PFSS diagrams of an simple rendering situation,  PFSS diagrams of an isotropic rendering situation,
189  showing how stretch-squish works.  showing why stretch-squish improves image quality.
190  a) Normal trilinear filtering.  The three rows show different sub-pixel translations of the texture.
191  b) Stretching the texture and squishing it allows more samples to be used  The trilinear filter kernel size varies strongly in both X and Y
192  when using an anisotropic filter. The footprint in XXX direction is much closer  directions, whereas the kernel size for
193  to the actual pixel; there is less blur in the output. Here, 2x anisotropy was  stretch-squish varies far less strongly for the direction of the stretch.
 used; using more anisotropy sharpens the filter further.  
194  }  }
195  \end{figure}  \end{figure}
196    
197    The quality of trilinear filtering result depends strongly on subpixel position,
198    as shown in Fig.~\ref{figstretchsquishwhyworks}. The anisotropic filter provides
199    a better
200    
201  - analogous to supersampling  - analogous to supersampling
202    
203    
204  - downside: if transformed nonorthogonally, blurs easier since the "aniso power" is already used  - downside: if transformed nonorthogonally, blurs easier since the "aniso power" is already used
205    
 \section{Example: text}  
   
 \begin{table}  
 \caption{  
 \label{tabperformance}  
 The performance of the different implementations on a XXX.  
 }  
 \end{table}  
206    
207  \def\fontexamplesize{8cm}  \def\fontexamplesize{8cm}
208  \begin{figure*}  \begin{figure*}
# Line 247  Magnified examples of text filtered usin Line 242  Magnified examples of text filtered usin
242    
243  \section{Conclusion}  \section{Conclusion}
244    
245    \begin{table*}
246    \begin{minipage}{\textwidth}
247    \begin{tabular}{p{3cm}|lllll}
248    Method              & HW req & Clarity     & Aliasing    & Code changes   & Relative time per pixel\\
249    \hline\\
250    Trilinear           & Any    & Blurry      & ---         & ---            & 1 \\
251    Trilinear, LOD bias & Any    & Less blurry & Bad         & trivial        & 1---2 \\
252    Stretch-squish 2x   & NV1X+  & Better      & ---         & almost trivial & 1.5---2 \\
253    4x FSAA supersampling & NV1X\footnote{Not available on NVIDIA Linux drivers 44.96 on NV25 or NV31, only on NV1X}  & Good        & ---         & trivial        & 4\footnote{With FSAA, the entire scene slows down, not just the polygon to be improved} \\
254    Vertex-based supersampling &
255                            NV2X+ & Good       & ---         & significant    & 4---6 \\
256    Fragment-based supersampling &
257                            NV3X+ & Good       & ---         & trivial        & 10---20 \\
258    \hline
259    \end{tabular}
260    \end{minipage}
261    \caption{
262    \label{tabperformance}
263    A comparison of different ways of improving the filtering of \emph{isotropically} rendered
264    textured polygons.
265    The hardware requirements are given as NVIDIA architectures; the architectures from
266    other manufacturers such as ATI mostly follow a similar pattern.
267    }
268    \end{table*}
269    
270  - In this article, we argue that isotropic situations should be explicitly avoided  - In this article, we argue that isotropic situations should be explicitly avoided
271    in 2D orthogonal rendering - better quality with aniso    in 2D orthogonal rendering - better quality with aniso
272    
# Line 279  Seveeral aspects we did not touch Line 299  Seveeral aspects we did not touch
299    
300  \section{Acknowledgments}  \section{Acknowledgments}
301    
302    \bibliographystyle{abbrv}
303    \bibliography{gzigzag}
304    
305    \appendix
306    
307  \appendix  \appendix
308  \section*{Appendix}  \section*{Appendix}
309    
310  \section{Probing hardware texture filters for drawing realistic PFSS snapshots}  \section{Probing hardware texture filters in OpenGL}
311  \label{secprobing}  \label{secprobing}
312    
313  In our investigations for this article, we found the pixel footprint  In our investigations for this article, we found the pixel footprint
# Line 362  antialiasing patterns, (XXX should we?). Line 387  antialiasing patterns, (XXX should we?).
387        as the contribution of four texels on a higher mimap level as per        as the contribution of four texels on a higher mimap level as per
388        the assumption of generating the mipmap levels in the usual way        the assumption of generating the mipmap levels in the usual way
389    
 \bibliographystyle{abbrv}  
 \bibliography{gzigzag}  
   
 \appendix  
   
 \section{Probing texture filters}  
390    
391  \begin{enumerate}  \begin{enumerate}
392    

Legend:
Removed from v.1.22  
changed lines
  Added in v.1.23

savannah-hackers-public@gnu.org
ViewVC Help
Powered by ViewVC 1.1.26