Incorrect parenthesization in fp error bounds computation #181
Comments
mmp
added a commit
that referenced
this issue
Oct 12, 2019
MartinNose
pushed a commit
to MartinNose/pbrt-v3-rpf
that referenced
this issue
Nov 23, 2022
0oTinaKimo0
added a commit
to MartinNose/pbrt-v3-rpf
that referenced
this issue
Nov 28, 2022
commit fc8a19b Author: MartinNose <[email protected]> Date: Sun Nov 27 20:03:25 2022 -0800 update algo 3 commit e16019b Merge: c258b81 df2008c Author: MartinNose <[email protected]> Date: Wed Nov 23 00:14:06 2022 -0800 Merge branch 'jdt' into ljl merge from jdt commit c258b81 Merge: cae1c9d ff84c15 Author: MartinNose <[email protected]> Date: Tue Nov 22 23:15:19 2022 -0800 merge from jdt commit cae1c9d Merge: 9a26da4 8d011c8 Author: MartinNose <[email protected]> Date: Tue Nov 22 23:13:56 2022 -0800 excluded sample.dat commit 9a26da4 Merge: ed9a8ef a42418f Author: MartinNose <[email protected]> Date: Thu Nov 17 09:52:46 2022 -0800 merge from jdt commit a42418f Author: 0oTinaKimo0 <[email protected]> Date: Tue Nov 15 12:53:57 2022 -0800 Update rpf.ipynb commit ed9a8ef Author: MartinNose <[email protected]> Date: Mon Nov 14 00:46:09 2022 -0800 Feature Vector done commit 7eab29d Author: 0oTinaKimo0 <[email protected]> Date: Mon Nov 14 00:14:30 2022 -0800 Notebook to process samples commit 7f2d8c1 Merge: 449bb9d 332e952 Author: 0oTinaKimo0 <[email protected]> Date: Sun Nov 13 23:47:01 2022 -0800 Merge branch 'ljl' into jdt commit 332e952 Author: MartinNose <[email protected]> Date: Sun Nov 13 23:30:01 2022 -0800 modified description file to have a lower resolution commit c943489 Author: MartinNose <[email protected]> Date: Sun Nov 13 23:28:59 2022 -0800 featureVector result commit 7308bc8 Author: MartinNose <[email protected]> Date: Sun Nov 13 23:27:47 2022 -0800 add python process tool commit f3e334e Author: MartinNose <[email protected]> Date: Sun Nov 13 23:27:16 2022 -0800 fix featureVector bug commit 3f57ba1 Author: MartinNose <[email protected]> Date: Sun Nov 13 18:16:28 2022 -0800 add getters in all material class commit 449bb9d Merge: fbba9b3 7661bea Author: 0oTinaKimo0 <[email protected]> Date: Sun Nov 13 15:19:54 2022 -0800 Merge branch 'ljl' into jdt commit 7661bea Author: MartinNose <[email protected]> Date: Sun Nov 13 15:14:42 2022 -0800 modified pbrt files from 128 spp to 8 spp commit 4e43976 Author: MartinNose <[email protected]> Date: Sun Nov 13 15:14:03 2022 -0800 modified CMakeLists.txt to compile commit d319e5a Author: MartinNose <[email protected]> Date: Sun Nov 13 15:12:38 2022 -0800 add getters during rendering process\n todo add texture getter during all the material class commit fbba9b3 Author: 0oTinaKimo0 <[email protected]> Date: Wed Nov 9 14:20:03 2022 -0800 header file commit 4d0f974 Author: 0oTinaKimo0 <[email protected]> Date: Wed Nov 9 00:01:53 2022 -0800 Update RPF.cpp commit 49203cc Author: MartinNose <[email protected]> Date: Tue Nov 8 23:10:27 2022 -0800 add FeatureVector implementation commit c66e483 Author: 0oTinaKimo0 <[email protected]> Date: Tue Nov 1 20:30:10 2022 -0700 Create RPF.cpp commit 3ca6c13 Author: MartinNose <[email protected]> Date: Thu Oct 27 17:26:52 2022 -0700 add FeatureVector declaration commit 6d79443 Author: Artem Kharytoniuk <[email protected]> Date: Sat Apr 11 20:46:33 2020 +0200 Fix trilinear filtering A texture footprint with 'width' characteristic size corresponds to texture sampling rate with 1 sample per screen pixel. In order to satisfy the Nyquist limit we need a filter of size '2*width' - and that's what we have in the original code. This computation does not take into account though, that after computing lod and selecting mip levels, we apply bilinear filtering to sample each mip. This effectively increases filter size to '4*width' which results in too blurry images. There could be at least two solutions here: the first one is to use '2*width' filter size and then use point sampling when working with separate mip levels. Another solution proposed here is to select the mip levels based on the 'width' filter size and then rely on bilinear filtering to get correct result. I didn't check this but the second method could provide better results because the last step when we apply bilinear filtering takes into account specific use case and in the case of point sampling the final result is baked into the mip level, so we don't have an opportunity to select proper position between texels as we do with bilinear filtering. The proposed solution also matched HW filtering results, for example, in this project https://github.com/kennyalive/vulkan-raytracing, the RTX raytracing code computes texture lod using just 'width' and then applies bilinear filter. The result closely matches rasterization version. Color encoding of lod levels used in that demo allows to visualize differences in lod selection. By modifying the code to use '2*width' it can be shown that is produces wrong result. commit 593b859 Author: Matt Pharr <[email protected]> Date: Fri Jun 26 14:43:41 2020 -0700 Fix bug in MicrofacetTransmission Return 0 for f() and pdf if both directions are on the same side of the microsurface. Issue mmp#228, via Mandy Xia. commit 3cb0349 Author: Matt Pharr <[email protected]> Date: Fri Jun 26 13:36:10 2020 -0700 Fix Henyey Greenstein importance sampling sign conventions (via Jiayin Cao) commit 577d76d Author: Jakub Hendrich <[email protected]> Date: Wed Nov 13 20:10:51 2019 +0100 Fix memory leak in FourierBSDFTable commit eff5b43 Author: jim price <[email protected]> Date: Sat Mar 14 18:20:00 2020 -0700 Initialize alpha with lower bound to avoid NaN Alpha is as a denominator in the TrowbridgeReitzDistribution and BeckmannDistribution so we initialize to a lower bound of 0.001. This value was choosen as its less than the RoughnessToAlpha()'s lower bound and matches the bound in the Disney material. commit 4f0a794 Author: Matt Pharr <[email protected]> Date: Fri Jun 26 10:06:19 2020 -0700 Fix rare NaN case in CosDPhi(). Issue mmp#272, via Pascal Grittmann. commit e5c7d6b Author: Girish Ramesh <[email protected]> Date: Sat Oct 26 00:09:57 2019 +0100 Correct importance sampling as per d'Eon paper commit b90949c Author: Matt Pharr <[email protected]> Date: Wed Dec 4 10:07:10 2019 -0800 Fix memory leak introduced in 8445c00. Issue mmp#264. commit 0170681 Author: Matt Pharr <[email protected]> Date: Fri Nov 22 17:30:32 2019 -0800 SPPM: move MemoryArena vector out of "nIterations" loop This saves a bunch of unnecessary MemoryArena creation and deletion, which in turn was hurting parallel scalability. commit bc58a2a Author: Matt Pharr <[email protected]> Date: Sat Oct 26 17:07:44 2019 -0700 Fix build for recent versions of xcode commit 3c12841 Author: Matt Pharr <[email protected]> Date: Sun Oct 20 15:09:51 2019 -0700 Fix pbrtObjectEnd to not mess up attribute/object end nesting when --cat/--toply is used commit 1b8bbe8 Author: Matt Pharr <[email protected]> Date: Tue Oct 15 14:03:26 2019 -0700 Fix orientation of wh in MicrofacetReflection::f Always orient wh to be on the same side of the surface as the normal; in turn, this means that the cos theta value passed to Fresnel::Evaluate() has the right sign. This in particular fixes the handling of total internal reflection with this BxDF, both accounting for it when it should be and not incorrectly including it when there is none. (Rendered images may change due to this, especially noticeable in cases where TIR was incorrectly happening with non-transmissive surfaces, giving bright highlights.) Fixes issues mmp#214 and mmp#254. commit b7923b8 Author: Matt Pharr <[email protected]> Date: Tue Oct 15 11:33:30 2019 -0700 Apply ReverseOrientation to per-vertex normals Issue mmp#229 detritus. commit fd2ec9b Author: Matt Pharr <[email protected]> Date: Mon Oct 14 13:23:15 2019 -0700 Fix handling of per-vertex normals with transforms that swap handedness With triangle meshes, even when the shading normal's direction was authoratative (coming via per-vertex normal vectors), we'd still flip the shading normal if the transform swapped the coordinate system handedness. This in turn led to bugs where the wrong side of the light would be emissive when a ray intersected it. Issue mmp#229 commit c49a318 Author: Matt Pharr <[email protected]> Date: Mon Oct 14 08:12:33 2019 -0700 Fix DisneyMaterial microfacet model Issue mmp#224. commit 84393b0 Author: Matt Pharr <[email protected]> Date: Sun Oct 13 14:36:46 2019 -0700 Remove pError scale in OffsetRayOrigin() when PBRT_FLOAT_AS_DOUBLE is enabled. Before, pError was scaled by 1024 in that case; the idea was that there is so much excess precision with 64-bit floats that it wouldn't matter to bump up the error that much. The problem with that is the Curve shape: because it has the behavior of facing the incident ray, the ray epsilon has to be fairly large--half the curve's width. Multiplying that by 1024 in turn totally broke spawned rays, which effectively wouldn't intersect anything... Fixes issue mmp#221. commit db3fe7d Author: Matt Pharr <[email protected]> Date: Sun Oct 13 14:35:40 2019 -0700 Fix Curve Log2 computation when Float=Double Before, it would end up calling the 64-bit FloatToBits, then letting the result be truncated to 32-bits, and then interpreting some bits of that as exponent. For a double, it ended up using random bits from the significand, rather than the exponent. Discovered while chasing down mmp#221, though not the cause of it. commit 2aeb128 Author: Matt Pharr <[email protected]> Date: Sat Oct 12 16:05:27 2019 -0700 Fix parenthesization in Transform::operator() implementations. Issue mmp#181 commit b5a3e14 Author: Matt Pharr <[email protected]> Date: Sat Oct 12 14:28:55 2019 -0700 Improve EFloat::GetAbsoluteError(). Issue mmp#167 commit a5a2eba Author: Matt Pharr <[email protected]> Date: Sat Oct 12 14:11:31 2019 -0700 Actually use quadratic probing in the TransformCache. Issue mmp#194. commit 8f7c841 Author: Matt Pharr <[email protected]> Date: Sat Oct 12 11:14:22 2019 -0700 Fix variable naming in AnimatedTransform constructor. Issue mmp#231 commit 699bcb1 Author: Matt Pharr <[email protected]> Date: Sat Oct 12 11:13:57 2019 -0700 Request c++11 features for obj2pbrt commit 0bb1e8c Author: Stefan Reinhold <[email protected]> Date: Thu Jul 4 21:21:56 2019 +0200 Fixing ResampleLinearSpectrum method in spectrum.cpp commit 97946c6 Author: Mudream <[email protected]> Date: Sat Jan 19 18:05:52 2019 +0800 update cmake version commit 6c25dba Author: Matt Pharr <[email protected]> Date: Sat Oct 12 10:11:30 2019 -0700 Fix out of bounds index in MarbleTexture. Issue mmp#252 commit d82d9bb Author: Matt Pharr <[email protected]> Date: Wed Sep 4 11:48:35 2019 -0700 Fix bug in Film::AddSpat(). (Due to the fact that the Camera importance methods return >0 importance within the sample bounds but then Film::AddSplat() just uses a box filter.) With BDPT, the pixels along the left and top edges would receive ~2x the contributions they were supposed to if a pixel filter with > 1x1 extent was used. The issue was that p would come in with negative values in one of their coordinates between -1 and 0. The conversion from Point2f to Point2i would turn those to zero and then their contribution would be included, even though it wasn't supposed to be. commit 6ed8c71 Author: Matt Pharr <[email protected]> Date: Thu May 23 10:36:00 2019 -0700 Fix bug in BSDF::rho(). We need to transform the given wo direction to local coordinates before passing it along to the BxDF. commit ab34e43 Author: Matt Pharr <[email protected]> Date: Fri Jan 4 16:23:44 2019 -0800 Fix ray differential computation in SpecularTransmit() See pbrt-v2 PR mmp/pbrt-v2#38 and this discussion https://groups.google.com/forum/#!topic/pbrt/THex27EOrOE for context. Notes on the derivation of the differentials are included for reference and for future use in PBR book text. commit 67ed7af Author: SOH <[email protected]> Date: Sat Dec 29 14:57:32 2018 +0800 更新计算射线和Curve相交时候的参数 commit 6369892 Author: Matt Pharr <[email protected]> Date: Sat Dec 29 16:31:10 2018 -0800 Try two at disk parameterization and partial derivatives. The fix from issue mmp#154 led to a different bug (issue mmp#183). This switches back to the old partial derivative vectors via adjusting the disk parameterization. (Now (0,0) is on the outer ring, which isn't great, but seems like the least worst option.) commit bf00edb Author: Matt Pharr <[email protected]> Date: Sat Dec 29 16:26:58 2018 -0800 Revert "Fix disk dPdv computation." This reverts commit 2b55c0c. commit f42d4f1 Author: Matt Pharr <[email protected]> Date: Sat Dec 15 15:04:46 2018 -0800 Fix typo in preprocessor definition. Via Hiroyuki Sakai, issue mmp#199. commit 50e0710 Author: Yoshiyuki Yamariku <[email protected]> Date: Wed Dec 12 21:13:40 2018 +0900 Update sphere.cpp Fix compile error on gcc when PBRT_FLOAT_AS_DOUBLE enabled. commit 989666b Author: Matt Pharr <[email protected]> Date: Sat Nov 24 11:37:25 2018 -0800 Clean up code formatting to match pbrt-v3 and minimize diffs vs. the book commit f0199ed Author: Wenzel Jakob <[email protected]> Date: Sun Nov 18 11:43:03 2018 +0100 Improve numerical behavior of Sphere::Sample() The current implementation of ``Sphere::Sample()`` computes ``sinTheta`` from ``cosTheta`` using the expression ``std::sqrt(1-sinTheta*sinTheta)``. While convenient and computationally cheap, this approach suffers from severe cancellation errors for small angles, causing the sphere to effectively collapse down to a point. Numerical errors start to kick in below ``sinThetaMax`` < 1.5 degrees in single precision. As far as numerical issues go, this isn't that terrible — a point light is a relatively good approximation to a distant sphere. Yet, 1.5 degrees seems relatively large and we can easily do better by computing ``sinTheta`` directly using a Taylor series expansion. This commit introduces such an approximation and also reorganizes the computation to be a bit more efficient. commit eeaed2b Author: Peihong <[email protected]> Date: Mon Jul 16 00:09:20 2018 -0700 Bug fix for TransformCache `hashTable.resize(512)` should be done after `hashTable.clear()`, otherwise the size of the hash table will be 0 after `Clear` is called. When pbrt is compiled as a library, this would lead to crash in multiple rendering passes. commit 0cca3d3 Author: linhsu0723 <[email protected]> Date: Fri Nov 9 05:56:58 2018 +0800 ScopedAssignment: move assignment operator commit da42851 Author: Andrew Tinits <[email protected]> Date: Thu Nov 8 17:26:36 2018 -0500 Remove unused variable wi commit 90f7d92 Author: Sébastien Rombauts <[email protected]> Date: Thu Nov 8 01:36:35 2018 +0100 Add Visual Studio folders using source_group() and add headers to them (mmp#206) Without that, the Visual Studio solution does not show headers, and does group all source files from all sub-directories into on big mess. commit 33800d3 Author: Matt Pharr <[email protected]> Date: Fri Nov 2 11:19:18 2018 -0700 Fix overflow bug in I0(). ifact would overflow a 32-bit integer. via Matthias Raab. commit d47c52f Author: Matthias Moulin <[email protected]> Date: Thu Nov 1 21:22:07 2018 +0100 Removed unused baseColor texture (mmp#205) The single channel `baseColor` texture is already "replaced" by the n-channel `color` texture. Note that `base color` is the actual terminology used by Burley. commit 8ef190d Author: Matt Pharr <[email protected]> Date: Wed Oct 17 14:46:33 2018 -0700 Update README.md for online edition of the book commit 700e799 Author: Matt Pharr <[email protected]> Date: Mon Oct 8 16:28:57 2018 -0700 Fix incorrect error message. (via Ingo Wald.) commit d46fda4 Author: Matt Pharr <[email protected]> Date: Mon Oct 1 17:26:34 2018 -0700 Update to latest OpenEXR. Fixes mmp#196. commit fe3e55d Author: Matt Pharr <[email protected]> Date: Mon Oct 1 09:25:02 2018 -0700 Generate RGB EXRs; the alpha channel wasn't being used anyway. commit c0986cf Author: Matt Pharr <[email protected]> Date: Wed Aug 15 07:17:56 2018 -0700 Remove CopyIlmBaseLibs from CMakeLists.txt Part of mmp#196 fix. commit c74dae1 Author: Matt Pharr <[email protected]> Date: Mon Jul 30 07:23:42 2018 -0700 Don't expand Include directives with --cat and --toply commit cd7d7a9 Author: Matt Pharr <[email protected]> Date: Sun Jul 29 20:48:59 2018 -0700 Update Dockerfile commit 7d624ad Author: Matt Pharr <[email protected]> Date: Wed Jul 25 15:38:25 2018 -0700 Also check for _WIN32 in little endian check commit ce1d69e Author: Matt Pharr <[email protected]> Date: Wed Jul 25 15:06:50 2018 -0700 Disable more chatty MSVC warnings commit c3f337d Author: Matt Pharr <[email protected]> Date: Wed Jul 25 15:02:22 2018 -0700 Fix LNK4049 warnings from glog in MSVC build commit 56d8d19 Author: Matt Pharr <[email protected]> Date: Wed Jul 25 14:52:59 2018 -0700 Use folders in cmake to clean up generated MSVC solution file commit 6346592 Author: Matt Pharr <[email protected]> Date: Wed Jul 25 14:38:32 2018 -0700 Fix MSVC warning about tr1 namespace in gtest.h commit b975881 Author: Matt Pharr <[email protected]> Date: Wed Jul 25 13:42:21 2018 -0700 Improve README.md Add more detailed instructions about MSVC builds. commit 71b03dc Author: Matt Pharr <[email protected]> Date: Wed Jul 25 13:41:55 2018 -0700 Specify float/double and RGB/spectral rendering via cmake config variables commit 69bc471 Author: Charles de Rousiers <[email protected]> Date: Mon Jul 16 00:01:02 2018 +0200 Fix double radians conversion, reducing the effect of the alpha parameter on hair appearance (mmp#189) commit b35c2fa Author: Matt Pharr <[email protected]> Date: Fri Jul 13 16:30:42 2018 -0700 Fix Windows file memory mapping to work with large files (>2GB). Fix thanks to Steve Booth. Also, remove a gratuitous and inefficient string copy of the mapped file. commit 095f6fd Author: Matt Pharr <[email protected]> Date: Mon Jun 18 16:26:11 2018 -0700 Fix (for real this time) buggy ProjectionLight cosTotalWidth initialization commit e077418 Author: Matt Pharr <[email protected]> Date: Mon Jun 18 16:24:12 2018 -0700 Fix race condition in StatRegisterer constructor commit 6435419 Author: Matt Pharr <[email protected]> Date: Mon May 14 16:53:39 2018 -0700 Allow empty quoted strings in the parser commit 0761c1f Author: Matt Pharr <[email protected]> Date: Fri May 11 19:41:57 2018 -0700 Parser: exit out more quickly on errors Fixes some crashes on things like unterminated strings. commit 0fa4f53 Author: Matt Pharr <[email protected]> Date: Fri May 11 19:41:04 2018 -0700 SPPM: don't trace rays with zero contributions commit 54d5e3d Author: soma <[email protected]> Date: Tue May 8 05:32:18 2018 +0900 Fix arguments handling (mmp#187) commit 64873f4 Author: Matt Pharr <[email protected]> Date: Thu Apr 26 16:56:35 2018 -0700 Fix error handling in decodeEscaped() Fixes mmp#186. commit c289647 Author: Matt Pharr <[email protected]> Date: Mon Apr 9 09:45:56 2018 -0700 Update CHECK_NOTNULL calls commit 02a6162 Author: Matt Pharr <[email protected]> Date: Sat Apr 7 14:43:21 2018 -0700 ptex build cleanup in CMakeLists.txt Disable building shared libs (we don’t need them, and they’re problematic when we’re building zlib as well, since we only build a static zlib) commit 56bd7cc Author: Matt Pharr <[email protected]> Date: Sat Apr 7 14:29:04 2018 -0700 Update to latest ptex commit dd42551 Author: Matt Pharr <[email protected]> Date: Thu Mar 29 17:16:46 2018 -0700 Always evaluate all BxDF components in BSDF::Sample_f As noted by 卡卡西 on the pbrt mailing list, the current implementation can have light leaks if the sampled direction is under the surface. commit 73c06fe Author: Matt Pharr <[email protected]> Date: Thu Mar 29 12:57:12 2018 -0700 Remove comment and update DisneyClearcoat f() to be more clear. commit 174d4c7 Author: Matt Pharr <[email protected]> Date: Thu Mar 29 12:27:52 2018 -0700 Clean up handling of zlib in CMakeLists.txt Possibly helps issue mmp#174 commit e4f5681 Author: Matt Pharr <[email protected]> Date: Thu Mar 29 12:26:25 2018 -0700 Fix DisneyBSSRDF::S(). Sw() was adding an extra Fresnel term and an unneeded normalization scale. Issue mmp#173 commit d644d2a Author: Matt Pharr <[email protected]> Date: Thu Mar 29 08:30:58 2018 -0700 Remove excess cos theta terms from DisneyClearCoat::f() These are actually already present, via SmithG_GGX(). Empirically, adding these also made the BxDF non-energy conserving. (Issue mmp#172). commit 40fe422 Author: liftA42 <[email protected]> Date: Fri Mar 30 02:36:45 2018 +0800 Fix unit test for print util (mmp#166) commit c24e8e5 Author: Matt Pharr <[email protected]> Date: Tue Mar 6 16:41:53 2018 -0800 Add distance scale factor to DisneyBSSRDF commit b02c169 Author: Matt Pharr <[email protected]> Date: Mon Mar 5 11:40:39 2018 -0800 Run clang-format on parser.cpp commit 7d35bc0 Author: Immocat <[email protected]> Date: Mon Mar 5 08:23:21 2018 -0800 fix_DisneyClearcoat_bug v1 (mmp#172) commit 7aad635 Author: Matt Pharr <[email protected]> Date: Mon Mar 5 08:00:09 2018 -0800 DisneyMaterial updates Address a few issues identified by Evgenii Golubev; thanks to Brent Burley for additional clarification. commit 99feb88 Author: Matt Pharr <[email protected]> Date: Fri Mar 2 11:55:41 2018 -0800 Work around MSVC2013 limitation commit 53b8cb6 Author: Matt Pharr <[email protected]> Date: Fri Mar 2 10:23:56 2018 -0800 Allow specifying image crop window via the command-line commit 96b9410 Author: Matt Pharr <[email protected]> Date: Tue Feb 27 15:35:46 2018 -0800 Record VolPathIntegrator path length with an integer distribution Issue mmp#176 via @abusch. commit c951c54 Author: Alexandre Baron <[email protected]> Date: Mon Feb 19 00:28:03 2018 +0100 Make the parser use file memory-mapping on Windows too. (mmp#175) * Make the parser use file memory-mapping on Windows too. * Omit \r characters in comments and scope the test tokenizer to make unit tests happy commit dddc27a Author: Matt Pharr <[email protected]> Date: Sun Feb 18 14:43:54 2018 -0800 Triangle: compute dndu and dndv even if the parameterization is degenerate commit a03ba64 Author: Matt Pharr <[email protected]> Date: Sat Feb 17 15:38:52 2018 -0800 Fix rare attempt to Normalize() degenerate vector in Triangle intersection With frame380 of the measure one scene, a ray would occasionally be reported hit a triangle with zero area that we'd then try to compute a normal for. Added a test for that specific case. commit 1f739ff Author: Matt Pharr <[email protected]> Date: Fri Feb 16 13:44:13 2018 -0800 #include cleanups commit 9a49bad Author: Matt Pharr <[email protected]> Date: Fri Feb 16 11:05:53 2018 -0800 Move ParseFile to official api: pbrtParseFile Also add pbrtParseString() commit eba742e Author: Matt Pharr <[email protected]> Date: Fri Feb 16 11:05:04 2018 -0800 Forward declare TransformedPrimitive in pbrt.h commit fa9fa2d Author: Matt Pharr <[email protected]> Date: Wed Feb 14 19:19:10 2018 -0800 Remove unused parser detritus commit 5d1fb69 Author: Matt Pharr <[email protected]> Date: Fri Jan 26 13:48:49 2018 -0800 Fix printing of boolean parameters in --cat/--fmt commit 222e7a3 Author: Matt Pharr <[email protected]> Date: Mon Jan 15 16:24:10 2018 -0800 homebrew modifications to hash functions are a bad idea commit 18e9fd4 Author: Matt Pharr <[email protected]> Date: Mon Jan 15 14:14:03 2018 -0800 Curve: look for splitdepth as an integer parameter (Still accept it as a float for backwards compatibility) Issue mmp#171. commit 736c939 Author: Matt Pharr <[email protected]> Date: Fri Jan 12 16:11:21 2018 -0800 Remove unneeded unistd.h check from CMakeLists commit 553545a Author: Matt Pharr <[email protected]> Date: Fri Jan 12 16:06:22 2018 -0800 Improve error checking for missing submodules commit fd89479 Author: Matt Pharr <[email protected]> Date: Fri Jan 12 15:56:57 2018 -0800 Remove mention of flex and bison from README.md commit d68e7c4 Author: Matt Pharr <[email protected]> Date: Fri Jan 12 15:56:46 2018 -0800 Fix possible invalid memory reference in ParseFile() commit 1dba79b Author: Matt Pharr <[email protected]> Date: Wed Jan 10 06:05:02 2018 -0800 Fix bug in handling of ungetToken in parser The parser would crash if an unget was done after the last token in a file was read (and thence, the file had been removed from fileStack) commit d7e6d71 Author: Matt Pharr <[email protected]> Date: Sun Jan 7 16:00:01 2018 -0800 Fix bug in ptex gamma correction commit 04426b9 Author: Matt Pharr <[email protected]> Date: Sun Jan 7 15:41:50 2018 -0800 Ptex improvements Optionally apply gamma correction to returned values (2.2 default) Set max texture cache size to 4GB commit a73a643 Author: Matt Pharr <[email protected]> Date: Sun Jan 7 15:18:30 2018 -0800 Improve performance of AttributeBegin/AttributeEnd Use a copy-on-write approach for textures and named materials to avoid making unnecessary copies of the underlying std::maps. commit 5b837b2 Author: Matt Pharr <[email protected]> Date: Sun Jan 7 15:17:38 2018 -0800 Bump printed copyright to 2018 commit fa1d4fd Author: Matt Pharr <[email protected]> Date: Sun Jan 7 15:16:55 2018 -0800 Rewrite scene file lexer and parser by hand (mmp#169) * Rewrite scene file lexer and parser by hand This substantially reduces time parsing very complex scene descriptions. Removed the old flex/bison-based parser.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
(Via Kostya Smolenskiy)
pages 228 and 229. In earlier section, You had mentioned that since the compiler can't rearrange floating point computation, it executes it like it's written in code. Code for transforming points is written like, for example:
But in derivation of error bounds on that computation parentheses are as if it is:
From which it stems that maximum gamma is gamma(3), but with former code, maximum gamma is gamma(4).
The text was updated successfully, but these errors were encountered: