//-***************************************************************************** // Copyright (c) 2012, Pixar. All rights reserved. * // * // This license governs use of the accompanying software. If you * // use the software, you accept this license. If you do not accept * // the license, do not use the software. * // * // 1. Definitions * // The terms "reproduce," "reproduction," "derivative works," and * // "distribution" have the same meaning here as under U.S. * // copyright law. A "contribution" is the original software, or * // any additions or changes to the software. * // A "contributor" is any person or entity that distributes its * // contribution under this license. * // "Licensed patents" are a contributor's patent claims that read * // directly on its contribution. * // * // 2. Grant of Rights * // (A) Copyright Grant- Subject to the terms of this license, * // including the license conditions and limitations in section 3, * // each contributor grants you a non-exclusive, worldwide, * // royalty-free copyright license to reproduce its contribution, * // prepare derivative works of its contribution, and distribute * // its contribution or any derivative works that you create. * // (B) Patent Grant- Subject to the terms of this license, * // including the license conditions and limitations in section 3, * // each contributor grants you a non-exclusive, worldwide, * // royalty-free license under its licensed patents to make, have * // made, use, sell, offer for sale, import, and/or otherwise * // dispose of its contribution in the software or derivative works * // of the contribution in the software. * // * // 3. Conditions and Limitations * // (A) No Trademark License- This license does not grant you * // rights to use any contributor's name, logo, or trademarks. * // (B) If you bring a patent claim against any contributor over * // patents that you claim are infringed by the software, your * // patent license from such contributor to the software ends * // automatically. * // (C) If you distribute any portion of the software, you must * // retain all copyright, patent, trademark, and attribution * // notices that are present in the software. * // (D) If you distribute any portion of the software in source * // code form, you may do so only under this license by including a * // complete copy of this license with your distribution. If you * // distribute any portion of the software in compiled or object * // code form, you may only do so under a license that complies * // with this license. * // (E) The software is licensed "as-is." You bear the risk of * // using it. The contributors give no express warranties, * // guarantees or conditions. You may have additional consumer * // rights under your local laws which this license cannot change. * // To the extent permitted under your local laws, the contributors * // exclude the implied warranties of merchantability, fitness for * // a particular purpose and non-infringement. * //-***************************************************************************** //-***************************************************************************** // Written by Pixar Animation Studios, 2011-2012. //-***************************************************************************** #ifndef _PxDeepUtils_h_ #define _PxDeepUtils_h_ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include using std::isfinite; namespace PxDeep { //-***************************************************************************** // The large block of comments below explains our working terminology and // the justification of our limits & magic numbers. In a few places, the // use of centimeters as a spatial unit does affect the absolute position of // various minima and maxima, but in normal usage those should be well outside // working ranges. //-***************************************************************************** //-***************************************************************************** // DENSITY //-***************************************************************************** // "Density" refers to the the optical density which, when integrated // through a line, produces an alpha. The relationship between alpha, density, // and line segment of a given length "dz" is as follows: // // alpha = 1.0 - exp( -dz * density ) // // We use a minimum non-zero density in some places in our code, which // represents the density of dry air at atmospheric pressure. Though // different wavelengths of light are attenuated differently, the average // attenuation is 10^-5 per meter. To make it very minimal, // we'll work with 1/10th that density (tiny tiny). Since our facility // works in centimeters, this works out to (using & rearranging the // equation above) // // 10^-6 = 1.0 - exp( -100.0 * MIN_NON_ZERO_DENSITY ) // exp( -100.0 * MIN_NON_ZERO_DENSITY ) = 1.0 - 10^-6 // -100.0 * MIN_NON_ZERO_DENSITY = log( 1.0 - 10^-6 ) // MIN_NON_ZERO_DENSITY = log( 1.0 - 10^-6 ) / -100.0 // MIN_NON_ZERO_DENSITY = 1.0000050000290891e-08 // // We use double precision for density and dz calculations. // //-***************************************************************************** // VISIBILITY (or 'VIZ', or 'TRANSMISSION') //-***************************************************************************** // Throughout the code below, we transform "alpha" into its inverse, which is // transmissivity, or visibility, or for short, 'viz'. The relationship between // alpha and viz is simple: // // alpha = 1.0 - viz, or viz = 1.0 - alpha. // // Similarly, the relationship between viz and density & dz is simple: // // viz = exp( -dz * density ) // log( viz ) = -dz * density // // Viz is easier to work with than alpha, because to accumulate a total // visibility of many adjacent samples, the relationship is just, for the // set of sample viz's: {viz0, viz1, viz2, ..., vizN-1} // // totalViz = viz0 * viz1 * viz2 * ... * vizN-1 // // It's interesting to note that for any given set of spans, their accumulated // visibility is the same regardless of what order they're accumulated in, // since A*B == B*A. // // When using viz, we use double precision because the operation // 1.0f - ( 1.0f - a ) loses precision, and we want as little of that as // possible! // //-***************************************************************************** // DEPTH RANGES //-***************************************************************************** // Because we need to be able to arithmetically manipulate depths, we place // a range on the valid depth values. Positive Infinity is a valid depth value // to be stored in a DTEX file, but in order to make everything else work, we // set the maximum depth to near (but not at) FLT_MAX, 10^30. Similarly, we // set the minimum depth to just slightly greater than zero, 10^-4. This // could potentially clip effects being deep composited with very small // distances and units of meters. // //-***************************************************************************** // DEEP OPACITY //-***************************************************************************** // "Deep Opacity" refers to a depth function in which the sample at each point // represents the total accumulated opacity at that depth. This represents // the way that deep shadows would have been produced by renderman with the // Display Driver Line: "deepshad" "deepopacity", except that the files actually // store the inverse (1.0-opacity) at each point. It is important to note // that for any given Dtex deepopacity sample, the value represents the // accumulation of visibility on the NEAR side of the sample - up to and // including the sample's depth, but no further in depth. Deep Opacity // functions are monotonically decreasing in depth, and are always // between 0 and 1. // // A complication arises when the 0'th continuous deep opacity sample has a // non-zero deep opacity, because we don't have enough information to infer // where the continuous span that ends at the 0th sample begins in depth. We // solve the problem by interrogating the entire deep pixel for the maximum // density of all its spans (see above), and then solving for what dz // would produce the given accumulated alpha for that max density. The // near point of the initial span is then 'dz' units in front of the 0th // sample depth. // // We sometimes use 'deepViz' to in the code below to refer to 1.0 - deepOpacity // //-***************************************************************************** // DEEP ALPHA //-***************************************************************************** // "Deep Alpha" refers to a depth function in which the sample at each point // represents the non-accumulated alpha of that single sample. When interpreting // the depth function as continuous instead of discrete, Deep Alpha represents // the alpha of the FAR side of the sample - from the depth of the sample // up to, but not including, the depth of the next sample. // // A complication arises when the last continuous deep alpha sample has a // non-zero deep alpha, because we don't have enough information to infer // where the continuous span that begins at the last sample ends in depth. We // solve this problem analagously to how we solve the DeepOpacity problem. // We get the maximum density along the entire deep pixel and extrapolate to // determine an end depth. // //-***************************************************************************** // DEEP RGBA //-***************************************************************************** // Deep RGBA is exactly the same as Deep Alpha, for both discrete and // continuous cases, with the additional R,G, and B channels carried along. // The RGB can be read as premultiplied by alpha, or not. The output deep // pixel expects RGB to be premultiplied by alpha. // The use of premultiplied alpha makes it possible to entangle emitted and // reflected light - basically "glows", when premultiplied R,G,B are non-zero // but alpha is zero. However, in order for us to collapse coindicent samples, // we need to temporarily store RGB unpremultiplied. We simply don't affect // the samples that have zero alpha, and don't remultiply samples that have // zero alpha. There's no way for uncombined samples that had non-zero alpha // to produce a combined sample with zero alpha, so any sample that has // zero alpha at the end of all the combining was entirely composed of zero // alpha samples to begin with. SO, if the alpha is zero, we don't // multiply by it! //-***************************************************************************** //-***************************************************************************** //-***************************************************************************** //-***************************************************************************** // UTILITY CONSTANTS AND FUNCTIONS //-***************************************************************************** //-***************************************************************************** //-***************************************************************************** //-***************************************************************************** // Explained above in the "Density" section of the comments. // We set this value to one tenth the attenuation of light in dry air // at atmospheric pressure. // 10^-6 = 1.0 - exp( -100.0 * MIN_NON_ZERO_DENSITY ) // exp( -100.0 * MIN_NON_ZERO_DENSITY ) = 1.0 - 10^-6 // -100.0 * MIN_NON_ZERO_DENSITY = log( 1.0 - 10^-6 ) // MIN_NON_ZERO_DENSITY = log( 1.0 - 10^-6 ) / -100.0 // MIN_NON_ZERO_DENSITY = 1.0000050000290891e-08 // static const double MIN_NON_ZERO_DENSITY = log( 1.0 - 1.0e-6 ) / -100.0; #define PXDU_MIN_NON_ZERO_DENSITY 1.0000050000290891e-08L //-***************************************************************************** // The change in depth which produces maximum alpha for maximum density. // We want this to be small without risking subnormality. // static const double DZ_OF_ALPHA_1 = 0.001; // static const double DZ_OF_VIZ_0 = 0.001; #define PXDU_DZ_OF_ALPHA_1 0.001 #define PXDU_DZ_OF_VIZ_0 0.001 //-***************************************************************************** // We set the max density of alpha 1 to the density which would produce // an alpha of 0.99999 in a distance of 0.001 centimeters (DZ_OF_ALPHA_1) // // 0.99999 = 1.0 - exp( -0.001 * MAX_DENSITY ) // exp( -0.001 * MAX_DENSITY ) = 1.0 - 0.99999 // -0.001 * MAX_DENSITY = log( 1.0 - 0.99999 ) // MAX_DENSITY = log( 1.0 - 0.99999 ) / -0.001 // MAX_DENSITY = 11512.925464974779 // static const double DENSITY_OF_ALPHA_1 = log( 1.0 - 0.99999 ) / -0.001; // static const double DENSITY_OF_VIZ_0 = DENSITY_OF_ALPHA_1; #define PXDU_DENSITY_OF_ALPHA_1 11512.92546497478 #define PXDU_DENSITY_OF_VIZ_0 11512.92546497478 //-***************************************************************************** // Just in case we need it. These are the constants used above. #define PXDU_MAX_NON_OPAQUE_ALPHA 0.99999 #define PXDU_MIN_NON_OPAQUE_VIZ 0.00001 #define PXDU_MIN_NON_TRANSPARENT_ALPHA 0.00001 #define PXDU_MAX_NON_TRANSPARENT_VIZ 0.99999 //-***************************************************************************** // Explained above in the "Depth" section of the comments. // static const double MIN_DEEP_DEPTH = 1.0e-4; // static const double MAX_DEEP_DEPTH = 1.0e30; #define PXDU_MIN_DEEP_DEPTH 1.0e-4 #define PXDU_MAX_DEEP_DEPTH 1.0e30 //-***************************************************************************** // A maximum depth change (dz) // static const double MAX_DZ = double( MAX_DEEP_DEPTH ) - // double( MIN_DEEP_DEPTH ); #define PXDU_MAX_DZ 1.0e30 //-***************************************************************************** // IEEE 754 floats can be incremented to the "next" positive float // in this manner, for positive float inputs. inline float IncrementPositiveFloat( float i_a, int32_t i_inc=1 ) { typedef union { int32_t i; float f; } intfloat; intfloat a; a.f = i_a; a.i += i_inc; return a.f; } //-***************************************************************************** // IEEE 754 floats can be decremented to the "previous" positive float // in this manner, for positive float inputs. inline float DecrementPositiveFloat( float i_a, int32_t i_inc=1 ) { typedef union { int32_t i; float f; } intfloat; intfloat a; a.f = i_a; a.i -= i_inc; return a.f; } //-***************************************************************************** // From: man isinf // isinf(x) returns 1 if x is positive infinity, and -1 if x is nega- // tive infinity. template inline bool IsInfinity( T i_f ) { return ( isinf( i_f ) == 1 ); } //-***************************************************************************** // A zero-nan functon, which actually zeros inf as well. template inline T ZeroNAN( T i_f ) { if ( !isfinite( i_f ) ) { return ( T )0; } else { return i_f; } } //-***************************************************************************** template inline T ClampDepth( T i_depth ) { if ( IsInfinity( i_depth ) ) { return PXDU_MAX_DEEP_DEPTH; } else { return Imath::clamp( i_depth, ( T )PXDU_MIN_DEEP_DEPTH, ( T )PXDU_MAX_DEEP_DEPTH ); } } //-***************************************************************************** template inline T ClampDz( T i_dz ) { return Imath::clamp( ZeroNAN( i_dz ), ( T )0, ( T )PXDU_MAX_DZ ); } //-***************************************************************************** template inline T ClampNonZeroDz( T i_dz ) { return Imath::clamp( ZeroNAN( i_dz ), ( T )PXDU_DZ_OF_ALPHA_1, ( T )PXDU_MAX_DZ ); } //-***************************************************************************** template inline T ClampAlpha( T i_alpha ) { return Imath::clamp( ZeroNAN( i_alpha ), ( T )0, ( T )1 ); } //-***************************************************************************** // "plausible" in this case means not completely transparent, nor // completely opaque. template inline T ClampPlausibleAlpha( T i_alpha ) { return Imath::clamp( ZeroNAN( i_alpha ), ( T )PXDU_MIN_NON_TRANSPARENT_ALPHA, ( T )PXDU_MAX_NON_OPAQUE_ALPHA ); } //-***************************************************************************** template inline double ClampViz( T i_viz ) { return Imath::clamp( ZeroNAN( i_viz ), ( T )0, ( T )1 ); } //-***************************************************************************** // "plausible" in this case means not completely transparent, nor // completely opaque. template inline T ClampPlausibleViz( T i_viz ) { return Imath::clamp( ZeroNAN( i_viz ), ( T )PXDU_MIN_NON_OPAQUE_VIZ, ( T )PXDU_MAX_NON_TRANSPARENT_VIZ ); } //-***************************************************************************** // Plausible density is clamped between min non-zero density // and density of alpha 1. template inline T ClampPlausibleDensity( T i_density ) { return Imath::clamp( ZeroNAN( i_density ), ( T )PXDU_MIN_NON_ZERO_DENSITY, ( T )PXDU_DENSITY_OF_ALPHA_1 ); } //-***************************************************************************** // Density/Viz/DZ calculations are always performed in double precision. // We try to leave them alone as much as possible, but the logarithm can get // weird for very very small numbers. The "isfinite" call basically rules // out NaN and Infinity results, though it doesn't bother with subnormal // numbers, since the error case we're worried about is log being too big. // viz = exp( -dz * density ) // log( viz ) = -dz * density // density = -log( viz ) / dz double DensityFromVizDz( double i_viz, double i_dz ); //-***************************************************************************** // We can often treat "density times dz" as a single quantity without // separating it. // viz = exp( -densityTimesDz ) // log( viz ) = -densityTimesDz // densityTimesDz = -log( viz ) double DensityTimesDzFromViz( double i_viz ); //-***************************************************************************** // Plausible density defined above. inline double PlausibleDensityFromVizDz( double i_viz, double i_dz ) { return ClampPlausibleDensity( DensityFromVizDz( i_viz, i_dz ) ); } //-***************************************************************************** // viz = exp( -dz * density ) // log( viz ) = -dz * density // dz = -log( viz ) / density // Note that this is basically the same as the computation above. double DzFromVizDensity( double i_viz, double i_density ); //-***************************************************************************** // viz = exp( -dz * density ) // valid for all finite numbers. // negative densities or dz's will give greater than 1 viz's, which will // get clamped! inline double VizFromDensityDz( double i_density, double i_dz ) { return ClampViz( exp( -ZeroNAN( i_density * i_dz ) ) ); } //-***************************************************************************** // same as above. inline double VizFromDensityTimesDz( double i_densityTimesDz ) { return ClampViz( exp( -ZeroNAN( i_densityTimesDz ) ) ); } //-***************************************************************************** //-***************************************************************************** //-***************************************************************************** // IMF SPECIFIC STUFF //-***************************************************************************** //-***************************************************************************** //-***************************************************************************** //-***************************************************************************** template Imf::PixelType ImfPixelType(); template <> inline Imf::PixelType ImfPixelType() { return Imf::HALF; } template <> inline Imf::PixelType ImfPixelType() { return Imf::FLOAT; } template <> inline Imf::PixelType ImfPixelType() { return Imf::UINT; } //-***************************************************************************** // Handy exception macro. #define PXDU_THROW( TEXT ) \ do \ { \ std::stringstream sstr; \ sstr << TEXT; \ std::runtime_error exc( sstr.str() ); \ throw( exc ); \ } \ while( 0 ) } // End namespace PxDeep #endif