• home
• « prev
• next »
• index

Output Shaders

Output shaders are user-written functions to operate on frame buffers. They have access to every pixel in all available frame buffers after rendering, and can be used to perform operations like post-filtering or compositing.

Output Files

Writing frame buffers of a certain data type to files on disk is enabled by specifying a file format and output file name.

In framebuffer statements, adding the file properties to a buffer marks it to be written to the file. For multi-channel formats, the same file name may be given to several buffers to write them as multiple layers to the same file.

Deprecated In file output statements, if the data type is omitted and more than a single target data type is supported by a format, then a default data type is chosen that is assumed to be the "best" type for the given image format. This default type is underlined in the table below.

mental ray supports a variety of standard color file formats, as well as special file formats for depth, vector, and label files. By listing the appropriate number and type of framebuffer statements (or output statements), it is possible to write multiple files. For example, both a filtered file and the unfiltered version can be written to separate files by listing multiple framebuffer statements (or output statements) : one to write the unfiltered image, one that runs an output shader that does the filtering into a secondary buffer, and finally another one to write the filtered image from that buffer. Output statements are executed in sequence.

The following file formats are natively supported for image output.

The supported compression algorithms include :

RLE

a simple lossless run-length encoding, which achieves best compression ratios on images with larger areas of constant color without noise like drawings and graphical paint. This method is implemented slightly differently by various formats, like separate encoding per color component to increase the compression rate in IFF. The compression ratio may reach 2:1 for appropriate input.

JPEG³

a lossy perception-based compression technique with adjustable degree of compression. It is well suited for final storage and distribution of photographic and photorealistic images with smooth variations of color. It is not well suited for further edits like in a compositing pipeline. The compression ratio can easily reach 10:1 for appropriate input.

ZIP

a lossless Huffman dictionary encoding, which tend to give slightly better compression ratios than RLE for photographic imagery at the expense of runtime performance. The compression ratio is about 2:1 for appropriate input.

ZIPS 3.11

like ZIP, but operating on individual scan lines rather than collecting them into blocks. Best suitable for scanline-based tools.

PIZ

a lossless compression based on wavelets coupled with Huffman encoding. This technique improves compression ratios especially for photographic imagery with film grain to about 3:1.

B44 3.11

a lossy compression for channels in "half" data format, designed towards realtime playback performance for such data. Any "integer" or "floating-point" channels are ignored and will be stored uncompressed.

DWAA
DWAB 3.13

a fast, lossy compression for HDR images. The compression ratio can be tuned, and defaults to a value of 45 for a high quality image at a low compression rate. A value of 2000 is suggested as a maximum for an acceptable image quality but highest compression. A value of 200 is a good compromise. There are two variations A and B, which differ in the size of the image chunks they are compressing and storing separately: the A version does smaller 32 scanlines while B does larger tiles of 256 scanlines, making the latter more attractive for full-resolution image transfers.

3.13 The OpenEXR ¹ implementation in mental ray is based on the version 2.2 libraries. This version introduces some new features that are not strictly backwards compatible to the previously used format, but mental ray can be enforced to stick to the older file format output using a registry setting. The following compression modes are supported: RLE, PIZ, ZIP, ZIPS, B44, DWAA, DWAB, and, optionally, PXR24 ², a lossy compression method if the data is stored in full floating-point precision but lossless otherwise. The fastest algorithms are normally RLE and ZIP; PIZ usually achieves better compression but is much slower. The OpenEXR manual recommends PXR24 for depth maps. The DWA modes are well suited for HDR image data. Note, that mental ray defaults to RLE, previous versions were using ZIP compression by default.

Each of the file formats implies a particular default data type (the entry underlined in column "supported data types"), like "jpg" implies the 8-bit RGB color type, and "zt" implies Z depth data. The data type may be overridden by explicitly specifying another data type, such as a 16-bit type, in the framebuffer statement (or output statement), as long as it is supported and appears in the above table.

The null file format is useful to create a stub file during rendering, that can be used to attach the imf_disp viewer, without leaving the file on disk when rendering finishes.

All mental ray native file formats, like ct or st, are following a simple layout. They contain a header followed by the image data, where

header

consists of a magic number byte identifying the format (as listed in table above), a null byte, width and height as unsigned shorts, as well as a version number as unsigned short ; and

image data

are simply uncompressed values written pixel by pixel (with increasing horizontal x coordinate) and line by line (with increasing vertical y coordinate), beginning in the lower left corner (x = y = 0). The pixel data consists of one to four 8-bit, 16-bit, or 32-bit component values, in RGBA, XYZ, or UV order. All shorts, integers, and floats are big-endian (most significant byte first).

The supported data types are:

The half-float data types rgb_h, rgba_h, and a_h inform mental ray to store floating-point color values only with half-float precision in the frame buffer, thus reducing memory requirements at the cost of small details. Note, that only the OpenEXR image file format supports to store half-float data natively. The other file formats can be used as well but may fall back to store only the low dynamic range of the data, which means values outside the [0…1] range will be clamped.

The RGBE color data type can store high dynamic range color values with RGB components which exceed the value of 1.0, the normal upper limit for standard RGB data. There are two specialized file formats HDR and CTH for RGBE, but mental ray will also allow storing RGBE data in any other 8-bit RGBA format, such as TIFF. This will result in images that can not be displayed by standard viewing programs because they would interpret the data as RGBA, but it has become common practice to use such formats to transport RGBE data to other tools that understand RGBE.

All the floating-point data types, like "rgba_fp" or a_fp, permit to store color and alpha values outside the normal range [0…1], often referred to as high dynamic range values. and no dithering is applied even if explicitly enabled. In contrast, any conversion to the 8-bit or 16-bit formats will clamp values outside this interval. Note that dithering tends to reduce the effectivity of RLE compression.

The difference between "vta" and "vts", and between n and m, is significant only when automatic conversions are done. The file contents are identical except for the magic number in the file header.

mental ray can combine samples within a pixel in different ways, called filtering or interpolation. The combination of existing samples can also pad the frame buffers to "bridge" unsampled pixels. Interpolation of colors, depths, normals, and motion vectors means that they are averaged, while interpolation of the labels means that the maximum label is used (taking the average label is not a good idea). Interpolation of depths only takes the average of non-infinite depths, and interpolation of normals and motion vectors only takes the average of vectors different from the null vector.

If interpolation is turned off for a frame buffer, the last sample value (color, normal, motion vector, or label) within each pixel is stored, and pixels without samples get a copy from one of the neighboring pixels. Interpolation off for depth images is an exception: rather than using the last sample depth, the min depth is used - this can be useful for compositing. Interpolation is on by default for color frame buffers (including alpha and intensity frame buffers) and off by default for depth, normal, motion vector, and label frame buffers.

Interpolation is controlled with the framebuffer filtering property.

Deprecated In backwards compatible output statements, it is turned on by writing a "+" in the beginning of the output type and turned off by writing a "-" there. For example, to interpolate the depth samples, write "+z" in the output statement:

• home
• « prev
• next »
• index

Copyright © 1986, 2015 NVIDIA ARC GmbH. All rights reserved.