Element A of RP 268-3 Reference Materials for DPX V2.0HDR Implementations
This repository is public.
Please consult CONTRIBUTING.md, CONFIDENTIALITY.md and PATENTS.md for important notices.
Unless specified otherwise, the contents of this element are licensed as indicated at LICENSE.md.
Please report issues at https://github.com/SMPTE/rp268-3-a/issues or at mailto:31fs-chair@smpte.org.
You need to install CMake in order to build the reference code.
Go to the directory corresponding to the example you would like to build (convert_descriptor, generate_color_test_pattern, or dump_dpx). Then:
cmake .
make
Building under Windows requires an installation of Microsoft Visual Studio. Go to the directory corresponding to the example you would like to build (convert_descriptor, generate_color_test_pattern, or dump_dpx). Then:
cmake .
CMake will create a .sln file that can be loaded using the MS Visual Studio IDE.
The convert_descriptor example takes in a DPX file with R, G, B, and optionally A components and rearranges the component and image element structure according to the setting on the command line.
Usage:
convert_descriptor input.dpx output.dpx <neworder> <planar>
- input.dpx - Name of input DPX file
- output.dpx - Name of output DPX file
- neworder - (e.g., BGR, BGRA, ARGB, etc.) order of components in output file. For an interleaved file, only some orders are supported. A is optional.
- planar - set to 0 for interleaved components within a single image element; set to 1 for planes of components, each in its own image element
The dump_dpx example takes in a DPX file, prints the header information, and dumps all of the samples as either text or raw binary files.
Usage:
dump_dpx <dpxfile> (-rawout <out_rawfile_base> (-plane_order <plane_order>) (-bit_depth_conv <bit_depth_conv>) (-dump_full_range))
- dpxfile - Name of DPX file to dump
- out_rawfile_base - Output filename base name (if specified, then the tool outputs to files using this base name rather than as text). The output extension of each file will be .#.ext, where # is the image element number and ext is an extension indicating the component type (y,u,v,r,g,b,a, or g0-7).
- bit_depth_conv - If specified, converts the samples to the specified bit depth before dumping.
- -dump_full_range - If this flag is present, the output raw files will be full range. If absent, output raw files will be limited range. Note that float/double output is always scaled from 0-1.0 based on the high & low code values if a bit depth conversion is specified.
A sample in an output file is stored within the smallest of (1, 2, 4, or 8) bytes that fits the sample (or if bit_depth_conv is specified, that fits within bit_depth_conv bits). For example, if there is an 8-bit DPX image element and bit_depth_conv = 12, then the image element is converted from 8 to 12 bits and each 12-bit sample is placed in a 2-byte (16-bit) field in the raw file output (right-justified). The byte order for raw files always matches the byte order of the machine
One way to use dump_dpx to sanity check images would be to specify the flags: -rawout out -bitdepth_conv 8. The resulting file(s) can be concatenated as needed and interpreted by tools that interpret raw pixel data. For example ffmpeg can be used with the -pix_fmt option set to yuv420p, yuv422p, yuv444p, yuva420p, yuva422p, yuva444p, rgb, or rgba based on the image data that is present and whether it is planar or interleaved.
The generate_color_test_pattern example creates a color bar style pattern for certain combinations of transfer function and bits per sample.
Usage:
generate_color_test_pattern -o out.dpx -tf (BT709|HLG|PQ) -corder (corder) -userfr (1|0) -bpc (8|10|12) -planar (1|0) -chroma (444|422|420) -w (width) -h (height) -dmd (l2r|r2l) -order (msbf|lsbf) -packing (packed|ma|mb) -encoding (1|0)
- out.dpx - Name of output file
- tf - Transfer function (BT709 = ITU-R BT.709, HLG = ITU-R BT.2100 HLG, PQ = ITU_R BT.2100 PQ)
- corder - component order of output file (e.g., CbYCr or BGR)
- usefr - 0 = use limited range; 1 = use full range
- bpc - Specifies bit depth of output file
- planar - 0 = interleaved samples, all components in 1 image element; 1 = planar samples, each component in separate image element
- chroma - 444 = 4:4:4; 422 = 4:2:2; 420 = 4:2:0
- w - output width
- h - output height
- dmd - datum mapping direction; either left to right (l2r) or right to left (r2l)
- order - byte order; either most-significant byte first (msbf) or least-significant byte first (lsbf)
- packing - packing to use; either packed (packed), Method A (ma), or Method B (mb). Note that ma and mb are only valid for 10 and 12 bpc
- encoding - 0 = no compression; 1 = RLE encoding used
The following is a list of supported formats: CbYCr, CbYCrA, CbYCrY422, CbYACrYA422, YCbCr422p, YCbCrA422p, CYY420, CYAYA420, YCbCr420p, YCbCrA420p, BGR, BGRA, ARGB RGB, RGBA, ABGR