From 74d3bcec05e693889b075ef64d88ee3ef985d13c Mon Sep 17 00:00:00 2001 From: WalterPlinge <22519813+WalterPlinge@users.noreply.github.com> Date: Sun, 1 May 2022 21:29:09 +0100 Subject: [PATCH] Fixed the depth values in the doc file, made some info more clear --- core/image/netpbm/doc.odin | 19 ++++++++++--------- core/image/netpbm/netpbm.odin | 2 +- 2 files changed, 11 insertions(+), 10 deletions(-) diff --git a/core/image/netpbm/doc.odin b/core/image/netpbm/doc.odin index baeb99968..1b5b46856 100644 --- a/core/image/netpbm/doc.odin +++ b/core/image/netpbm/doc.odin @@ -6,24 +6,25 @@ Formats: PAM (P7 ): Portable Arbitrary Map, stores arbitrary channel images (1 or 2 bytes per value) PFM (Pf, PF): Portable Float Map, stores floating-point images (Pf: 1 channel, PF: 3 channel) -Reading +Reading: All formats fill out header fields `format`, `width`, `height`, `channels`, `depth` Specific formats use more fields - PGM, PPM, and PAM set `maxval` - PAM also sets `tupltype`, and is able to set `channels` to an arbitrary value - PFM sets `scale` and `little_endian` + PGM, PPM, and PAM set `maxval` (maximum of 65535) + PAM sets `tupltype` if there is one, and can set `channels` to any value (not just 1 or 3) + PFM sets `scale` (float equivalent of `maxval`) and `little_endian` (endianness of stored floats) Currently doesn't support reading multiple images from one binary-format file -Writing +Writing: + You can use your own `Netpbm_Info` struct to control how images are written All formats require the header field `format` to be specified Additional header fields are required for specific formats - PGM, PPM, and PAM require `maxval` + PGM, PPM, and PAM require `maxval` (maximum of 65535) PAM also uses `tupltype`, though it may be left as default (empty or nil string) - PFM requires `scale` and `little_endian`, though the latter may be left untouched (default is false) + PFM requires `scale`, and optionally `little_endian` Some syntax differences from the specifications: - `channels` stores what the PAM specification calls `depth` - `depth` instead stores how many bytes will fit `maxval` (should only be 1, 2, or 4) + `channels` stores the number of values per pixel, what the PAM specification calls `depth` + `depth` instead is the number of bits for a single value (32 for PFM, 16 or 8 otherwise) `scale` and `little_endian` are separated, so the `header` will always store a positive `scale` `little_endian` will only be true for a negative `scale` PFM, every other format will be false `little_endian` only describes the netpbm data being read/written, the image buffer will be native diff --git a/core/image/netpbm/netpbm.odin b/core/image/netpbm/netpbm.odin index 768c06110..9574faa26 100644 --- a/core/image/netpbm/netpbm.odin +++ b/core/image/netpbm/netpbm.odin @@ -109,7 +109,7 @@ save_to_buffer :: proc(img: ^Image, custom_info: Info = {}, allocator := context // using info so we can just talk about the header using info - //? validation + // validation if header.format in (PBM + PGM + Formats{.Pf}) && img.channels != 1 \ || header.format in (PPM + Formats{.PF}) && img.channels != 3 { err = .Invalid_Number_Of_Channels