Base created sample with VITE
- Load GLTF
- + Draco (compress only mesh)
- + Quantize (Quantize geometry, reducing precision and memory)
- Load KTX2
- Load Basis
- + Meshopt (compress mesh + animations)
*** gltf-transform ***
Download from: https://gltf-transform.donmccurdy.com/cli.html
npm install --global @gltf-transform/cli*** toktx *** Download from: https://github.com/KhronosGroup/KTX-Software/releases
gltf-transform draco DamagedHelmetEmmbed.gltf DamagedHelmetEmmbedDraco.gltf
info: DamagedHelmetEmmbed.gltf (5.03 MB) → DamagedHelmetEmmbedDraco.gltf (3.29 MB)
------
gltf-transform draco DamagedHelmetEmmbed.gltf DamagedHelmetDracoUltra.gltf --encode-speed 10
info: DamagedHelmetEmmbed.gltf (5.03 MB) → DamagedHelmetDracoUltra.gltf (3.33 MB)gltf-transform draco DamagedHelmetEmmbed.gltf DamagedHelmetDraco.gltf
info: DamagedHelmetEmmbed.gltf (5.03 MB) → DamagedHelmetDraco.gltf (3.29 MB)gltf-transform meshopt DamagedHelmetEmmbed.gltf DamagedHelmetMeshopt.gltf
info: prune: Removed types... Accessor (4)
warn: quantize: Skipping TEXCOORD_0; out of [0,1] range.
info: prune: Removed types... Accessor (1)
info: DamagedHelmetEmmbed.gltf (5.03 MB) → DamagedHelmetMeshopt.gltf (3.4 MB)gltf-transform etc1s DamagedHelmetEmmbed.gltf DamagedHelmetKTX-etc1s.gltf
info: DamagedHelmetEmmbed.gltf (5.03 MB) → DamagedHelmetKTX-etc1s.gltf (2.65 MB)gltf-transform uastc DamagedHelmetEmmbed.gltf
info: DamagedHelmetEmmbed.gltf (5.03 MB) → DamagedHelmetKTX-uastc.gltf (13.44 MB)-
Preview: https://github.com/donmccurdy/three-gltf-viewer/blob/main/src/viewer.js
-
Draco: https://github.com/KhronosGroup/glTF/tree/main/extensions/2.0/Khronos/KHR_draco_mesh_compression
-
basis: https://discourse.threejs.org/t/compressed-texture-workflow-gltf-basis/10039/8
▸ gltf-transform draco <input> <output> [OPTIONS...]
Compress mesh geometry with the Draco library. This type of compression affects only geometry data — animation and textures are not compressed.
Compresses
- geometry (only triangle meshes)
Documentation
References
- draco: https://github.com/google/draco
- KHR_draco_mesh_compression: https://github.com/KhronosGroup/glTF/blob/master/extensions/2.0/Khronos/KHR_draco_mesh_compression/
ARGUMENTS
<input> Path to read glTF 2.0 (.glb, .gltf) model
<output> Path to write output
OPTIONS
--decode-speed <decodeSpeed> Decoding speed vs. compression level, 1–10.
number, default: 5
--encode-speed <encodeSpeed> Encoding speed vs. compression level, 1–10.
number, default: 5
--method <method> Compression method.
one of "edgebreaker","sequential", default:
"edgebreaker"
--quantization-volume <volume> Bounds for quantization grid.
one of "mesh","scene", default: "mesh"
--quantize-color <bits> Quantization bits for COLOR_*, 1-16.
number, default: 8
--quantize-generic <bits> Quantization bits for other attributes, 1-16.
number, default: 12
--quantize-normal <bits> Quantization bits for NORMAL, 1-16.
number, default: 10
--quantize-position <bits> Quantization bits for POSITION, 1-16.
number, default: 14
--quantize-texcoord <bits> Quantization bits for TEXCOORD_*, 1-16.
number, default: 12
GLOBAL OPTIONS
-h, --help Display global help or command-related help.
-V, --version Display version.
-v, --verbose Verbose mode: will also output debug messages.
--allow-http Allows reads from HTTP requests.
boolean
--vertex-layout <layout> Vertex buffer layout preset.
one of "interleaved","separate", default:
"interleaved"
SUMMARY
Compresses textures in the given file to .ktx2 GPU textures using the
ETC1S Basis Universal bitstream. GPU textures offer faster GPU upload
and less GPU memory consumption than traditional PNG or JPEG textures,
which are fully uncompressed in GPU memory. GPU texture formats require
more attention to compression settings to get similar visual results.
ETC1S, one of the two Basis Universal bitstreams, offers lower size and lower
quality than UASTC. In some cases it may be useful to increase the resolution
of the texture, to minimize compression artifacts while still retaining an
overall smaller filesize. Consider using less aggressive compression settings
for normal maps than for other texture types: you may want to use UASTC for
normal maps and ETC1S for other textures, for example.
Documentation:
https://gltf-transform.donmccurdy.com/extensions.html#khr_texture_basisu
Dependencies:
KTX-Software (https://github.com/KhronosGroup/KTX-Software/)
ARGUMENTS
<input> Path to read glTF 2.0 (.glb, .gltf) model
<output> Path to write output
OPTIONS
--compression <clevel> Compression level, an encoding speed vs. quality
tradeoff. Higher values are slower, but give higher
quality. Try --quality before experimenting with this
option.
one of 0,1,2,3,4,5, default: 1
--filter <filter> Specifies the filter to use when generating mipmaps.
one of
"box","tent","bell","b-spline","mitchell",
"lanczos3","lanczos4","lanczos6","lanczos12",
"blackman","kaiser","gaussian","catmullrom",
"quadratic_interp","quadratic_approx","quadratic_
mix", default: "lanczos4"
--filter-scale <fscale> Specifies the filter scale to use when generating
mipmaps.
number, default: 1
--max-endpoints <max_endpoints> Manually set the maximum number of color endpoint
clusters from 1-16128.
number
--max-selectors <max_selectors> Manually set the maximum number of color selector
clusters from 1-16128.
number
--power-of-two Resizes any non-power-of-two textures to the closest
power-of-two dimensions, not exceeding 2048x2048px.
Required for compatibility on some older devices and
APIs, particularly WebGL 1.0.
boolean
--quality <qlevel> Quality level. Range is 1 - 255. Lower gives better
compression, lower quality, and faster encoding. Higher
gives less compression, higher quality, and slower
encoding. Quality level determines values of
--max_endpoints and --max-selectors, unless those
values are explicitly set.
number, default: 128
--rdo-off Disable endpoint and selector RDO (slightly faster,
less noisy output, but lower quality per output bit).
boolean
--rdo-threshold <rdo_threshold> Set endpoint and selector RDO quality threshold. Lower
is higher quality but less quality per output bit (try
1.0-3.0). Overrides --quality.
number
--slots <slots> Texture slots to include (glob)
default: "*"
GLOBAL OPTIONS
-h, --help Display global help or command-related help.
-V, --version Display version.
-v, --verbose Verbose mode: will also output debug messages.
--allow-http Allows reads from HTTP requests.
boolean
--vertex-layout <layout> Vertex buffer layout preset.
one of "interleaved","separate", default:
"interleaved"
SUMMARY
Compresses textures in the given file to .ktx2 GPU textures using the
UASTC Basis Universal bitstream. GPU textures offer faster GPU upload
and less GPU memory consumption than traditional PNG or JPEG textures,
which are fully uncompressed in GPU memory. GPU texture formats require
more attention to compression settings to get similar visual results.
UASTC, one of the two Basis Universal bitstreams, offers higher size and higher
quality than ETC1S. While it is suitable for all texture types, you may find it
useful to apply UASTC only where higher quality is necessary, and apply ETC1S
for textures where the quality is sufficient.
Documentation:
https://gltf-transform.donmccurdy.com/extensions.html#khr_texture_basisu
Dependencies:
KTX-Software (https://github.com/KhronosGroup/KTX-Software/)
ARGUMENTS
<input> Path to read glTF 2.0 (.glb, .gltf) model
<output> Path to write output
OPTIONS
--filter <filter> Specifies the filter to use when generating mipmaps.
one of
"box","tent","bell","b-spline","mitchell",
"lanczos3","lanczos4","lanczos6","lanczos12",
"blackman","kaiser","gaussian","catmullrom",
"quadratic_interp","quadratic_approx","quadratic_
mix", default: "lanczos4"
--filter-scale <fscale> Specifies the filter scale to use when generating
mipmaps.
number, default: 1
--level <level> Create a texture in high-quality transcodable UASTC
format. The optional parameter <level> selects a speed
vs quality tradeoff as shown in the following table:
Level | Speed | Quality
——————|———————————|————————
0 | Fastest | 43.45dB
1 | Faster | 46.49dB
2 | Default | 47.47dB
3 | Slower | 48.01dB
4 | Very slow | 48.24dB
one of 0,1,2,3,4, default: 2
--power-of-two Resizes any non-power-of-two textures to the closest
power-of-two dimensions, not exceeding 2048x2048px.
Required for compatibility on some older devices and
APIs, particularly WebGL 1.0.
boolean
--rdo <uastc_rdo_l> Enable UASTC RDO post-processing and optionally set
UASTC RDO quality scalar (lambda). Lower values yield
higher quality/larger LZ compressed files, higher
values yield lower quality/smaller LZ compressed files.
A good range to try is [.25, 10]. For normal maps, try
[.25, .75]. Full range is [.001, 10.0].
number, default: 0
--rdo-block-scale <uastc_rdo_b> Set UASTC RDO max smooth block error scale. Range is
[1.0, 300.0]. Default is 10.0, 1.0 is disabled. Larger
values suppress more artifacts (and allocate more bits)
on smooth blocks.
number, default: 10
--rdo-dictionary-size <uastc_rdo_d> Set UASTC RDO dictionary size in bytes. Default is
32768. Lower values=faster, but give less compression.
Possible range is [256, 65536].
number, default: 32768
--rdo-multithreading <uastc_rdo_m> Enable RDO multithreading (slightly lower compression,
non-deterministic).
boolean, default: true
--rdo-std-dev <uastc_rdo_s> Set UASTC RDO max smooth block standard deviation.
Range is [.01, 65536.0]. Default is 18.0. Larger values
expand the range of blocks considered smooth.
number, default: 18
--slots <slots> Texture slots to include (glob)
default: "*"
--zstd <compressionLevel> Supercompress the data with Zstandard. Compression
level range is [1, 22], or 0 is uncompressed. Lower
values decode faster but offer less compression. Values
above 20 should be used with caution, requiring more
memory to decompress:
Level | Window Size |
——————|—————————————|
1 | 256 KB |
… | … |
10 | 2 MB |
… | … |
18 | 8 MB |
19 | 8 MB |
20 | 34 MB |
21 | 67 MB |
22 | 134 MB |
number, default: 18
GLOBAL OPTIONS
-h, --help Display global help or command-related help.
-V, --version Display version.
-v, --verbose Verbose mode: will also output debug messages.
--allow-http Allows reads from HTTP requests.
boolean
--vertex-layout <layout> Vertex buffer layout preset.
one of "interleaved","separate", default:
"interleaved"