v1.5.3

Highlights

gsplat.rasterization now supports render a batch of scenes in one pass. [In addition to its original support of rendering a batch of cameras per scene]

What's Changed

• Link Jonathan's Tutorial by @liruilong940607 in #688
• Fix format README.md by @liruilong940607 in #689
• Update README.md by @liruilong940607 in #690
• Batchify all APIs by @JunchenLiu77 in #670
• fix bug in gsplat-batch.mp4 by @JunchenLiu77 in #695
• Update LICENSE by @liruilong940607 in #699
• Update Cameras.cuh by @liruilong940607 in #700
• Update IntersectTile.cu by @liruilong940607 in #701
• Alpha shouldn't be normalized or inversed? by @RongLiu-Leo in #704
• Enforce eval3d with UT is set by @liruilong940607 in #712
• Update 3dgut.md by @liruilong940607 in #713
• New experimental feature: fused bilagrid by @ichsan2895 in #706
• Support 2dgs with the latest viewer by @RongLiu-Leo in #707
• fix: solves issue, that opacity is reset at iteration 0 by @kleinicke in #735
• Add option to disable world space normalization for 2DGS by @JojiJoseph in #719
• Fix Fisheye Undistortion. by @liruilong940607 in #736
• fix 3dgut sample viewer by @inuex35 in #734
• Remove unnecessary abs() calls in scale/opacity loss calculation by @vincentwoo in #743
• Fix lack of submodules in CI/CD by @randomze in #751
• F-Theta Camera Model Support by @liruilong940607 in #745
• Bump to 1.5.3 by @liruilong940607 in #752
• update windows server in CI by @liruilong940607 in #753
• Use windows 2022 in CI by @liruilong940607 in #754
• Add "-allow-unsupported-compiler" for win by @liruilong940607 in #755
• Exclude cu118 and cu121 for windows build by @liruilong940607 in #756

New Contributors

• @ichsan2895 made their first contribution in #706
• @kleinicke made their first contribution in #735
• @inuex35 made their first contribution in #734
• @randomze made their first contribution in #751

Full Changelog: v1.5.2...v1.5.3

v1.5.2

Highlights

• 3DGUT is integrated. Now distorted cameras are supported, along with rolling shutter.
• 2DGS now supports camera gradients (camera pose).
• full nerfstudio-style viewer supported in the repo.
• PyPI package fix.

What's Changed

• Fixed redundant PyTest mark that was applied to a fixture. by @findeevy in #642
• Retire 20.04 in workflow by @liruilong940607 in #643
• fix workflow PyPI upload by @liruilong940607 in #644
• add 2dgs viewmats gradient by @FantasticOven2 in #645
• Update rendering.py by @liruilong940607 in #646
• Fix spelling: 'principle' => 'principal' by @brentyi in #647
• Fix up side down flipping for most of the cases by @liruilong940607 in #648
• Fix inverted scene when flipped by @samuelm2 in #656
• New Viewer with nerfview=0.1.2 by @JunchenLiu77 in #657
• Simplify tight rectangular bounding box calculation by @zerolover in #662
• [BUG] Fix strict "invalid mask" in gsplat.utils.save_ply(...) by @Crezle in #663
• Multiple minor fixes and update citation by @liruilong940607 in #664
• Increase maximum allowed far plane in viewer by @vincentwoo in #672
• Implement 3DGUT by @liruilong940607 in #667
• Update nerfview dep by @liruilong940607 in #673
• use c++17 by @liruilong940607 in #675
• expose learning rates in simple_trainer.py by @MrNeRF in #676
• Minor fix by @RongLiu-Leo in #677
• fix instructions by @liruilong940607 in #679
• fix format by @liruilong940607 in #680
• Fix _jit_compile arguments for torch>=2.7.0 by @shiukaheng in #651
• Fix manifest and upgrade to 1.5.2 by @liruilong940607 in #685

New Contributors

• @findeevy made their first contribution in #642
• @samuelm2 made their first contribution in #656
• @zerolover made their first contribution in #662
• @Crezle made their first contribution in #663
• @vincentwoo made their first contribution in #672
• @RongLiu-Leo made their first contribution in #677

Full Changelog: v1.5.1...v1.5.2

v1.5.1

Highlights

New exporter python api: from gsplat import export_splats, supprts three exporting formats (#628, #640):

• .ply: A standard PLY file format. Supported by most viewers.
• .splat: A custom Splat file format. Supported by antimatter15 viewer.
• .ply_compressed: A compressed PLY file format. Used by Supersplat viewer.

What's Changed

• Update publish.yml by @liruilong940607 in #621
• faster fused-ssim by @MrNeRF in #622
• Added Gaussian Splat Exporters by @bolopenguin in #628
• Fix typo in strategy.rst code example by @buesma in #632
• Multiple small fixes by @liruilong940607 in #638
• slightly adjusting atol for test by @liruilong940607 in #639
• 🐛 Fix memory leak in exporters - splat2ply bytes by @mrcabellom in #640
• bump to 1.5.1 by @liruilong940607 in #641

New Contributors

• @bolopenguin made their first contribution in #628
• @buesma made their first contribution in #632
• @mrcabellom made their first contribution in #640

Full Changelog: v1.5.0...v1.5.1

v1.5.0

Highlights

• 3.5x faster compilation (4m19s -> 1m22s) via code refactoring.
• up to 25% faster rasterization function via more accurate GS-tile intersection.
• support CUDA fused Adam Optimizer with mask support (from gsplat.optimizers import SelectiveAdam)

What's Changed

• A set of comments and annotations to 2DGS CUDA kernels by @DylanWaken in #429
• densify after optimizer step. fixes #426 by @rahul-goel in #431
• Fix 2DGS for Correct Normal and Median Depth Calculation by @khokao in #430
• Selective Adam by @rahul-goel in #432
• Fix PyPI upload missing whl file by @liruilong940607 in #454
• add-support-for-non-trainable-params by @nikmo33 in #456
• remove unused code by @georghess in #482
• Reorganize file structure by @JunchenLiu77 in #480
• Fix inconsistency in the comments by @JiangWenPL in #441
• bugfix in rasterization2d_inria by @Golbstein in #499
• use mask in 2dgs by @Golbstein in #497
• Fix proj.cuh include guard by @desmondlzy in #504
• Fix evaluation scripts by @rjbaw in #510
• Reduce ram usage using fisheye as camera model by @aW4KeNiNG in #521
• Implement gaussian splatting ply file saving with o3d by @MrNeRF in #427
• Avoid jpeg compression artifacts when training Mip-NeRF360. by @jefequien in #507
• Update nerfview version in examples to avoid division by zero by @MotivaCG in #558
• minor fix to 2dgs corner case by @FantasticOven2 in #543
• refactor CUDA code for faster compilation and easier extensibility by @liruilong940607 in #583
• Fix minor typo in documentation about rasterization by @kikirizki in #580
• Fix incorrect comments for view matrices by @ScaryPG in #579
• fix the offset of K in projection packed fwd 2dgs kernel by @lapisliu in #569
• Fix channel padding for distortion and median depth computation by @jonathsch in #559
• Add disable_video flag for simple_trainer by @MotivaCG in #557
• fix: wrong torch cat dim when concatenating colors+depths in simple_trainer_2dgs.py by @Crescent-Saturn in #508
• Fixes build issues by @MrNeRF in #585
• move env yml to examples/ by @liruilong940607 in #587
• remove env yml by @liruilong940607 in #588
• Update _backend.py by @liruilong940607 in #589
• fix save ply if else by @liruilong940607 in #594
• Rect cull by @liruilong940607 in #596
• Optionally leverage opacity to tight the radii bounding box by @liruilong940607 in #597
• fix ALPHA_THRESHOLD by @liruilong940607 in #598
• fix setup.py for build by @liruilong940607 in #602
• Disable Auto Precompiled Header. Support Enabling it Manually. by @liruilong940607 in #603
• fix tight opa by @liruilong940607 in #606
• Fix build action: Upgraded upload-artifact@v3 / download-artifact@v3 -> upload-artifact@v4 / download-artifact@v4 by @shiukaheng in #616
• fix adam shape issue by @liruilong940607 in #618
• delay import for _check_and_build_extension_h_precompiler_headers by @liruilong940607 in #619
• bump to 1.5.0 by @liruilong940607 in #620

New Contributors

• @DylanWaken made their first contribution in #429
• @khokao made their first contribution in #430
• @nikmo33 made their first contribution in #456
• @georghess made their first contribution in #482
• @JunchenLiu77 made their first contribution in #480
• @JiangWenPL made their first contribution in #441
• @Golbstein made their first contribution in #499
• @desmondlzy made their first contribution in #504
• @rjbaw made their first contribution in #510
• @aW4KeNiNG made their first contribution in #521
• @MotivaCG made their first contribution in #558
• @kikirizki made their first contribution in #580
• @ScaryPG made their first contribution in #579
• @lapisliu made their first contribution in #569
• @jonathsch made their first contribution in #559
• @Crescent-Saturn made their first contribution in #508
• @shiukaheng made their first contribution in #616

Full Changelog: v1.4.0...v1.5.0

v1.4.0

Highlights

• support 2DGS
• Camera model: perspective, orthogonal, fish-eye
• Start to support pre-built wheels (python 10)

What's Changed

• Update build workflow by @liruilong940607 in #369
• fix (apparant) typos in compression docs by @christian-helms in #370
• eval supports multi-GPU by @liruilong940607 in #371
• correct a typo to fix #373 by @jckhng in #383
• Add option to render camera trajectory as an ellipse. by @jefequien in #380
• Add an option to disable world space normalization in examples/simple_trainer.py by @JojiJoseph in #379
• Fix points projection when sampling patches for training by @nnop in #386
• fix reshape error when multi GPU bachsize > 1 not packed by @Ultized in #382
• Fix typo by @cr333 in #392
• Make download script work on Windows, too by @cr333 in #397
• Fused differentiable SSIM by @rahul-goel in #396
• 2DGS by @FantasticOven2 in #208
• citation.bib by @maturk in #400
• Implement Bilateral Guided Gaussian Splatting. by @jefequien in #291
• Bump fused-ssim commit by @liruilong940607 in #405
• Update rendering.py by @liruilong940607 in #407
• Update CITATION.bib by @cr333 in #410
• Make commands easier in README by @MrNeRF in #420
• Implement Fisheye-GS. by @jefequien in #398
• Publish precompiled wheels in Github release and add wheel index by @martinResearch in #365
• 1.4.0 by @liruilong940607 in #421
• fix workflow by @liruilong940607 in #422
• Workflow fix by @liruilong940607 in #423

New Contributors

• @christian-helms made their first contribution in #370
• @jckhng made their first contribution in #383
• @JojiJoseph made their first contribution in #379
• @nnop made their first contribution in #386
• @Ultized made their first contribution in #382
• @cr333 made their first contribution in #392
• @rahul-goel made their first contribution in #396
• @FantasticOven2 made their first contribution in #208
• @MrNeRF made their first contribution in #420
• @martinResearch made their first contribution in #365

Full Changelog: v1.3.0...v1.4.0

v1.3.0

Highlight

• Support 3DGS Compression
• Support orthographic projection
• Support TanksAndTemples Dataset
• Various Bug Fix

What's Changed

• Make sure the build directory exists. by @keunhong in #319
• docs: update README.md by @eltociear in #322
• Fix typo in examples/simple_viewer.py by @simo23 in #321
• Consolidate simple_trainer.py and simple_trainer_mcmc.py by @brentyi in #325
• Compressing gaussian splats by @jefequien in #309
• Simplify tyro usage in simple_trainer.py by @brentyi in #332
• Cleanup C/CUDA code for C users by @liruilong940607 in #333
• Patch to make Strategy compatible with splatfacto by @kerrj in #334
• Fix minor typo by @zhiguotang5 in #342
• correct distCoeffs of SIMPLE_RADIAL by @Master-cai in #337
• Add np.random.choice fallback for many Gaussians exceeding torch.multinomial limits by @soskek in #338
• data parser for t&t dataset by @liruilong940607 in #347
• Added orthographic projection by @VladislavZavadskyy in #349
• Fix bug related to refine_scale2d and add pause_refine_after_reset to default strategy by @jb-ye in #354
• tt compression script by @liruilong940607 in #357
• Update to 1.3.0 by @liruilong940607 in #366

New Contributors

• @keunhong made their first contribution in #319
• @simo23 made their first contribution in #321
• @zhiguotang5 made their first contribution in #342
• @Master-cai made their first contribution in #337
• @soskek made their first contribution in #338
• @VladislavZavadskyy made their first contribution in #349

Full Changelog: v1.2.0...v1.3.0

v1.2.0

Highlight

Multi-GPU distributed rasterization is supported! E.g. 4 GPUs could lead to more than 3x speedup as well as 3x less memory usage at the same time.

What's Changed

• [Docs] Evaluation metrics for mcmc, antialiased, and absgrad. by @maturk in #300
• Update requirements.txt by @liruilong940607 in #308
• Fix projection for images with non-centered camera (e.g. crops) by @JonathonLuiten in #305
• prevent race condition when JIT in multiprocess by @liruilong940607 in #312
• Support Multi-GPU training based on the paper "On Scaling Up 3D Gaussian Splatting Training" by @liruilong940607 in #253
• Update version.py by @liruilong940607 in #313

New Contributors

• @JonathonLuiten made their first contribution in #305

Full Changelog: v1.1.1...v1.2.0

v0.1.13

Full Changelog: v0.1.12...v0.1.13

v1.1.1

What's Changed

hot fix

Full Changelog: v1.1.0...v1.1.1

v1.1.0

Highlight

Multiple densification strategies are now supported in gsplat, including

• DefaultStrategy(): The original 3DGS densification process.
• DefaultStrategy(absgrad=True): Densification process from "AbsGS: Recovering Fine Details for 3D Gaussian Splatting"
• MCMCStrategy(). Densification process from "3D Gaussian Splatting as Markov Chain Monte Carlo"

What's Changed

• Update 3DGS examples to use the latest nerfview from pypi by @hangg7 in #206
• docs: update README.md by @eltociear in #210
• Added a guidance for installing gsplat on Windows by @KevinXu02 in #213
• Minor fix for _QuatScaleToCovarPreci back with sparse grad by @liruilong940607 in #214
• Fix wrong description in rasterization API doc by @yifanlu0227 in #216
• support 2^n+1 channels for rasterize_to_pixels by @liruilong940607 in #218
• workaround for n-dim rendering with n>32 by @wuzirui in #217
• Minor fix for _RasterizeToPixels back to avoid NaNs by @opipari in #235
• clean up the shape of the colors by @liruilong940607 in #228
• static shared mem -> dynamic shared mem by @liruilong940607 in #229
• updating background for different rendering modes [D, ED, RGB+D, RGB+… by @vye16 in https://github.com//pull/240
• Fix max_range in _torch_impl._rasterize_to_pixels() by @liruilong940607 in #242
• Implement 3DGS-MCMC in gsplat. by @jefequien in #238
• sqrt int to float by @liruilong940607 in #246
• Clamp ratios in MCMC's compute_relocation by @yzslab in #247
• add poetry files, and fix numpy error and no data error. by @MasahiroOgawa in #251
• Separate C source into many translation units by @fwilliams in #255
• Fwilliams/fix headers and ninja by @fwilliams in #256
• template everything but don't actually support multiple precisions yet by @fwilliams in #259
• fix glm path in JIT building by @liruilong940607 in #260
• Fwilliams/atomics by @fwilliams in #261
• set the MAX_JOBS to 10 if not set. by @liruilong940607 in #262
• float3 -> vec3; float2 -> vec2; by @liruilong940607 in #267
• Start adding support for double, float16, and bfloat16 by @fwilliams in #270
• fix test_fully_fused_projection_packed test by @liruilong940607 in #289
• Add optional masks in rasterize_to_pixels() to support Grendel by @alexis-mmm in #284
• Abstracting out the GS heuristics into a Strategy class by @liruilong940607 in #278
• Bump to 1.1.0 by @liruilong940607 in #290

New Contributors

• @yifanlu0227 made their first contribution in #216
• @wuzirui made their first contribution in #217
• @opipari made their first contribution in #235
• @jefequien made their first contribution in #238
• @yzslab made their first contribution in #247
• @MasahiroOgawa made their first contribution in #251
• @fwilliams made their first contribution in #255
• @alexis-mmm made their first contribution in #284

Full Changelog: v1.0.0...v1.1.0