{"paper":{"title":"A Single Atlas is All You Need: Decoder-Side Gaussian Splatting for Immersive Video","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Decoder-side Gaussian splatting from a single compressed atlas outperforms depth estimation for immersive video quality and consistency.","cross_cats":["cs.MM","eess.IV"],"primary_cat":"cs.GR","authors_text":"Dawid Mieloch, Stuart Perry","submitted_at":"2026-05-16T13:54:36Z","abstract_excerpt":"Immersive video delivery is bottlenecked by pixel-rate constraints, making the transmission of high-resolution depth maps or explicit 3D volumetric data expensive. Decoder-Side Depth Estimation (DSDE) shifts depth computation to the client, but struggles with complex geometries, inter-view flickering, and non-Lambertian reflections. Conversely, 3D Gaussian Splatting (3DGS) offers state-of-the-art view synthesis, but transmitting splats (or their projected 2D maps) incurs prohibitive bandwidth costs and is poorly aligned with standard video codecs. We propose Decoder-Side Gaussian Splatting (DS"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Under extreme view sparsity (one 2D atlas comprising 4 input views), DSGS achieves a +5.79 dB BD-PSNR and +0.054 BD-SSIM gain over the DSDE anchor while reducing maximum inter-view Delta IV-PSNR from 17.2 dB to 6.4 dB.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"Lossy compression in the transmitted bitstream functions as an implicit low-pass filter that stabilizes feed-forward splat prediction without introducing artifacts that degrade the final rendered views.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"DSGS performs feed-forward 3D Gaussian Splatting on the decoder from one compressed 2D atlas of four views, yielding +5.79 dB BD-PSNR and lower inter-view variance than decoder-side depth estimation.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Decoder-side Gaussian splatting from a single compressed atlas outperforms depth estimation for immersive video quality and consistency.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"928c71294b8200b1a8d936e77d46160554b51eb0d6caa621a366b017d575433c"},"source":{"id":"2605.17002","kind":"arxiv","version":1},"verdict":{"id":"7c320dd7-03ee-4fde-9d6e-1d41eb680b0d","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-19T18:40:04.254605Z","strongest_claim":"Under extreme view sparsity (one 2D atlas comprising 4 input views), DSGS achieves a +5.79 dB BD-PSNR and +0.054 BD-SSIM gain over the DSDE anchor while reducing maximum inter-view Delta IV-PSNR from 17.2 dB to 6.4 dB.","one_line_summary":"DSGS performs feed-forward 3D Gaussian Splatting on the decoder from one compressed 2D atlas of four views, yielding +5.79 dB BD-PSNR and lower inter-view variance than decoder-side depth estimation.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"Lossy compression in the transmitted bitstream functions as an implicit low-pass filter that stabilizes feed-forward splat prediction without introducing artifacts that degrade the final rendered views.","pith_extraction_headline":"Decoder-side Gaussian splatting from a single compressed atlas outperforms depth estimation for immersive video quality and consistency."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.17002/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"citation_quote_validity","ran_at":"2026-05-19T19:49:56.291895Z","status":"completed","version":"0.1.0","findings_count":0},{"name":"cited_work_retraction","ran_at":"2026-05-19T19:23:34.925225Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_title_agreement","ran_at":"2026-05-19T19:01:18.829469Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T18:51:20.748701Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"claim_evidence","ran_at":"2026-05-19T18:41:56.197973Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"ai_meta_artifact","ran_at":"2026-05-19T18:33:26.287891Z","status":"skipped","version":"1.0.0","findings_count":0}],"snapshot_sha256":"a222119fb2919156aedf8dc2365a0a12d6cdd858d20164b336ec299d1eac45fc"},"references":{"count":34,"sample":[{"doi":"","year":2021,"title":"MPEG immersive video coding st andard,","work_id":"fa32bad6-7c9f-4aea-8743-0c1d01b561d2","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2022,"title":"Overview and efficiency of deco der-side depth estimation in MPEG immersive video,","work_id":"304d402b-6ad9-4eba-bc51-30d5055404e1","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"A new approach to decoder-side depth estimation in immersive video transmission,","work_id":"6eb688d8-3d37-4a16-b881-6d7fe4d3a5dd","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"Non-Lam bertian Surfaces and Their Challenges for Visual SLAM,","work_id":"44525c70-0d60-484d-a131-2e78f5a307df","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"3D Gaussian splatting for real-tim e radiance field rendering,","work_id":"0eb012b8-6baa-4189-afd0-faf4a4bc54fd","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":34,"snapshot_sha256":"184036ce39f30da764011273e94efd926070dfac41e3404092292558d6b453ed","internal_anchors":1},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}