{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,8,13]],"date-time":"2026-08-13T11:36:06Z","timestamp":1786620966838,"version":"3.56.0"},"reference-count":27,"publisher":"Wiley","license":[{"start":{"date-parts":[[2026,8,13]],"date-time":"2026-08-13T00:00:00Z","timestamp":1786579200000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"},{"start":{"date-parts":[[2026,8,13]],"date-time":"2026-08-13T00:00:00Z","timestamp":1786579200000},"content-version":"tdm","delay-in-days":0,"URL":"http:\/\/doi.wiley.com\/10.1002\/tdm_license_1.1"}],"content-domain":{"domain":["onlinelibrary.wiley.com"],"crossmark-restriction":true},"short-container-title":["Computer Graphics Forum"],"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:p>3D Gaussian Splatting (3DGS) has recently enabled real\u2010time, high\u2010quality novel view synthesis. However, its formulation treats each Gaussian independently, lacking explicit information exchange between neighboring elements and limiting its ability to capture local structure. This leads to suboptimal trade\u2010offs between reconstruction quality and representation compactness. A natural extension is to introduce interactions among neighboring Gaussians, yet naive message passing often results in training instability and over\u2010smoothing, degrading fine details. In this work, we present FlowGS, a graph\u2010based interaction framework that explicitly models and controls information flow among Gaussians. Our key insight is that neighborhood interaction in 3DGS is fundamentally a problem of controlling information flow, rather than simply enabling feature propagation. To this end, we represent Gaussians as nodes in a spatial graph to define the flow topology, and enable content\u2010aware feature aggregation for information exchange. Crucially, we introduce a learnable dynamic gating mechanism that adaptively controls, for each Gaussian, how much information to preserve from itself and how much to incorporate from its neighbors. We further incorporate lightweight regularization and a simple yet effective stabilization strategy to ensure robust and stable optimization. Extensive experiments show that, under the same training budget as vanilla 3DGS, our method consistently achieves higher reconstruction quality while using significantly fewer Gaussians, with up to 6.74 dB PSNR gain (on the Deep Blending playroom scene) and an average reduction of approximately 64.3% in Gaussian count. These results demonstrate that controlling information flow is key to enabling efficient and high\u2010quality point\u2010based scene representations.<\/jats:p>","DOI":"10.1111\/cgf.70548","type":"journal-article","created":{"date-parts":[[2026,8,13]],"date-time":"2026-08-13T11:04:46Z","timestamp":1786619086000},"update-policy":"https:\/\/doi.org\/10.1002\/crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["FlowGS: Controlling Information Flow for Graph\u2010Based Interaction in 3D Gaussian Splatting"],"prefix":"10.1111","author":[{"ORCID":"https:\/\/orcid.org\/0009-0001-9686-8554","authenticated-orcid":false,"given":"T.","family":"Jia","sequence":"first","affiliation":[{"name":"College of Software Nankai University  Tianjin China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0005-9741-0683","authenticated-orcid":false,"given":"R.","family":"Sun","sequence":"additional","affiliation":[{"name":"College of Software Nankai University  Tianjin China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0009-3067-2100","authenticated-orcid":false,"given":"W.","family":"Shi","sequence":"additional","affiliation":[{"name":"College of Software Nankai University  Tianjin China"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"311","published-online":{"date-parts":[[2026,8,13]]},"reference":[{"key":"e_1_2_7_2_2","unstructured":"BagdasarianM. T. KnollP. LiY.\u2010H. BarthelF. HilsmannA. EisertP. MorgensternW.:3DGS.zip: A Survey on 3D Gaussian Splatting Compression Methods.e70078. 2"},{"key":"e_1_2_7_3_2","doi-asserted-by":"crossref","unstructured":"BarronJ. T. MildenhallB. VerbinD. SrinivasanP. P. HedmanP.: Mip\u2010NeRF 360: Unbounded Anti\u2010Aliased Neural Radiance Fields. InProceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR)(2022) pp.5460\u20135469. doi:10.1109\/CVPR52688.2022.00539. 6 7 8","DOI":"10.1109\/CVPR52688.2022.00539"},{"key":"e_1_2_7_4_2","unstructured":"ChengC. SongG. YaoY. ZhouQ. ZhangG. WangH.: Graph\u2010Guided Scene Reconstruction from Images with 3D Gaussian Splatting. InProceedings of the International Conference on Learning Representations (ICLR)(2025). 2"},{"key":"e_1_2_7_5_2","unstructured":"DarmonF. PorziL. Rota\u2010Bul\u00f2S. KontschiederP.: Robust Gaussian Splatting. InEuropean Conference on Computer Vision (ECCV)(2024). 2"},{"key":"e_1_2_7_6_2","doi-asserted-by":"crossref","unstructured":"Fridovich\u2010KeilS. YuA. TancikM. ChenQ. RechtB. KanazawaA.: Plenoxels: Radiance Fields without Neural Networks. InProceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR)(2022) pp.5491\u20135500. doi:10.1109\/CVPR52688.2022.00542. 1 2","DOI":"10.1109\/CVPR52688.2022.00542"},{"key":"e_1_2_7_7_2","doi-asserted-by":"crossref","unstructured":"FangG. WangB.: Mini\u2010Splatting: Representing Scenes with a Constrained Number of Gaussians. InEuropean Conference on Computer Vision (ECCV)(2024) pp.165\u2013181. 2 7 8","DOI":"10.1007\/978-3-031-72980-5_10"},{"key":"e_1_2_7_8_2","doi-asserted-by":"publisher","DOI":"10.1145\/3272127.3275084"},{"key":"e_1_2_7_9_2","doi-asserted-by":"publisher","DOI":"10.1162\/neco.1997.9.8.1735"},{"key":"e_1_2_7_10_2","doi-asserted-by":"crossref","unstructured":"HuJ. ShenL. AlbanieS. SunG. WuE.: Squeeze\u2010and\u2010Excitation Networks. InProceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR)(2018) pp.7132\u20137141. doi:10.1109\/CVPR.2018.00745. 3","DOI":"10.1109\/CVPR.2018.00745"},{"key":"e_1_2_7_11_2","doi-asserted-by":"crossref","unstructured":"HeK. ZhangX. RenS. SunJ.: Deep Residual Learning for Image Recognition. InProceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR)(2016) pp.770\u2013778. doi:10.1109\/CVPR.2016.90. 3","DOI":"10.1109\/CVPR.2016.90"},{"key":"e_1_2_7_12_2","doi-asserted-by":"publisher","DOI":"10.1145\/3592433"},{"key":"e_1_2_7_13_2","doi-asserted-by":"publisher","DOI":"10.1145\/3072959.3073599"},{"key":"e_1_2_7_14_2","doi-asserted-by":"publisher","DOI":"10.52202\/079017\u20102573"},{"key":"e_1_2_7_15_2","doi-asserted-by":"crossref","unstructured":"LeeJ. C. RhoD. SunX. KoJ. H. ParkE.: Compact 3D Gaussian Representation for Radiance Field. InProceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR)(2024) pp.20777\u201320786. (CVPR 2024 Highlight). doi:10.1109\/CVPR52733.2024.02052. 2 7 8","DOI":"10.1109\/CVPR52733.2024.02052"},{"key":"e_1_2_7_16_2","doi-asserted-by":"crossref","unstructured":"LyuX. SunY.\u2010T. HuangY.\u2010H. WuX. YangZ. ChenY. PangJ. QiX.: 3DGSR: Implicit Surface Reconstruction with 3D Gaussian Splatting.arXiv preprint arXiv:2404.00409(2024). 2","DOI":"10.1145\/3687952"},{"key":"e_1_2_7_17_2","doi-asserted-by":"crossref","unstructured":"LuT. YuM. XuL. XiangliY. WangL. LinD. DaiB.: Scaffold\u2010GS: Structured 3D Gaussians for View\u2010Adaptive Rendering. InProceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR)(2024) pp.20654\u201320664. (CVPR 2024 Highlight). doi:10.1109\/CVPR52733.2024.01952. 7 8","DOI":"10.1109\/CVPR52733.2024.01952"},{"key":"e_1_2_7_18_2","doi-asserted-by":"publisher","DOI":"10.1145\/3528223.3530127"},{"key":"e_1_2_7_19_2","doi-asserted-by":"crossref","unstructured":"MildenhallB. SrinivasanP. P. TancikM. BarronJ. T. RamamoorthiR. NgR.: NeRF: Representing Scenes as Neural Radiance Fields for View Synthesis. InEuropean Conference on Computer Vision (ECCV)(2020) pp.405\u2013421. doi:10.1007\/978\u20103\u2010030\u201058452\u20108_24. 2","DOI":"10.1007\/978-3-030-58452-8_24"},{"key":"e_1_2_7_20_2","doi-asserted-by":"crossref","unstructured":"QiC. R. SuH. MoK. GuibasL. J.: PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation. InProceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR)(2017) pp.652\u2013660. doi:10.1109\/CVPR.2017.16. 3","DOI":"10.1109\/CVPR.2017.16"},{"key":"e_1_2_7_21_2","volume-title":"Attention Residuals: Rethinking Residual Connections for Deep Learning","author":"Team K.","year":"2026"},{"key":"e_1_2_7_22_2","unstructured":"VerverasE. PotamiasR. A. SongJ. DengJ. ZafeiriouS.: SAGS: Structure\u2010Aware 3D Gaussian Splatting. InEuropean Conference on Computer Vision (ECCV)(2024). doi:10.1007\/978\u20103\u2010031\u201072655\u20102_13. 2 7 8"},{"key":"e_1_2_7_23_2","doi-asserted-by":"publisher","DOI":"10.1145\/3326362"},{"key":"e_1_2_7_24_2","doi-asserted-by":"publisher","DOI":"10.1007\/s41095\u2010024\u20100436\u2010y"},{"key":"e_1_2_7_25_2","doi-asserted-by":"crossref","unstructured":"XuQ. XuZ. PhilipJ. BiS. ShuZ. SunkavalliK. NeumannU.: Point\u2010NeRF: Point\u2010Based Neural Radiance Fields. InProceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR)(2022) pp.5438\u20135448. doi:10.1109\/CVPR52688.2022.00536. 1","DOI":"10.1109\/CVPR52688.2022.00536"},{"key":"e_1_2_7_26_2","doi-asserted-by":"crossref","unstructured":"ZhangQ. BaekS.\u2010H. RusinkiewiczS. HeideF.: Differentiable Point\u2010Based Radiance Fields for Efficient View Synthesis. InProceedings of SIGGRAPH Asia 2022 Conference Papers(2022) pp.1\u201312. doi:10.1145\/3550469.3555413. 1","DOI":"10.1145\/3550469.3555413"},{"key":"e_1_2_7_27_2","unstructured":"ZhangS. FeiX. LiuF. SongH. DuanY.: Gaussian Graph Network: Learning Efficient and Generalizable Gaussian Representations from Multi\u2010view Images.arXiv preprint(2025). doi:10.48550\/arXiv.2503.16338. 2"},{"key":"e_1_2_7_28_2","doi-asserted-by":"crossref","unstructured":"ZhaoB. ZhouY. WangZ. WangH.: EG\u2010Gaussian: Epipolar Geometry and Graph Network enhanced 3D Gaussian Splatting.arXiv preprint(2025). doi:10.48550\/arXiv.2504.13540. 1 2 7","DOI":"10.1109\/ICME59968.2025.11209117"}],"container-title":["Computer Graphics Forum"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/onlinelibrary.wiley.com\/doi\/pdf\/10.1111\/cgf.70548","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/onlinelibrary.wiley.com\/doi\/full-xml\/10.1111\/cgf.70548","content-type":"application\/xml","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/onlinelibrary.wiley.com\/doi\/pdf\/10.1111\/cgf.70548","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,8,13]],"date-time":"2026-08-13T11:04:58Z","timestamp":1786619098000},"score":1,"resource":{"primary":{"URL":"https:\/\/onlinelibrary.wiley.com\/doi\/10.1111\/cgf.70548"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,8,13]]},"references-count":27,"alternative-id":["10.1111\/cgf.70548"],"URL":"https:\/\/doi.org\/10.1111\/cgf.70548","archive":["Portico"],"relation":{},"ISSN":["0167-7055","1467-8659"],"issn-type":[{"value":"0167-7055","type":"print"},{"value":"1467-8659","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,8,13]]},"assertion":[{"value":"2026-08-13","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}],"article-number":"e70548"}}