Gaussian Splashing Enables Direct Volumetric Rendering Underwater

preprint OA: closed CC-BY-4.0
📄 Open PDF Full text JSON View at publisher
AI-generated deep summary by claude@2026-07, 2026-07-03 · read from full text

The paper studies fast 3D reconstruction and novel-view volumetric rendering for underwater scenes, where scattering and geometric imaging conditions occlude scene features and cause failures of in-air methods such as NeRF and 3D Gaussian Splatting. Using an approach called Gaussian Splashing, the authors start from an SfM-derived point cloud and camera poses, then optimize underwater rendering via a scattering-aware rendering equation, depth estimation, and a modified 3DGS loss function, including periodic re-estimation of backscatter for convergence. They report reconstruction taking only minutes and rendering at up to 140 FPS, with improved distant-scene clarity and comparisons showing better or comparable quality to existing underwater splatting methods, while the preprint explicitly notes it has not been peer reviewed. This paper does not explicitly discuss endometriosis or adenomyosis; it was included in the corpus via a keyword match in the upstream search index.

Read from the paper's body, not the abstract. Not a substitute for reading the paper. No clinical advice. How this works

Abstract

Abstract In underwater images, most useful features are occluded by water. The extent of the occlusion depends on imaging geometry and can vary even across a sequence of burst images. As a result, 3D reconstruction methods robust on in-air scenes,like Neural Radiance Field methods (NeRFs) or 3D Gaussian Splatting (3DGS), fail on underwater scenes. While a recent underwater adaptation of NeRFs achieved state-of-the-art results, it is impractically slow: reconstruction takes hours and its rendering rate, in frames per second (FPS), is less than 1. Here, we present a new method that takes only a few minutes for reconstruction and renders novel underwater scenes at 140 FPS. Named Gaussian Splashing, our method unifies the strengths and speed of 3DGS with an image formation model for capturing scattering, introducing innovations in the rendering and depth estimation procedures and in the 3DGS loss function. Despite the complexities of underwater adaptation, our method produces images at unparalleled speeds with superior details. Moreover, it reveals distant scene details with far greater clarity than other methods, dramatically improving reconstructed and rendered images. We demonstrate results on existing datasets and a new dataset we have collected.
Full text 17,221 characters · extracted from preprint-html · click to expand
Gaussian Splashing Enables Direct Volumetric Rendering Underwater | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Gaussian Splashing Enables Direct Volumetric Rendering Underwater Nir Mualem, Roy Amoyal, Oren Freifeld, Derya Akkaynak This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9611886/v1 This work is licensed under a CC BY 4.0 License Status: Under Revision Version 1 posted 10 You are reading this latest preprint version Abstract In underwater images, most useful features are occluded by water. The extent of the occlusion depends on imaging geometry and can vary even across a sequence of burst images. As a result, 3D reconstruction methods robust on in-air scenes,like Neural Radiance Field methods (NeRFs) or 3D Gaussian Splatting (3DGS), fail on underwater scenes. While a recent underwater adaptation of NeRFs achieved state-of-the-art results, it is impractically slow: reconstruction takes hours and its rendering rate, in frames per second (FPS), is less than 1. Here, we present a new method that takes only a few minutes for reconstruction and renders novel underwater scenes at 140 FPS. Named Gaussian Splashing, our method unifies the strengths and speed of 3DGS with an image formation model for capturing scattering, introducing innovations in the rendering and depth estimation procedures and in the 3DGS loss function. Despite the complexities of underwater adaptation, our method produces images at unparalleled speeds with superior details. Moreover, it reveals distant scene details with far greater clarity than other methods, dramatically improving reconstructed and rendered images. We demonstrate results on existing datasets and a new dataset we have collected. Physical sciences/Engineering Physical sciences/Mathematics and computing Earth and environmental sciences/Ocean sciences Physical sciences/Optics and photonics Figures Figure 1 Figure 2 Figure 3 Figure 4 Full Text Additional Declarations No competing interests reported. Supplementary Files 02supplementaryinformation.pdf Cite Share Download PDF Status: Under Revision Version 1 posted Editorial decision: Revision requested 18 May, 2026 Reviews received at journal 14 May, 2026 Reviews received at journal 11 May, 2026 Reviewers agreed at journal 11 May, 2026 Reviewers agreed at journal 11 May, 2026 Reviewers invited by journal 11 May, 2026 Editor assigned by journal 11 May, 2026 Editor invited by journal 11 May, 2026 Submission checks completed at journal 09 May, 2026 First submitted to journal 09 May, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-9611886","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":638646402,"identity":"fe8f1ef8-4eac-498b-837d-fbda44151b22","order_by":0,"name":"Nir Mualem","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA1UlEQVRIiWNgGAWjYDACZhBRAWayQYUYGw8Q1nIGrAGupQG/FrCaNhQtDAx4tZiz8z7dzDvvjpz8/Aa2h1/bGOT5G5jx22LZzG52m3fbM2ODYwzsxrJtDIYzDhBwmMFhNjaglsOJG4AOk5ZsY2DcQMgvEC1zDifOb4NosSdSS8PhxIZjDGySH9sYEonScnPOMZBfEtuNGc5JJM84TEjL+WNsN97UAEOs+fCxhz/KbGz729sfPsCnBQpAxjI2MPMwSEAjlzgtQE0/iFM9CkbBKBgFIwwAAMS+SPodARujAAAAAElFTkSuQmCC","orcid":"","institution":"Ben-Gurion University of the Negev","correspondingAuthor":true,"prefix":"","firstName":"Nir","middleName":"","lastName":"Mualem","suffix":""},{"id":638646403,"identity":"e597709a-2222-4cc4-9d0e-b9096f4e157d","order_by":1,"name":"Roy Amoyal","email":"","orcid":"","institution":"Ben-Gurion University of the Negev","correspondingAuthor":false,"prefix":"","firstName":"Roy","middleName":"","lastName":"Amoyal","suffix":""},{"id":638646404,"identity":"40bd2351-81a5-4e68-88d0-523535809280","order_by":2,"name":"Oren Freifeld","email":"","orcid":"","institution":"Ben-Gurion University of the Negev","correspondingAuthor":false,"prefix":"","firstName":"Oren","middleName":"","lastName":"Freifeld","suffix":""},{"id":638646405,"identity":"0a622741-f564-4a66-8530-bebc0d0a1e7d","order_by":3,"name":"Derya Akkaynak","email":"","orcid":"","institution":"University of Haifa","correspondingAuthor":false,"prefix":"","firstName":"Derya","middleName":"","lastName":"Akkaynak","suffix":""}],"badges":[],"createdAt":"2026-05-04 20:23:57","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9611886/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9611886/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":109249106,"identity":"aa3a7a56-51f2-4ec4-8c4a-ec9ead4b7671","added_by":"auto","created_at":"2026-05-14 08:42:43","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":11408,"visible":true,"origin":"","legend":"\u003cp\u003ePerformance comparison. Comparison of methods by Frames Per Second (FPS) and Peak Signal-to-Noise Ratio (PSNR). Our approach (red diamonds) achieves a PSNR average of 29.11 while maintaining high inference speeds for real-time rendering. The PSNR values were averaged over the Red Sea, Curacao, Panama, and TableDB datasets; the FPS values are fairly consistent across datasets.\u003c/p\u003e","description":"","filename":"03figure1performancecomparison.png","url":"https://assets-eu.researchsquare.com/files/rs-9611886/v1/d219cd6fd50f67ac3c148d19.png"},{"id":109249144,"identity":"e3be5969-a874-4cfa-b567-d37e000ded87","added_by":"auto","created_at":"2026-05-14 08:42:51","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":322928,"visible":true,"origin":"","legend":"\u003cp\u003eMethod overview. We first use Structure from Motion (SfM) to obtain an initial point cloud and camera poses. We then optimize the model using our underwater rendering equation and modified tile rasterization so that scattering-related distortions are accounted for directly in the rendering process. Backscatter is re-estimated every 500 steps to guide convergence toward accurate medium coefficients. The base figure is adapted from the original 3D Gaussian Splatting method in [14].\u003c/p\u003e","description":"","filename":"04figure2methodoverview.png","url":"https://assets-eu.researchsquare.com/files/rs-9611886/v1/944e2f0d76e8f606949a33d0.png"},{"id":109297660,"identity":"f1edf33d-a9f5-43ca-9e52-d86512c4cbec","added_by":"auto","created_at":"2026-05-15 09:01:36","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":25441168,"visible":true,"origin":"","legend":"\u003cp\u003eVisual comparison of novel views. Row 1 shows an example from the Red Sea scene. Rows 3, 5, and 7 show examples from the TableDB dataset. Rows 2, 4, and 6 provide zoomed-in views of the red rectangles highlighted in Rows 1, 3, and 5. Row 7 highlights STNeRFacto’s failure on unbounded scenes.\u003c/p\u003e","description":"","filename":"05figure3visualcomparisonnovelviews.png","url":"https://assets-eu.researchsquare.com/files/rs-9611886/v1/b0fd29da677808f817c08463.png"},{"id":109215100,"identity":"e37de1cd-6e93-4abb-9af1-b79229175931","added_by":"auto","created_at":"2026-05-13 17:51:54","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":29587832,"visible":true,"origin":"","legend":"\u003cp\u003eVisual comparison across underwater splatting methods. Novel views rendered by different underwater splatting methods are compared. Our results are better than SeaSplat’s and are comparable to those of WaterSplatting, while remaining substantially faster. In regions far from the cameras, our method typically yields sharper reconstructions than WaterSplatting.\u003c/p\u003e","description":"","filename":"06figure4underwatersplattingcomparison.png","url":"https://assets-eu.researchsquare.com/files/rs-9611886/v1/d4e4eab3845ba920dcc07c72.png"},{"id":109300647,"identity":"0d6c0148-915d-44fb-b1eb-e755a3afa2f1","added_by":"auto","created_at":"2026-05-15 09:22:51","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":31232314,"visible":true,"origin":"","legend":"","description":"","filename":"01manuscriptlatexsource.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9611886/v1_covered_70294be8-f9ea-4308-931f-1acd7eecd2d5.pdf"},{"id":109215097,"identity":"d1c7c1bc-6bba-4e60-b05a-8188c15d1325","added_by":"auto","created_at":"2026-05-13 17:51:54","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":16786925,"visible":true,"origin":"","legend":"","description":"","filename":"02supplementaryinformation.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9611886/v1/a710aca5a1c61fa9e0db327e.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Gaussian Splashing Enables Direct Volumetric Rendering Underwater","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-9611886/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9611886/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"In underwater images, most useful features are occluded by water. The extent of the occlusion depends on imaging geometry and can vary even across a sequence of burst images. As a result, 3D reconstruction methods robust on in-air scenes,like Neural Radiance Field methods (NeRFs) or 3D Gaussian Splatting (3DGS), fail on underwater scenes. While a recent underwater adaptation of NeRFs achieved state-of-the-art results, it is impractically slow: reconstruction takes hours and its rendering rate, in frames per second (FPS), is less than 1. Here, we present a new method that takes only a few minutes for reconstruction and renders novel underwater scenes at 140 FPS. Named Gaussian Splashing, our method unifies the strengths and speed of 3DGS with an image formation model for capturing scattering, introducing innovations in the rendering and depth estimation procedures and in the 3DGS loss function. Despite the complexities of underwater adaptation, our method produces images at unparalleled speeds with superior details. Moreover, it reveals distant scene details with far greater clarity than other methods, dramatically improving reconstructed and rendered images. We demonstrate results on existing datasets and a new dataset we have collected.","manuscriptTitle":"Gaussian Splashing Enables Direct Volumetric Rendering Underwater","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-05-13 17:51:44","doi":"10.21203/rs.3.rs-9611886/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2026-05-18T09:33:00+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-05-14T13:50:02+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-05-12T02:48:21+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"134274833076557886635537381881750131388","date":"2026-05-12T00:27:44+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"115548658825137598935944456706803422655","date":"2026-05-12T00:01:17+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-05-11T23:59:02+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-05-11T23:55:52+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2026-05-11T14:42:53+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-05-09T09:28:00+00:00","index":"","fulltext":""},{"type":"submitted","content":"Scientific Reports","date":"2026-05-09T09:24:48+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"c7058c64-c291-46b1-b924-1b0d205a6a27","owner":[],"postedDate":"May 13th, 2026","published":true,"recentEditorialEvents":[{"type":"decision","content":"Revision requested","date":"2026-05-18T09:33:00+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-05-14T13:50:02+00:00","index":34,"fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-05-12T02:48:21+00:00","index":32,"fulltext":""},{"type":"reviewerAgreed","content":"134274833076557886635537381881750131388","date":"2026-05-12T00:27:44+00:00","index":31,"fulltext":""},{"type":"reviewerAgreed","content":"115548658825137598935944456706803422655","date":"2026-05-12T00:01:17+00:00","index":30,"fulltext":""},{"type":"reviewersInvited","content":"11","date":"2026-05-11T23:59:02+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-05-11T23:55:52+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2026-05-11T14:42:53+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-05-09T09:28:00+00:00","index":"","fulltext":""},{"type":"submitted","content":"Scientific Reports","date":"2026-05-09T09:24:48+00:00","index":"","fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"in-revision","subjectAreas":[{"id":67969538,"name":"Physical sciences/Engineering"},{"id":67969539,"name":"Physical sciences/Mathematics and computing"},{"id":67969540,"name":"Earth and environmental sciences/Ocean sciences"},{"id":67969541,"name":"Physical sciences/Optics and photonics"}],"tags":[],"updatedAt":"2026-05-18T09:40:35+00:00","versionOfRecord":[],"versionCreatedAt":"2026-05-13 17:51:44","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9611886","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9611886","identity":"rs-9611886","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}

Text is read by the "Ask this paper" AI Q&A widget below. Extraction quality varies by source — PMC NXML preserves structure cleanly, OA-HTML may include some navigation residue, and OA-PDF can have broken hyphenation. The publisher copy (via DOI) is the canonical version.

My notes (saved in your browser only)

Ask this paper AI returns verbatim quotes from the full text · source: preprint-html

Answers must be backed by verbatim quotes from this paper's full text. Hallucinated quotes are dropped automatically; if no verbatim passage answers the question, we say so. How this works

Citation neighborhood (no data yet)

We don't have any in-corpus citations linked to this paper yet. This is a recent paper (2026) — citers typically take a year or two to land, and the OpenAlex reference graph may still be filling in.

Source provenance

europepmc
last seen: 2026-05-20T01:45:00.602351+00:00
unpaywall
last seen: 2026-05-24T02:00:01.246996+00:00
License: CC-BY-4.0