{"id":32,"date":"2023-10-08T10:23:01","date_gmt":"2023-10-08T10:23:01","guid":{"rendered":"http:\/\/e-muse.it\/blog\/?p=32"},"modified":"2023-10-11T22:47:51","modified_gmt":"2023-10-11T22:47:51","slug":"a-3d-reconstruction-comparison","status":"publish","type":"post","link":"https:\/\/e-muse.it\/blog\/2023\/10\/08\/a-3d-reconstruction-comparison\/","title":{"rendered":"A 3D reconstruction comparison"},"content":{"rendered":"\n<p class=\"wp-block-paragraph\">In the past few weeks a new reconstruction technique has been taking the community by storm, called <a href=\"https:\/\/repo-sam.inria.fr\/fungraph\/3d-gaussian-splatting\/\" data-type=\"link\" data-id=\"https:\/\/repo-sam.inria.fr\/fungraph\/3d-gaussian-splatting\/\">Gaussian Splatting<\/a>. It is sort of an evolution on NeRFs, and mathematically it&#8217;s not dissimilar from the kind of reconstructions we do for spatial audio.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Luckily a couple of companies have already implemented Gaussian Splatting pipelines, and here we are comparing two of them with photogrammetry.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Here is Luma AI:<\/p>\n\n\n\n<iframe src=\"https:\/\/lumalabs.ai\/embed\/003a6b01-5f92-4ed8-abdd-b64aa48ed0ae?mode=sparkles&#038;background=%23ffffff&#038;color=%23000000&#038;showTitle=true&#038;loadBg=true&#038;logoPosition=bottom-left&#038;infoPosition=bottom-right&#038;cinematicVideo=undefined&#038;showMenu=false\"  frameborder=\"0\" title=\"luma embed\" style=\"display: block; margin-left: auto; margin-right: auto; border: none; width: 100%; min-height: 300px\"><\/iframe>\n\n\n\n<p class=\"wp-block-paragraph\">Polycam<\/p>\n\n\n\n<iframe src=\"https:\/\/poly.cam\/capture\/095df594-8b07-49bc-a8e6-8939784db7db\" title=\"polycam capture viewer\" style=\"display:block;margin-left:auto;margin-right:auto;height:100%;width:100%;max-height:720px;min-height:480px;min-width:380px\" frameborder=\"0\"><\/iframe>\n\n\n\n<p class=\"wp-block-paragraph\">And finally a photogrammetry computed with PhotoCatch (which uses Apple&#8217;s ObjectCapture API, like Polycam&#8217;s photogrammetry), converted to GLB in Blender to stay under Sketchfab&#8217;s size limit. In practice the only modification is that the textures are in JPEG instead of PNG.<\/p>\n\n\n\n<div class=\"sketchfab-embed-wrapper\"> <iframe style=\"display: block; margin-left:auto; margin-right: auto; width: 100%; min-height: 300px;\" title=\"Malignano 2023-10-07\" frameborder=\"0\" allowfullscreen mozallowfullscreen=\"true\" webkitallowfullscreen=\"true\" allow=\"autoplay; fullscreen; xr-spatial-tracking\" xr-spatial-tracking execution-while-out-of-viewport execution-while-not-rendered web-share src=\"https:\/\/sketchfab.com\/models\/3b5cb1280053495a9adf2b0084c059fc\/embed\"> <\/iframe> <p style=\"font-size: 13px; font-weight: normal; margin: 5px; color: #4A4A4A;\"> <a href=\"https:\/\/sketchfab.com\/3d-models\/malignano-2023-10-07-3b5cb1280053495a9adf2b0084c059fc?utm_medium=embed&#038;utm_campaign=share-popup&#038;utm_content=3b5cb1280053495a9adf2b0084c059fc\" target=\"_blank\" rel=\"nofollow noopener\" style=\"font-weight: bold; color: #1CAAD9;\"> Malignano 2023-10-07 <\/a> by <a href=\"https:\/\/sketchfab.com\/xorgol?utm_medium=embed&#038;utm_campaign=share-popup&#038;utm_content=3b5cb1280053495a9adf2b0084c059fc\" target=\"_blank\" rel=\"nofollow noopener\" style=\"font-weight: bold; color: #1CAAD9;\"> xorgol <\/a> on <a href=\"https:\/\/sketchfab.com?utm_medium=embed&#038;utm_campaign=share-popup&#038;utm_content=3b5cb1280053495a9adf2b0084c059fc\" target=\"_blank\" rel=\"nofollow noopener\" style=\"font-weight: bold; color: #1CAAD9;\">Sketchfab<\/a><\/p><\/div>\n\n\n\n<p class=\"wp-block-paragraph\">The difference is particularly evident in the textures on the building, and it&#8217;s just staggering on the vegetation.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\" id=\"gaussian-splatting-comparison\">Also there is a small caveat: Luma was given a video, Polycam was given 200 photos, PotoCatch was given around 250 photos. This reflects differences in the platforms themselves.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Update 2023-10-12: There is already an <a href=\"https:\/\/github.com\/quadjr\/aframe-gaussian-splatting\" data-type=\"link\" data-id=\"https:\/\/github.com\/quadjr\/aframe-gaussian-splatting\">A-Frame component for self-hosting Gaussian Splats<\/a>, we&#8217;re going to try it real soon.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>In the past few weeks a new reconstruction technique has been taking the community by storm, called Gaussian Splatting. It is sort of an evolution on NeRFs, and mathematically it&#8217;s not dissimilar from the kind of reconstructions we do for spatial audio. Luckily a couple of companies have already implemented Gaussian Splatting pipelines, and here [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":33,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"activitypub_content_warning":"","activitypub_content_visibility":"","activitypub_max_image_attachments":3,"activitypub_interaction_policy_quote":"anyone","activitypub_status":"","footnotes":""},"categories":[5],"tags":[17,10,16,8,6,11],"class_list":["post-32","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-reconstruction","tag-1caad9","tag-3d","tag-4a4a4a","tag-nerf","tag-photogrammetry","tag-radiant-fields"],"_links":{"self":[{"href":"https:\/\/e-muse.it\/blog\/wp-json\/wp\/v2\/posts\/32","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/e-muse.it\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/e-muse.it\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/e-muse.it\/blog\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/e-muse.it\/blog\/wp-json\/wp\/v2\/comments?post=32"}],"version-history":[{"count":8,"href":"https:\/\/e-muse.it\/blog\/wp-json\/wp\/v2\/posts\/32\/revisions"}],"predecessor-version":[{"id":45,"href":"https:\/\/e-muse.it\/blog\/wp-json\/wp\/v2\/posts\/32\/revisions\/45"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/e-muse.it\/blog\/wp-json\/wp\/v2\/media\/33"}],"wp:attachment":[{"href":"https:\/\/e-muse.it\/blog\/wp-json\/wp\/v2\/media?parent=32"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/e-muse.it\/blog\/wp-json\/wp\/v2\/categories?post=32"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/e-muse.it\/blog\/wp-json\/wp\/v2\/tags?post=32"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}