{"id":71980,"date":"2026-09-21T00:01:25","date_gmt":"2026-09-20T22:01:25","guid":{"rendered":"https:\/\/www.3dnatives.com\/en\/?p=71980"},"modified":"2026-09-18T17:50:07","modified_gmt":"2026-09-18T15:50:07","slug":"direct-ink-writing-llnl-21092026","status":"publish","type":"post","link":"https:\/\/www.3dnatives.com\/en\/direct-ink-writing-llnl-21092026\/","title":{"rendered":"Lawrence Livermore&#8217;s AI Vision System Catches Direct Ink Writing Defects Mid-Print"},"content":{"rendered":"<p dir=\"ltr\" style=\"text-align: justify;\"><a href=\"https:\/\/www.3dnatives.com\/en\/direct-ink-writing-guide-23122025\/\">Direct ink writing<\/a> is one of those 3D printing technologies that has a process that can vary significantly. It&#8217;s not very commercialized, and mostly reserved for research labs.<\/p>\n<p dir=\"ltr\" style=\"text-align: justify;\">Now, a team of scientists and engineers at <strong>Lawrence Livermore National Laboratory (LLNL)<\/strong> in California have developed a camera-based inspection system that can monitor complex 3D-printed structures using AI and machine learning to measure small variations and potentially identify problems before a part leaves the printer. Described in <em>npj Advanced Manufacturing<\/em>, the system pairs printer-mounted cameras with machine learning and computer vision to turn thousands of in-print images into measurements and spatial maps of the deposited material.<\/p>\n<h2 dir=\"ltr\" style=\"text-align: justify;\">Catching Problems Earlier<\/h2>\n<p dir=\"ltr\" style=\"text-align: justify;\">Direct ink writing can produce flexible cushions and pads that depend on strands only a fraction of a millimeter thick. Gaps, breaks or diameter changes can hurt performance.<\/p><div class=\"dnati-inside-article-leaderboard dnati-entity-placement\" style=\"text-align: center;\" id=\"dnati-1498233752\"><div id=\"dnati-3943223622\"><a data-no-instant=\"1\" href=\"https:\/\/us06web.zoom.us\/webinar\/register\/4117888011843\/WN_-wcvopznTX-DUxxXmQR72w\" rel=\"noopener\" class=\"a2t-link\" target=\"_blank\" aria-label=\"leaderboard\"><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.3dnatives.com\/en\/wp-content\/uploads\/sites\/2\/2026\/09\/leaderboard.gif\" alt=\"\"  width=\"850\" height=\"150\"   \/><\/a><\/div><\/div>\n<p dir=\"ltr\" style=\"text-align: justify;\">Traditionally, inspection happens after printing, via X-ray imaging or mechanical testing. It&#8217;s costly, slow and only reveals failures after the fact.<\/p>\n<p dir=\"ltr\" style=\"text-align: justify;\">The LLNL system adds an earlier check. A camera images each layer as it&#8217;s deposited, and software measures the newest strands, including filament diameter.<\/p>\n<p dir=\"ltr\" style=\"text-align: justify;\"><em>&#8220;It&#8217;s a first-pass check,&#8221;<\/em> said project technical lead Brian Weston. <em>&#8220;It allows us to see things before we do some very expensive tests and to fail parts earlier if we already know they have broken strands or other problems.&#8221;<\/em><\/p>\n<div id=\"attachment_71992\" style=\"width: 710px\" class=\"wp-caption aligncenter\"><img loading=\"lazy\" decoding=\"async\" aria-describedby=\"caption-attachment-71992\" class=\"size-full wp-image-71992\" src=\"https:\/\/www.3dnatives.com\/en\/wp-content\/uploads\/sites\/2\/2026\/09\/1-2.jpg\" alt=\"\" width=\"700\" height=\"400\" srcset=\"https:\/\/www.3dnatives.com\/en\/wp-content\/uploads\/sites\/2\/2026\/09\/1-2.jpg 700w, https:\/\/www.3dnatives.com\/en\/wp-content\/uploads\/sites\/2\/2026\/09\/1-2-600x343.jpg 600w, https:\/\/www.3dnatives.com\/en\/wp-content\/uploads\/sites\/2\/2026\/09\/1-2-160x91.jpg 160w\" sizes=\"auto, (max-width: 700px) 100vw, 700px\" \/><p id=\"caption-attachment-71992\" class=\"wp-caption-text\">LLNL researchers (from left) data scientist Michael Zelinski, engineer Hamed Ziad Ammar and principal investigator Brian Giera beside the DIW 3D printer. (Photo: Blaise Douros\/LLNL)<\/p><\/div>\n<h2 dir=\"ltr\" style=\"text-align: justify;\">How It Performs<\/h2>\n<ul dir=\"ltr\" style=\"text-align: justify;\">\n<li>Trained on nearly 15,000 human-annotated images across several lattice geometries<\/li>\n<li>Across 55 parts, measurements typically landed within a few micrometers of human ones<\/li>\n<li>Analysis takes milliseconds versus 20 minutes to an hour by hand, roughly 100,000 times faster on average<\/li>\n<li>Reusable calibration and dataset work means far less training data for new cameras and parts<\/li>\n<\/ul>\n<h2 dir=\"ltr\" style=\"text-align: justify;\">Seeing the Whole Print<\/h2>\n<p dir=\"ltr\" style=\"text-align: justify;\">On a cushion roughly 25 by 25 centimeters, researchers combined about 2,500 images from one layer into a map of the interior.<\/p>\n<p dir=\"ltr\" style=\"text-align: justify;\">The map showed filament diameters drifting across the print, revealing a slight platform tilt that an overall average could have hidden.<\/p>\n<p dir=\"ltr\" style=\"text-align: justify;\">Because the camera sits on the printer, it can also inspect parts too large for practical high-resolution CT.<\/p>\n<h2 dir=\"ltr\" style=\"text-align: justify;\">What&#8217;s Next<\/h2>\n<p dir=\"ltr\" style=\"text-align: justify;\">Principal investigator Brian Giera said the approach can extend to other additive, subtractive and experimental systems. Near term, it could scrap heavily defective parts mid-print and flag when X-ray CT is warranted.<\/p>\n<p dir=\"ltr\" style=\"text-align: justify;\">The capability is headed to the Kansas City National Security Complex for evaluation. Longer term, the data could feed digital twins linking a part&#8217;s structure to predicted performance.<\/p>\n<p dir=\"ltr\" style=\"text-align: justify;\"><em>&#8220;When the printer can inspect its own work, we can start thinking about the system making its own accept\/reject calls,&#8221;<\/em> Weston said. <em>&#8220;If it sees a defect, maybe it can assess whether that makes the part nonconforming.&#8221;\u00a0<\/em>If you want to learn more about the research, find the study <a href=\"https:\/\/www.nature.com\/articles\/s44334-026-00107-7\" target=\"_blank\" rel=\"noopener\">here<\/a>.<\/p>\n<p dir=\"ltr\" style=\"text-align: justify;\">What do you think of this research? Let us know in a comment below or on our\u00a0<a href=\"https:\/\/www.linkedin.com\/company\/4987104\/\">LinkedIn<\/a>\u00a0or\u00a0<a href=\"https:\/\/www.facebook.com\/3Dnatives\/\">Facebook<\/a>\u00a0pages! Plus, don\u2019t forget to sign up for our free weekly\u00a0<a href=\"https:\/\/www.3dnatives.com\/en\/3d-printing-newsletter\/\">Newsletter<\/a>\u00a0to get the latest 3D printing news straight to your inbox. You can also find all our videos on our\u00a0<a href=\"https:\/\/www.youtube.com\/channel\/UCMWrNpdLOXa7BffRKXZoaZw\">YouTube<\/a>\u00a0channel.<\/p>\n<p style=\"text-align: justify;\"><em>*Cover photo credit: Garry McLeod\/LLNL<\/em><\/p>\n<div class=\"dnati-after-content dnati-entity-placement\" id=\"dnati-3146416521\"><div id=\"dnati-981139696\"><a data-no-instant=\"1\" href=\"https:\/\/www.npe.org\/about-npe?utm_campaign=41990190-NPE2027%20Attendee%20Marketing&#038;utm_source=3D%20Natives&#038;utm_medium=3D%20Newsletter%20Banner&#038;utm_term=Reg%20Opens%20Sept%2014\" rel=\"noopener\" class=\"a2t-link\" target=\"_blank\" aria-label=\"NPE2027_850x150\"><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.3dnatives.com\/en\/wp-content\/uploads\/sites\/2\/2026\/09\/NPE2027_850x150.png\" alt=\"\"  srcset=\"https:\/\/www.3dnatives.com\/en\/wp-content\/uploads\/sites\/2\/2026\/09\/NPE2027_850x150.png 850w, https:\/\/www.3dnatives.com\/en\/wp-content\/uploads\/sites\/2\/2026\/09\/NPE2027_850x150-600x106.png 600w, https:\/\/www.3dnatives.com\/en\/wp-content\/uploads\/sites\/2\/2026\/09\/NPE2027_850x150-768x136.png 768w, https:\/\/www.3dnatives.com\/en\/wp-content\/uploads\/sites\/2\/2026\/09\/NPE2027_850x150-160x28.png 160w\" sizes=\"(max-width: 850px) 100vw, 850px\" width=\"850\" height=\"150\"   \/><\/a><\/div><\/div>","protected":false},"excerpt":{"rendered":"<p>Direct ink writing is one of those 3D printing technologies that has a process that can vary significantly. It&#8217;s not very commercialized, and mostly reserved for research labs. Now, a team of scientists and engineers at Lawrence Livermore National Laboratory&hellip;<\/p>\n","protected":false},"author":6114,"featured_media":71991,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"content-type":"","footnotes":""},"categories":[28,1,10],"tags":[],"class_list":["post-71980","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-3d-technologies","category-news","category-research"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.3dnatives.com\/en\/wp-json\/wp\/v2\/posts\/71980","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.3dnatives.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.3dnatives.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.3dnatives.com\/en\/wp-json\/wp\/v2\/users\/6114"}],"replies":[{"embeddable":true,"href":"https:\/\/www.3dnatives.com\/en\/wp-json\/wp\/v2\/comments?post=71980"}],"version-history":[{"count":2,"href":"https:\/\/www.3dnatives.com\/en\/wp-json\/wp\/v2\/posts\/71980\/revisions"}],"predecessor-version":[{"id":71996,"href":"https:\/\/www.3dnatives.com\/en\/wp-json\/wp\/v2\/posts\/71980\/revisions\/71996"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.3dnatives.com\/en\/wp-json\/wp\/v2\/media\/71991"}],"wp:attachment":[{"href":"https:\/\/www.3dnatives.com\/en\/wp-json\/wp\/v2\/media?parent=71980"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.3dnatives.com\/en\/wp-json\/wp\/v2\/categories?post=71980"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.3dnatives.com\/en\/wp-json\/wp\/v2\/tags?post=71980"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}