<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Pann on Medium]]></title>
        <description><![CDATA[Stories by Pann on Medium]]></description>
        <link>https://medium.com/@yukisophia?source=rss-b1d21ee9d31d------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*UfBar-UL8eV5Kxeg83n_wA.png</url>
            <title>Stories by Pann on Medium</title>
            <link>https://medium.com/@yukisophia?source=rss-b1d21ee9d31d------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 04:30:55 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@yukisophia/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[What Satellites Saw Before and After the 2026 Kumamoto Earthquake]]></title>
            <link>https://yukisophia.medium.com/what-satellites-saw-before-and-after-the-2026-kumamoto-earthquake-2558f7e31d94?source=rss-b1d21ee9d31d------2</link>
            <guid isPermaLink="false">https://medium.com/p/2558f7e31d94</guid>
            <category><![CDATA[remote-sensing]]></category>
            <category><![CDATA[change-detection]]></category>
            <dc:creator><![CDATA[Pann]]></dc:creator>
            <pubDate>Thu, 13 Aug 2026 10:16:34 GMT</pubDate>
            <atom:updated>2026-08-14T05:19:58.326Z</atom:updated>
            <content:encoded><![CDATA[<p><em>A quick, automated damage and change map from two satellite images</em></p><p>When an earthquake hits, the first question is always the same: how bad, and where? Ground surveys take time.<br>A visual assessment of the M7.1 Kumamoto earthquake (July 28, 2026) was conducted using AxelGlobe GRUS satellite imagery acquired on May 18, 2026 (pre-event) and August 5, 2026 (post-event).<br>A big thanks to <a href="https://www.axelglobe.com/ja/the-kumamoto-earthquake-in-2026"><strong>AxelGlobe</strong></a> for making this pre/post imagery data openly available.</p><h4>Approach</h4><p>Change Vector Analysis (CVA) magnitude was calculated across multi-spectral bands (5 meter) to map damage, applying radiometric normalization to adjust for 2.5 months of seasonal drift in sun angle and atmosphere, alongside sub-pixel correlation to account for seismic ground displacement.<br>A location flagged in both maps; spectral change and physical displacement is higher confidence change than either alone.<br><strong>Note</strong>: the 2.5-month gap between timestamps means seasonal growth and unrelated construction can also show up as “change.”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*fLZ36lEkIf6ANPrGbQof1w.png" /><figcaption><strong>Change Analysis</strong></figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*HdAWxzzn56haLR9CLM9Ieg.png" /><figcaption><strong>Displacement Vectors</strong></figcaption></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2558f7e31d94" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[From 2D Images to 3D Reconstruction: My First Experiment with VGGT]]></title>
            <link>https://yukisophia.medium.com/from-2d-images-to-3d-reconstruction-my-first-experiment-with-vggt-f550b23582d3?source=rss-b1d21ee9d31d------2</link>
            <guid isPermaLink="false">https://medium.com/p/f550b23582d3</guid>
            <category><![CDATA[vggt]]></category>
            <category><![CDATA[3d-to-2d]]></category>
            <category><![CDATA[3d-reconstruction]]></category>
            <dc:creator><![CDATA[Pann]]></dc:creator>
            <pubDate>Fri, 07 Aug 2026 05:59:54 GMT</pubDate>
            <atom:updated>2026-08-13T10:27:00.975Z</atom:updated>
            <content:encoded><![CDATA[<p>Recently, I experimented with <strong>VGGT (Visual Geometry Grounded Transformer)</strong> to see how well it can reconstruct a 3D scene from 2D images. For this experiment, I tested two different image sources:</p><ul><li>Aerial drone images</li><li>Android smartphone images</li></ul><p>Both datasets consisted of multiple overlapping images captured from different viewpoints. The objective was to evaluate how effectively VGGT could recover the scene geometry from these images.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*YPqJQSDwaQjzLjs-Sura7Q.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/533/1*ohQa-4q4SjhplFbcEtJqaw.png" /><figcaption><strong>My Adorable Cat That Became 3D 🐾🐾🐾</strong></figcaption></figure><h4><strong>What I Observed</strong></h4><p>Some interesting observations from these experiments:</p><ul><li>VGGT handled both aerial and smartphone images successfully.</li><li>Good image overlap is essential for accurate reconstruction.</li><li>Image quality and camera viewpoints have a significant impact on the final 3D model.</li></ul><h4>Conclusion</h4><p>This experiment demonstrated that VGGT is a flexible framework for generating 3D reconstructions from ordinary 2D images. Whether the images come from a drone or normal camera, the model can recover meaningful scene geometry from large aerial landscapes to close-up building structures when provided with enough overlapping views.</p><p>This was an enjoyable hands-on project, and I look forward to exploring the next version, <strong>VGGT-Ω</strong>! <br>Thank you for reading!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f550b23582d3" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>