WonderHuman: 3D avatars from single-view video

This link (which is not underlined and so easy to miss, <i>boooo</i>) includes videos and more: <a href="https:&#x2F;&#x2F;wyiguanw.github.io&#x2F;WonderHuman&#x2F;" rel="nofollow">https:&#x2F;&#x2F;wyiguanw.github.io&#x2F;WonderHuman&#x2F;</a>

Not sure if there are similar techniques applied, but this definitely reminds of of the AI &quot;microwave&quot; filter that&#x27;s been very popular on TikTok for the past few weeks. It creates equal numbers of impressive and hilarious 360 models from single images. One example: <a href="https:&#x2F;&#x2F;www.tiktok.com&#x2F;@beccasthesia&#x2F;video&#x2F;7471455213115755806" rel="nofollow">https:&#x2F;&#x2F;www.tiktok.com&#x2F;@beccasthesia&#x2F;video&#x2F;74714552131157558...</a>

Reminds me of that scene [1] from Enemy of the State where they &quot;Rotate us 75 degrees around the vertical&quot; in one of those &quot;enhance, enhance&quot; tropes.<p>[1] <a href="https:&#x2F;&#x2F;youtu.be&#x2F;3EwZQddc3kY?t=7" rel="nofollow">https:&#x2F;&#x2F;youtu.be&#x2F;3EwZQddc3kY?t=7</a>

Based on the video this looks great in the 200x200px previews typical for a paper but has substantial visual artifacts that become obvious at higher resolution.<p>Maybe there are straightforward ways to improve the results by spending more compute time. But as presented it&#x27;s more &quot;turn one picture into a convincing character in your Sims game&quot;, not &quot;turn one picture into an animated character in your next TikTok video&quot;

Looking forward to the next paper on this. Hopefully debuting the FaceBack app!