Conversation

Notices

1. Embed this notice
   Rich Felker (dalias@hachyderm.io)'s status on Wednesday, 04-Mar-2026 00:12:43 JST Rich Felker

   Wow, a moment in #3dprinting history and totally not who I expected to pull it off!

   • Embed this notice
     Rich Felker (dalias@hachyderm.io)'s status on Wednesday, 04-Mar-2026 00:20:35 JST Rich Felker

     in reply to

     This is of course what folks in the scene call a BSO - a "Benchy-Shaped Object". But getting even this in under a minute is bonkers.

   • Embed this notice
     Rich Felker (dalias@hachyderm.io)'s status on Wednesday, 04-Mar-2026 00:30:58 JST Rich Felker

     in reply to

     A Benchy at competition specs is 192 layers, so this was printed at approximately 280 milliseconds per layer. 🤯

     Just cooling the plastic sufficiently to build the next layer on top of is near-impossible below around 600 ms/layer with even high-end known approaches to cooling. I have no idea how he made it work here.

   • Embed this notice
     Rich Felker (dalias@hachyderm.io)'s status on Wednesday, 04-Mar-2026 00:36:58 JST Rich Felker

     in reply to

     • John Regehr

     @regehr Pretty sure they're open-loop steppers and no gears except in the extruder. Mostly just belts and pulleys and rails. The printed frame (vs a super rigid metal one) absorbs a lot of the forces and trades inaccuracy for being able to get physically-unrealizable nominal accelerations. But the result was still way closer to accurate than most attempts I've seen at half this speed.

   • Embed this notice
     John Regehr (regehr@mastodon.social)'s status on Wednesday, 04-Mar-2026 00:36:59 JST John Regehr

     in reply to

     @dalias I thought the video was sped up at first. I felt bad for those poor gears and servomotors.

   • Embed this notice
     Rich Felker (dalias@hachyderm.io)'s status on Wednesday, 04-Mar-2026 00:49:41 JST Rich Felker

     in reply to

     • Florian 'floe' Echtler

     @floe Keeping temperature isn't that hard. You need about 0.5 W/(mm³/s) of flow to keep up with the heat going into the plastic, and even standard cartridges go up to 120W. I don't know what this custom hotend is running, but the harder part is getting that much heat into the plastic (which has low heatconductivity, and needs solutions that increase the surface area and decrease the depth the heat has to travel inward) not keeping the metal hot.

   • Embed this notice
     Florian 'floe' Echtler (floe@hci.social)'s status on Wednesday, 04-Mar-2026 00:49:42 JST Florian 'floe' Echtler

     in reply to

     @dalias that, and also the opposite question: how does the hotend keep its temperature at these flow speeds? Pretty insane 👌

   • Embed this notice
     John Regehr (regehr@mastodon.social)'s status on Wednesday, 04-Mar-2026 00:54:50 JST John Regehr

     in reply to

     @dalias I don't pay that much attention to 3d printing but I've never seen anything like this. it was moving fast enough that I'd not have stood close by while it was printing!

   • Embed this notice
     Rich Felker (dalias@hachyderm.io)'s status on Wednesday, 04-Mar-2026 00:54:50 JST Rich Felker

     in reply to

     • John Regehr

     @regehr This machine is probably one of the safest to stand near. 😂 He says it's only 500g of moving mass, which sounds about right. The really scary ones are using overbuilt all-steel frames and gantries and motors with enough torque to crush a limb, which tends to be the way folks targeting a high-price industrial market design stuff.

   • Embed this notice
     Rich Felker (dalias@hachyderm.io)'s status on Wednesday, 04-Mar-2026 06:57:03 JST Rich Felker

     in reply to

     Update on this: There have been allegations that he faked it, and while I normally just trust that folks in this community are doing what they say they're doing because there aren't like any prized for faking shit, the evidence (spectrum analysis of the audio vs his other videos, including fan noise and frame vibrations) suggests it was recorded at 30 fps and speed-doubled. Guess we'll see how this plays out. This does match the cooling performance being roughly double what I thought possible with this equipment.

   • Embed this notice
     Rich Felker (dalias@hachyderm.io)'s status on Thursday, 05-Mar-2026 00:09:10 JST Rich Felker

     in reply to

     Wow, the forensic work folks have been doing on this is wild!

     Pendulum period analysis for objects in the background, spectrum analysis of audio, looking up the framerate and resolution modes of the camera it was recorded on, reading the source code of the javascript stopwatch to find its refresh interval, etc.

     All point to video having been recorded at 30 fps and altered to play at 60 fps.

   • Embed this notice
     Rich Felker (dalias@hachyderm.io)'s status on Thursday, 05-Mar-2026 01:43:00 JST Rich Felker

     in reply to

     • Koz Ross

     @koz Annex Discord (speedboat_discussion) and Rolohaun Discord (thread under general) are the places I'm aware of with analysis going on.

   • Embed this notice
     Koz Ross (koz@chaos.social)'s status on Thursday, 05-Mar-2026 01:43:01 JST Koz Ross

     in reply to

     @dalias Do you have a link to some of these analyses? Just so I have something to aim folks at.

   • Embed this notice
     Rich Felker (dalias@hachyderm.io)'s status on Sunday, 08-Mar-2026 12:28:32 JST Rich Felker

     in reply to

     OK, I'm going to go ahead and declare this to have been determined fake. Matt has been "on vacation" since he posted it, but the comments are moderated and anything critical or questioning the validity is not getting thru while plenty of praise is. And at least one piece of evidence, the js clock updating each frame while the source code for the website only has a 32 ms refresh interval, seems irrefutable.

     I don't know what personal circumstances he might be going thru to make "fake a world record, jack up Patreon subscription cost for folks trying to find more about it, and disappear for a while" make sense to him, but this is rather disappointing and souring.

   • Embed this notice
     Rich Felker (dalias@hachyderm.io)'s status on Sunday, 08-Mar-2026 13:07:12 JST Rich Felker

     in reply to

     @koz That sounds sadly plausible.

   • Embed this notice
     Koz Ross (koz@chaos.social)'s status on Sunday, 08-Mar-2026 13:07:14 JST Koz Ross

     in reply to

     @dalias Highly circumstantial and second-hand evidence, but someone I trust who had access to Matt's Patreon posts said that they lately seem to be LLM slop. Given that GPU-powered plagiarism machines have echoed people into _far_ worse things than faking a (niche) world record, I wouldn't be surprised if this is LLM 'psychosis' on display.

   • Embed this notice
     Simon Richter (gyrosgeier@hachyderm.io)'s status on Sunday, 08-Mar-2026 18:01:02 JST Simon Richter

     in reply to

     @dalias liquid nitrogen heat exchanger printing bed?