BenderDrummer

Mostly retweet tech and avoid politics. Previously #3 Sheertex (YC W18). People before products.

[BenderDrummer @BenderDrummer]

@yacineMTB it’d be crazy if you didn’t

[BenderDrummer @BenderDrummer]

@yacineMTB El Segundo

[BenderDrummer @BenderDrummer]

@macjshiggins Oooh that’s is siiick

[BenderDrummer @BenderDrummer]

@___Dario_____ @lrocket Castings are great in compression and not in tension

3D printed parts take forever, too slow for mass production cycles

3D prints have to follow the same paths for every volume of material, can’t produce complex geometry as fast as metal liquid

[BenderDrummer @BenderDrummer]

@Tyler_Plummer Data here is super old, but it explains the decline

https://t.co/09fYIpjmak

[BenderDrummer @BenderDrummer]

onshore ferrous casting is a huge issue by the way..

[Tyler Plummer @Tyler_Plummer]

Need Die Cast Clamps ASAP. Don’t have print right now. Pic is best I have. Cast out of Iron. DM me with quote and lead time please.

[BenderDrummer @BenderDrummer]

@Tyler_Plummer I’ve been reading about the closure of these companies, and it almost puts the continent at risk

[BenderDrummer @BenderDrummer]

@JoshuaSteinman @aphysicist - Ferrous Casting (upstream of machine tools)

[BenderDrummer @BenderDrummer]

@boxcardavid the prop toque would twist that barrel of magnets I bet

[Chris Sweeney @the_shweenz]

2) Given the stereo overlap in Gen 2 we can leverage a robust stereo model like FoundationStereo to obtain dense correspondences from stereo-rectified images. With some optimizations we were able to get FoundationStereo to run in real-time -- this enables live depth streaming!

[Daniel DeTone @ddetone]

Tired of 1B+ parameter VLMs that require massive GPUs and take 1s+ to process an image? Do you miss live demos?
Come to the CVPR Meta booth Friday + Saturday 10am-12pm, I'll be giving a new live demo running on a LAPTOP of our egocentric 3DBB demo of our latest work, Boxer

[Eliot Pence @EliotPence]

EPIC

[TBPN @tbpn]

BREAKING: @bchesky is launching a new AI lab to build models, focused on UI and design

[BenderDrummer @BenderDrummer]

@TheDarkGoldMan Looks great! 🚀

[em m0shouris @emm0sh]

the solidworks error watching you add a third guide rail to a loft

[BenderDrummer @BenderDrummer]

sometimes you just wanna lay out the nesting manually in vector layers first, drill holes 🕳️ then go back and manually outline the vectors for tool comp, if.. the hole drilling the change to cutting has less rapids.. algo sometimes is never optimal as just step and repeat CAM path arrays.. but it’s faster to generate if you don’t care about cut time

[BenderDrummer @BenderDrummer]

If your uncle calls and needs you to send bitcoin, don’t do it

[Aoden Teo @AodenTeoMT]

Today, we’re excited to introduce Miso One, the most emotive voice model in the world.

Miso One is an 8-billion-parameter text-to-speech model for highly expressive speech generation. It emotes like a human and responds faster than a human, with just 110 milliseconds of latency.

We’ve open-sourced the model weights, with API access coming soon.

Hear how Miso One sounds in the thread below.

[Joey @aijoey]

real world run with @NVIDIAAI Cosmos 3

Ran every frame of a 15s highway clip (450 frames @ 30fps) through LocateAnything-3B on DGX Spark:

- 450 frames × 14,105 boxes — every frame, real detections
- 5.2s/frame avg (vs 12.7 BPS on H100) — GB10 holds its own
- 7.8 GB VRAM — fits with room to spare on 128GB unified
- IoU tracking + lane assignment — real trajectories, not synthetic

cc: @PavloMolchanov