🚀 Just created something completely new! I Rewrote Apple's Sharp to generate 3D Gaussian Splats from a single 360° equirectangular image made from FLUX 2 (Flex) One equirectangular image → Full spherical 3DGS world! ✨ New web interface - drag & drop your pano, select quality, download PLY ⚡... 2-3 sec on a older laptop Quality tiers: ▫️ Low: 500K splats ▫️ Medium: 2M splats ▫️ High: 4.7M splats ▫️ Ultra: 8.4M splats The pipeline: 360° depth estimation from equirectangular input Spherical projection maps pixels → 3D points Each point becomes a Gaussian with depth-scaled radius - Export to standard PLY for any 3DGS viewer (Unity) No multi-view capture. No COLMAP. No NeRF training. Just upload a pano, pick quality, get your splats. #3DGS #GaussianSplatting #ComputerVision #AI #madewithunity #Splats Sharp Monocular View Synthesis in Less Than a Secondshow more

Generate SHARP splat -> shift perspective by (x,y,z) offset -> render new equirectangular image from result -> infill missing detail with FLUX -> generate new SHARP splat -> merge into existing scene with known (x,y,z) offset -> rinse, repeat

Any chance of a technical write up or blog post? I’ve been doing research into this very approach. But not for VR, more for a home version of an ILM volume type setup, think projectors.

yes will do:-)

How is it different from extruding the depth map? Why do you need gsplats in this case?

Hi I dont use gsplats have a nice weekend :-)

Damn, Daniel

Thanks Han! , much appreciated have a nice weekend 😃

Great stuff Daniel! Anywhere we can test this?

Thanks Martin😍right now it just runs locally on my laptop but will find a provider and upload it soon it runs really smooth and fast but need to tweek it bit more have a nice sunday I thanks for sharing my work!

@MartinNebelong cant wait!! good work!

@MartinNebelong Thanks Fedox🤓

do you feed the whole image to sharp or chunks of it ?

Hello Leo, Yes the whole image :-)

Wow amazing. It’s still limited by Apples license I guess?

Amazing, good job!

can't wait for you to host this

How can I try it?

will find a provider to upload soon

Incredible work! Can't wait to try it

@ReflctWillie Artistic !

Whoa!! So cool

This is awesome!

How did you do it as the model is trained with 1530x1530 size?

Awesome work! 👏

Why not share the code?

nice work. I think you can convert pano photo to cubeface first. ML-Sharp was trained on perspective photo.

what depth model are you using?

Incredible work

how accurate might this be for architecture stuff? when can we test this?

this project also implemented a pano 360 recon via sharp

looking forward the online demo

Does Flux 2 generate the 360 photos?

What was the process and methods you you applied to produce this? Looks great!

looks like shit. congrats

Hey Daniel. Is this out yet? Would love to try this

Awesome! Could you share your fork so that we can also try it out locally?