Splattr: Create 3D Gaussian Splats in the cloud, no setup required

Hi all, for the past month I've been working like crazy on splattr, a cloud-based app to create nerfs and gaussian splats easily using powerful GPUs on the cloud. I know there are other great tools out there, but I wanted to build something that gave me the flexibility of a tool like Nerfstudio but with the convenience of a cloud UI. I'm launching it today and would like to get some feedback about it, good or bad. Some of its features are Flexible Workflows: Upload raw media (videos, or ZIPs of photos) for the full pipeline, or upload pre-processed NeRF Studio datasets (transforms.json) to skip the COLMAP stage, which is much cheaper and faster. Various options to choose from: number of downscales, fps, training method, max number of iterations and maching strategy. I will add more in the future. Iterative refinement: You can run a job, view the result, and if it's not perfect, you can go back, add more photos or videos to the same job, and re-process. The app uses the context from the previous run to refine the camera poses and improve the final splat. Public sharing: Share your creation with the world by generating a public link! All splats are private by default. Team collaboration: You can add multiple people to your splat, and manage their permissions. Free Online Viewer: There's also a simple, free .ply viewer under the "Tools" menu for anyone to use, no signup required. As much as I'd love to have a generous free tier, GPU time isn't free, so, as for now, I've set up a credit-based system. The price of every job is calculated upfront based on all your settings (model choice, resolution, iterations, etc.), so you know the exact cost before you commit without surprises. To get things started and to gather as much feedback as possible, I'm giving to the first 20 users 100 free credits (I will increase that in case more people are interested, check the discord server). I hope that this will be enough to experiment, and hopefully create something cool. My only ask is that you share your thoughts, bug reports, or feature ideas. This is a stable version, but I have so many ideas for the future. Before I build anything else, I want to hear from you all. If you are really interested, consider also joining the discord server and share your experience. Here are all the links you might need: Main app: https://splattr.app Discord server: https://discord.gg/5eyUW5YAeT Docs: https://docs.splattr.app Thank you for reading and checking it out

Universal-3 Pro by AssemblyAI — Speech-to-text that finally understands context

Speech-to-text that finally understands context

Promoted

Maker

📌

6mo ago

GPT-4o

Whoa, cloud-based NeRF creation? That's truely awesome! The iterative refinement feature is kinda genius imo – being able to add more photos/videos to a job and re-process is a HUGE time saver, right? No more starting from scratch if you need to tweak something. So cool! How's the performance on larger datasets?

Sellinger AI

Hey,

Congrats on the release, I supported, we’re live today too and your quick look would help.