High level tools to make the power of Geometry Nodes accessible to any user familiar with modifiers. The focus here is (opposite to builtin Geometry Nodes) to combine lots of options and functionality into one convenient package, that can be extended by editing the nodes, or integrating it into a node-setup, but is focused on being used without node editing.
Design: I started by generating cohesive concept images in Midjourney, with sleek white interiors with yellow accents to define the overall vibe.
Generate: Using World Labs, I transformed those images into fully explorable and persistent 3D environments in minutes.
Assemble: I cropped out doorways inside the Gaussian splats, then aligned and stitched multiple rooms together using PlayCanvas Supersplat, creating a connected spaceship layout.
Experience: Just a few hours later, I was walking through a custom interactive game level that started as a simple idea earlier that day.
The AI Toolkit UI is a web interface for the AI Toolkit. It allows you to easily start, stop, and monitor jobs. It also allows you to easily train models with a few clicks. It also allows you to set a token for the UI to prevent unauthorized access so it is mostly safe to run on an exposed server.
Over 600+ production-ready models for image, video, audio, 3D. Fal AI
Serverless / On-demand Compute
You don’t have to set up GPU clusters yourself. It offers serverless GPUs with no cold starts or autoscaler setup. Fal AI
Custom / Private Deployments
Support for bringing your own model weights, private endpoints, and secure model serving. Fal AI
High Throughput & Speed
fal claims their inference engine for diffusion models is “up to 10× faster” and built for scale (100M+ daily inference calls) with “99.99% uptime.” Fal AI
Enterprise / Compliance
SOC 2 compliance, single sign-on, analytics, priority support, and tooling aimed at enterprise deployment and procurement. Fal AI
Flexible Pricing
Options include per-output (serverless) or hourly GPU pricing (for more custom compute). Fal AI
Use Cases & Positioning
Useful for rapid prototyping or productionizing generative media features (e.g. image generation, video, voice).
Appeals to teams that don’t want to manage MLOps/infra — it abstracts a lot of the “plumbing.”
Targets both startups and enterprises — they emphasize scale, reliability, and security.
They also showcase that fal is used by recognized companies in AI, design, and media (testimonials on site
– player and number detection with RF-DETR – player tracking with SAM2 – team clustering with SigLIP, UMAP and K-means – number recognition with SmolVLM2
“The Lionsgate catalog is too small to create a model,” a source tells The Wrap. “In fact, the Disney catalog is too small to create a model.” … Another issue is the rights of actors and the model for remuneration if their likeness appears in an AI-generated clip. It is a legal gray area with no clear path.
“Simon Willison created a Datasette browser to explore WebVid-10M, one of the two datasets used to train the video generation model, and quickly learned that all 10.7 million video clips were scraped from Shutterstock, watermarks and all.”
“In addition to the Shutterstock clips, Meta also used 10 million video clips from this 100M video dataset from Microsoft Research Asia. It’s not mentioned on their GitHub, but if you dig into the paper, you learn that every clip came from over 3 million YouTube videos.”
“It’s become standard practice for technology companies working with AI to commercially use datasets and models collected and trained by non-commercial research entities like universities or non-profits.”
“Like with the artists, photographers, and other creators found in the 2.3 billion images that trained Stable Diffusion, I can’t help but wonder how the creators of those 3 million YouTube videos feel about Meta using their work to train their new model.”
