🏗️ Building on HF

Suvaditya Mukherjee PRO

suvadityamuk

43 27 23

https://www.suvadityamuk.com

AI & ML interests

Computer Vision, Text-to-Image/3D Generation, Multimodal Models

Recent Activity

updated a Space 3 days ago

suvadityamuk/lights-camera-causality

published a Space 3 days ago

suvadityamuk/lights-camera-causality

liked a Space 4 days ago

ariG23498/flow-matching

View all activity

Organizations

suvadityamuk 's Spaces 34

Lights, Camera, Causality

🌍

Lights, Camera, Causality -- what even is a world model?

World Lab

🌌

Explore action outcomes in a simple 2D world

Matter Sampled — Geometry Lab

🫧

Explore a torus as points, mesh, or splats

Motion from Noise

🎞

Explore video diffusion by revealing images from noise

World Model Arena

🌐

Run a CPU‑only web service with FastAPI

A Hitchhiker's Guide to the 3D Ecosystem

💡

An interactive guide to 3D representations.

Agentic 3D Worlds Gallery

🌐

Browse and explore the worlds in the collection

Atelier Blender Kitchen

🪴

Blender kitchen with 3.88M triangles and lighting controls

One Prompt, Three Realities

🤗

An interactive tour of video, 3D, and world generation

Display an interactive tracking dashboard

Qwen3.8 Spatial

🧭

Spatial reasoning with Qwen3.8-27B

Trellis2 Geoflow Lora Trackio

🎯

Show visual summary of tracked I/O activity

Blender Agent

🧊

Chat with an open LLM that builds 3D scenes in Blender

Hugging Trace Viewer

🔍

Browse and share Perfetto traces from Hugging Face

TRELLIS Text-to-3D + LLM Prompt Upscaler

🧊

Text to 3D mesh - LLM prompt upscaler + TRELLIS

Inspiration Seeds

🌱

Fuse two images into diverse compositions, no prompt

MAOAM Sa2VA Object & Material Selection

🎯

Object & material selection (Sa2VA, Qwen2.5-VL + SAM2)

MAOAM GLaMM Object & Material Selection

🎯

Click/text object & material selection (GLaMM + SAM ViT-H).

ComboStoc-XL/2 ImageNet 256

🎲

Class-conditional ImageNet 256 (ComboStoc-XL/2)

Terrain Diffusion Infinite Panorama

🏔

Infinite-width panorama generation with SD1.5

Colorful Noise

🎨

Training-free color-conditioned SDXL via FFT noise mixing