08 / Feb 17 – 23, 2025 · 8 RELEASES

AI News Week 8, 2025

Week 8 2025 AI releases.

Vision & Image (1)

  • Tight Inversion

    Tel Aviv University & Snap Research · Image Editing

    A novel image inversion method for diffusion models that conditions on the input image itself during both inversion and reconstruction, improving the tradeoff between reconstruction fidelity and editability for real image editing tasks.

Video & Animation (2)

  • Diffusion Forcing Transformer

    kiwhansong · Video Understanding

    A transformer-based model that applies diffusion forcing techniques for improved sequence generation and video prediction tasks. Combines diffusion models with transformer architecture for enhanced temporal consistency.

  • TrajectoryCrafter

    Yu, Hu, Xing, Shan (ICCV 2025) · Video Editing

    Diffusion model-based system for redirecting camera trajectories in monocular videos, enabling creative cinematographic effects like bullet time, frozen camera, and dolly zoom without manual specification.

3D & Spatial (1)

  • MIDI: Multi-Instance Diffusion for Single Image to 3D Scene Generation

    Beihang University, VAST, Tsinghua University, University of Hong Kong · 3D Generation

    MIDI extends pre-trained image-to-3D models into multi-instance diffusion models that simultaneously generate multiple 3D objects with accurate spatial relationships from a single image using novel multi-instance attention mechanisms. The method processes complete scenes in ~40 seconds by conditioning on segmented object images and global scene context.

Agents & Automation (1)

  • OpenAI Agents SDK

    OpenAI · Agent Framework

    Production-ready Python SDK for building agentic AI applications with core primitives of Agents, Handoffs (agent-to-agent delegation), and Guardrails for validation. Includes built-in agent loop, tool orchestration, persistent sessions, tracing/visualization, and realtime voice agent capabilities.

Developer Tools (3)

  • WanTraining

    spacepxl · AI Infrastructure

    Training framework for vision models with focus on efficient distributed training and optimization techniques. Includes utilities for data loading, model architecture implementations, and training pipelines.

  • Math-To-Manim

    HarleyCoops · AI Infrastructure

    A tool that converts mathematical expressions and equations into animated visualizations using the Manim animation engine. Enables programmatic generation of mathematical animation content for education and presentation purposes.

  • ResumeConvertor Latex

    Abdullah Shafiq · Prompt Engineering

    A tool that converts resumes to LaTeX format, likely using parsing and template generation to automate resume formatting and styling.

Published by Quanta AI Labs, an AI agency in India. Browse every week in AI News Weekly.