FLOSS real-time visuals for media arts with ossia score (lgm2026)

This {presentation/workshop} will introduce ossia score to the audience. It is a free and open-source desktop software, running on Linux (including Raspberry Pi), FreeBSD, macOS and Windows, and has been used in artworks showcased in high-profile venues such as Ars Electronica, Nemo Biennale, Venice Architecture Biennale and others. Originally centered on pure show control and music, recent versions have focused more and more on providing a state-of-the-art real-time pipeline for live visual arts. Combining timelines and patch-based metaphors, it now allows visual artists, VJs, installation and performance artists, to freely combine shaders, geometry processing, audio-reactive behaviours and generative AI in a performant, GPU-only rendering pipeline, while being compatible with industry standards in show control, media processing and A/V formats: OSC, DMX, MIDI, Spout, Syphon, NDI and many others. It also enables more advanced users to live-code any kind of shaders on the fly. This presentation will give a quick overview of the general features and concept of the software, and will then delve more in depth in the graphics pipeline, based on Qt RHI, which supports OpenGL, Vulkan, Metal and Direct3D. We will in particular discuss the real-time compute-shader-based geometry pipeline which enables handling tens of millions of particles at interactive frame rates. We will also present the plug-in system, which allows to create from the same code, audio, data, texture and geometry plug-ins for ossia, as well as for alternative software such as Max/MSP and TouchDesigner. In particular, we will focus on AI-based plug-ins which leverage onnxruntime and TensorRT for real-time inference of AI models such as GANs or diffusion models, without requiring complex Python-based runtime dependencies. Licensed to the public under https://creativecommons.org/licenses/by/4.0/ about this event: https://pretalx.c3voc.de/lgm-2026/talk/CSR9QJ/