← All videos

Stable Diffusion AI VFX & No‑Code Game Engine: SIGGRAPH Vancouver Report

· 7:27

SIGGRAPH Vancouver mini report: generative VFX spoof and a no‑code game engine that builds worlds in one click. Which demo surprised you more—the Stable…

Notes & highlights

Published:  ·  7:27

SIGGRAPH Vancouver mini report: generative VFX spoof and a no‑code game engine that builds worlds in one click. Which demo surprised you more—the Stable Diffusion/Nuke fake TV spot or the AI‑assisted procedural worlds?
------------

SIGGRAPH week in Vancouver was electric. In this lightning share from the Vancouver AI community meetup, we spotlight two rapid‑fire demos: a generative VFX spoof built with Stable Diffusion + Nuke and a sneak peek of a no‑code game engine that spins up procedural worlds, physics, and one‑click publishing.

The last three weeks have been absolutely wild in AI film. Here’s the quick-hit roundup:

- WAN 2.2 dropped with day-one ComfyUI support. 1080p, true 24p output, dual-adversarial modeling for big-picture flow and tiny details in tandem. Local models are officially nipping at the heels of the big boys.

- Runway Aleph is sneaking out of partner-only land — conversational editing of uploaded video. Angle changes, style shifts, mood tweaks… just by telling it what you want.

- Luma Labs Modify hit the scene with instruction-based video manipulation, giving precise post-control inside their ecosystem.

- Nano-Banana (Google’s new engine under Gemini 2.5) slid into ComfyUI. Multi-reference workflows are looking crazy consistent. This could be the new standard for character and scene coherence.

- Ludus AI is experimenting with vibe-coded Unreal Engine blueprints. Big potential for virtual production and 3D workflows.

- ComfyUI MCP servers went live, letting you spin up and stress-test custom local pipelines.

- TouchDesigner LOPs now tie LLMs straight into realtime visual, audio, MIDI, and DMX wizardry.

SIGGRAPH 2025 Vancouver highlights:

- Lightcraft compositing + vibe-coded 3D sets, straight off an iPhone.
- Bria AI: ethically trained API toolset.
- GripTape Nodes: modular workflows running at AWS scale.
- Creadev: rapid Unreal mockups for XR and film pre-viz.
- ActionVFX: stock assets + training courses marrying GenAI and Nuke.
- Autodesk Wonder Studio: now with a free tier, bridging GenAI and VFX.
- And because “one release week” is never enough:
- ElevenLabs launched into music gen.
- MidJourney rolled out HD mode.
- Blackmagic added 360° / Apple Immersive support.
- Tencent upgraded Hunyuan 3D.
- DeepMind teased a world model for environment gen.

⚡ Takeaway: It’s no longer just “what can these models spit out?” The game is shifting fast toward control, consistency, and production-readiness.

👤 Featuring

• John — VFX compositor/director/photographer — AI‑assisted fake TV shopping spot (Stable Diffusion, AUTOMATIC1111, Nuke)

• Ahmed — game dev veteran (ex‑engineering manager, instructor) — “Creative” no‑code engine: procedural terrains, water/cloth/rope physics, destruction, AI‑guided walkthroughs, multi‑platform publish

🌱 Why it matters for BC + AI Grassroots creators in B.C. are testing what’s possible at the boundary of VFX, games, and AI tools. This episode captures that hands‑on energy—and invites you to build with us.

👉 Join the community
• Vancouver AI Meetups → https://luma.com/vancouver-ai
• BC + AI Ecosystem Association newsletter → https://bc-ai.ca/

🔔 Support the work
Like, subscribe, and share with someone exploring ethical, creative AI.

Comment below: what should we deep‑dive next?

CHAPTERS

00:00 Intro — SIGGRAPH Mini Report (Vancouver AI)
00:40 SIGGRAPH vibe + cross-meetup at Northeastern University
01:30 Guest intro: John Mutter (VFX compositor/director)
01:50 Generative VFX demo — Stable Diffusion + Nuke (fake TV spot)
03:20 Takeaways: AI-assisted compositing workflow & tools
03:40 Guest intro: Ahmed — “Creative” no-code game engine
04:20 Procedural open worlds: vegetation, water, weather systems
05:10 Physics stack: cloth, rope, rigid bodies, destruction
05:55 Gen-AI text-to-scene & AI-guided interactive walkthroughs
06:30 One-click publishing, platforms, October release + film cohort invite

Watch on YouTube ↗