Innovations Session

Sponsored by

Dome First: A Real Production Workflow — or a Beautiful Theory?

Can a single capture workflow serve both 8K x 8K Fulldome and 1.43 IMAX with Laser without compromise? This session moves beyond speculation into a controlled, real-world experiment. Using a Blackmagic 12K large-format sensor, the team simultaneously captures fisheye footage for an 8K Fulldome Master, rectilinear footage for 1.43 IMAX, and a stitched 12K x 12K 180° fisheye for comparative testing.

For IMAX with Laser a 1.43 extraction from the fisheye and the rectilinear footage is processed, AI-enhanced, and cut side-by-side to see if they integrate seamlessly. Later we will do the 8K-by-8K Fulldome version. We evaluate here at GSCA 2026 in 4K IMAX with Laser and at ASTC September 2026 in Arizona in 8K LED Fulldome.

Horizon placement, extraction quality, and lens strategies are rigorously tested to determine whether “Dome First” is a viable production model — or simply a beautiful theory. Real data, real results, and insights that could reshape lensing, post workflows, and budgeting for next-generation immersive filmmaking.

AI on the Giant Screen: Breakthrough or Half-Baked?

Artificial intelligence promises everything. Fix overexposed highlights. Extend shot borders. Remove logos. Rewrite dialogue. Upscale HD to giant screen. Remix music without paying for it. But what actually survives projection at scale? This session cuts through the noise and examines what AI really delivers for giant screen production — and where it quietly creates new risk. We will draw a clear line between:

  • Procedural AI — trained on your own footage, generally lower legal exposure
  • Generative AI — creating new content, introducing copyright, ethical, and continuity risk

What We’ll Show:

  • Scene Extensions: Extending frames for IMAX and dome. Still frames vs moving shots. Where it works. Where it breaks. How quickly “easy” turns into real VFX labor.
  • Highlight Recovery: Salvaging blown-out windows. Recovering underexposed footage.
  • Practical, documentary-friendly tools — and where instability creeps in.
  • Cleanup & Rotoscoping: Logo removal. Object removal. Human masking.
  • AI accelerates the work — but doesn’t eliminate finishing.
  • Upscaling & Denoising: How far can SD or HD really be pushed? When enhancement becomes “plastic.” Why dome may forgive what flat screen exposes.
  • Audio Manipulation: Voice isolation: useful and safe. Voice cloning: legally explosive. Music stem separation: technically simple, contractually dangerous. Removing preview watermarks from music libraries is now trivial. Publishers can now detect unauthorized use just as easily. Just because you can doesn’t mean you should.
  • Scripting, Previsualization:  AI can be a powerful tool for scriptwriting, and prompts can be written to generate storyboards from scripts.

Open Source Generative Systems: Tools like ComfyUI (comfy.org) offer powerful, node-based generative workflows — free, flexible, and largely unguarded. They expand creative power. They also expand liability.

The Reality: AI does not eliminate artists. It increases productivity — and expectation. Fewer people. More output. Cloud-based cost models. Uncertain bidding structures. And a production landscape that rewards those who understand the limits as much as the capabilities. This session shows the wins, the failures, the budgeting implications, and the legal guardrails. No hype, just clarity.