Lightcraft's Spark Levels Up Indie Film

In partnership with

Welcome to VP Land! A new museum in Colorado aims to 'teach' people what life was like in the 90s...complete with a mock Blockbuster.

In our last poll on using AI to reimagine classic films (like The Wizard of Oz at the Sphere), most of you were excited or open to the idea, with some raising concerns about artistic integrity. A few prefer the classics left untouched.

In today's edition:

Google DeepMind unveils Genie 3
Lightcraft launches Spark for production coordination
ChatGPT-5 ends the reasoning debate
Tons of new AI tool updates

Lightcraft Spark Unifies AI and Filmmaking

Lightcraft Technology announced Spark, a browser-based platform that combines AI, 3D tools, and traditional filmmaking to help teams create shots and coordinate productions from pre-production through post.

What Spark includes:

Spark Shot lets filmmakers assemble interactive scenes in the browser using 3D models, animation, audio, AI tools, and realistic camera simulations to test shots before expensive production days.
Spark Live connects team communications by linking voice, video, chat, and 3D interactions directly to specific shots or assets, integrating with external tools like Zoom.
Spark Atlas serves as an artist-focused database that parses scripts and connects every production element to its associated scene or dialogue for streamlined coordination.
Spark Forge automates post-production VFX tasks like tracking and compositing using timeline exports, generating application-specific scripts for Blender, Unreal Engine, Nuke, and other industry tools.

Spark builds on Lightcraft's Emmy-winning Jetset technology and targets a 2026 release, with beta testing starting in the coming months. The platform addresses traditional filmmaking barriers like funding delays and scheduling conflicts by enabling teams to visualize and iterate on creative ideas immediately, potentially helping indie filmmakers secure funding by demonstrating their vision more effectively.

SPONSOR MESSAGE

Join 400,000+ executives and professionals who trust The AI Report for daily, practical AI updates.

Built for business—not engineers—this newsletter delivers expert prompts, real-world use cases, and decision-ready insights.

No hype. No jargon. Just results.

Subscribe free—trusted by leaders

GPT-5 Launches With Unified AI Power

OpenAI has launched GPT-5 to all ChatGPT users, marking a major upgrade that combines multiple AI capabilities into one unified system. The new model automatically chooses between fast responses and deeper reasoning without requiring you to select different model versions.

GPT-5 merges OpenAI's reasoning models with standard GPT capabilities, so you no longer need to pick between different model types for different tasks.
The system now handles multi-step tasks like generating complete software applications, managing calendars, and drafting complex research briefs with minimal direction.
Accuracy improvements span coding, content creation, and professional writing, with significant reductions in AI-generated false information that plagued earlier versions.
Free users get limited GPT-5 queries with automatic fallback to a mini version, while paid subscribers get default access to the full model with expanded usage limits.
Enterprise and educational customers will receive access next week with customized deployment options and higher service guarantees.

Google's Genie 3 Creates AI Worlds

Google DeepMind released Genie 3, an AI system that generates interactive, navigable virtual environments from simple text prompts in real-time. Unlike previous AI models that create static images or linear videos, Genie 3 produces playable worlds you can explore and modify on the fly.

Genie 3 runs at 24 frames per second in 720p resolution, letting you move around and interact with generated environments rather than just watch them.
The system maintains "world memory" that persists your changes and interactions for several minutes, so actions you take stick around as you explore.
You can type new commands like "make it rain" or "add mountains" to reshape environments without restarting, enabling rapid iteration during creative workflows.
The technology works across diverse visual styles from photorealistic landscapes to animated fantasy worlds, all generated procedurally without pre-made assets.
Current limitations include flatscreen-only display (no VR integration yet) and temporary world persistence, though DeepMind positions this as a stepping stone toward more advanced interactive media.

Currently in a limited preview, Genie 3 opens up rapid prototyping possibilities that could compress environment visualization from weeks to seconds.

Huge Week for AI Tools — Here’s What Dropped

ElevenLabs launched Eleven Music, an AI-powered music generation platform that generates full songs from text prompts.

Alibaba Cloud open-sourced Qwen-Image, a 20B-parameter model that rivals GPT-4o in English and leads all open models in Chinese text-to-image generation.

OpenAI released gpt-oss, its first open-weight language models since GPT-2, enabling local deployment and customization.

ComfyUI launched its long-awaited Subgraph feature, allowing creators to package complex AI workflows into reusable, modular nodes.

xAI introduced Grok Imagine, an image and video generator with a “spicy mode” for NSFW content. Some people were able to create nude deepfakes of Taylor Swift, bypassing standard safety filters.

Imagen debuted Imagen for Video (beta) with automated, frame-by-frame color correction that integrates with Premiere.

MeshBlend, a new Unreal Engine 5 plugin by Tore Lervik, lets artists blend any mesh with another in real time for seamless 3D asset integration.

Google released Jules, its AI coding agent, shifting from code assistance to fully autonomous, asynchronous task execution.

Podcast platform Riverside rolled out Co-Creator, an AI suite that turns raw recordings into clips, blog posts, thumbnails, and social media content automatically.

Midjourney introduced HD Video mode (Pro and Mega only), offering 4x the resolution at approximately 3.2x the processing cost.

Anthropic rolled out Claude Opus 4.1, with upgrades in agentic tasks, real-world coding, and reasoning.

Team 2 Films reviews DaVinci Resolve 20.1's new features and key improvements to existing tools.

Stories, projects, and links that caught our attention from around the web:

🖥️ ACES development will now be guided by the Academy Software Foundation, ensuring continuity with leadership from both technical and industry communities.

🎓 Runway has launched a comprehensive 10-course series focused on teaching professionals how to apply AI for advertising throughout the entire creative process. They also launched a 6-part series on integrating Runway's AI tools into VFX pipelines.

⚖️ Universal Pictures has warned that it will sue Big Tech companies if their generative AI tools use Universal's films without permission.

📺 Disney will fully integrate Hulu into Disney+ in 2026, consolidating both streaming services under a single platform.

✨ SideFX previews new Houdini 21 features, highlighting workflow enhancements and powerful visual effects tools for VFX and animation.

🚀 LumaTouch has revealed major software advancements aimed at streamlining mobile video editing for professional filmmakers, including opening doors to third-party plugins.

Let’s unpack the Sphere’s Wizard of Oz AI controversy, check out what Showrunner AI is doing, and see how filmmakers are blending practical effects with generative tools to craft stunning low-budget productions.

Read the show notes or watch the full episode.

Watch/Listen & Subscribe

Spotify | Apple Podcast | YouTube