In the latest episode of the Denoised Podcast, hosts Joey Daoud and Addy Ghani dive into the upcoming NAB Show 2025, breaking down the most significant pre-show announcements and discussing OpenAI's strategic move toward open-weights. From leaked camera updates to workflow innovations, this episode provides a comprehensive preview of what professionals can expect at this year's event.

Blackmagic's Leaked PYXIS 12K Camera

One of the most exciting pre-NAB developments comes from Blackmagic Design, with a leaked image revealing the upcoming PYXIS 12K camera. According to the hosts, this new model appears to feature Blackmagic's RGBW sensor technology, which offers significant advantages over traditional camera sensors.

"The really cool thing about that is if you shoot different resolutions, it automatically can remap on the sensor and you don't have to deal with the crop issue," Joey explains.

This means when shooting in 4K rather than the full 12K resolution, users won't face the typical crop factor issues. Instead, they can maintain the wider field of view while still benefiting from the full sensor's photosensitivity.

The hosts speculate on pricing, noting that with the current PYXIS model at $3,000, the 12K version might land around $4,000-$4,500—though Addy suggests Blackmagic might surprise everyone by keeping it at $3,000 to maintain competitive advantage against cameras like the RED Komodo.

Key takeaways:

• The PYXIS 12K will feature Blackmagic's RGBW sensor technology

• Sensor remapping allows for shooting at lower resolutions without significant crop factor changes

• Pricing remains uncertain but likely competitive within the affordable cinema camera market

Joey also raises an interesting economic question about the impact of recent tariffs on camera pricing, noting that most cinema camera sensors are not manufactured in the US, potentially affecting consumer prices across all manufacturers.

Strada Agents: Direct Peer-to-Peer Media Access

Workflow innovation takes center stage with Strada Agents, transforming how remote teams access media files. This new software solution allows users to make any connected hard drive remotely accessible without requiring cloud storage intermediaries.

"It's basically a piece of software you can load onto your computer and it will make any hard drive you have plugged into your computer remotely accessible," Joey explains.

The technology enables direct peer-to-peer connection, allowing users to stream or download media files from remote hard drives. Joey tested an alpha version, noting that the playback was smooth and the implementation was surprisingly effective.

Unlike traditional media asset management systems that require uploading to cloud services before sharing, Strada Agents creates direct connections to the original media, eliminating redundant copies and potentially reducing costs.

Key benefits:

• Direct access to remote hard drives without cloud intermediaries

• Stream or download original media files from anywhere

• Eliminates the need for duplicate files and cloud storage costs

• Designed specifically for video file workflows

The hosts compare this approach to Blackmagic Cloud, noting that while Blackmagic offers a polished camera-to-cloud ecosystem, it still requires cloud storage as an intermediary step. Strada Agents bypasses this requirement altogether, essentially turning any hard drive into a personal cloud storage solution.

Adobe's AI Integration and Feature Updates

Adobe announced several significant updates to its Creative Cloud applications, with two particularly notable features for video professionals: Smart Search and the official release of Generative Extend.

Smart Search introduces contextual video analysis capabilities in Premiere Pro, allowing editors to search footage based on visual content rather than just metadata.

"It can analyze your footage and you can start searching based on context," Joey explains. "You can search your footage, you know, 'the plate' or like 'the medium shot of a person.'"

This represents a practical application of AI that addresses a real workflow challenge—quickly finding specific shots within large amounts of footage.

Generative Extend, while previously available in beta, is now officially included in the public release. This feature uses Adobe's Firefly AI model to generate additional frames when extending clips beyond their recorded length—useful for creating smooth transitions or fixing timing issues.

Key Adobe updates:

• Smart Search enables contextual video content searching within Premiere Pro

• Generative Extend moves from beta to official release

• Frame.io adds native transcription capabilities

• Adobe emphasizes they don't train on user content, addressing previous concerns

The hosts note that Adobe's Frame.io also received an update, adding built-in transcription capabilities that keep more of the workflow within a single ecosystem. However, Joey points out that Frame.io version 4 still lacks API access a year after its release, limiting integration options.

Gaussian Splats: The Rising Star of Virtual Production

A significant trend emerging from pre-NAB announcements is the increased adoption of Gaussian Splats technology—a new approach to creating photorealistic 3D environments that's gaining traction in virtual production workflows.

Mo-Sys announced a "scan-to-shoot VP workflow" that combines LiDAR scanning with Gaussian Splats to create realistic virtual environments. This allows productions to scan locations in one part of the world and recreate them on LED volumes elsewhere.

"They scan something in London, sent it to Studio X in Thailand... and they shot as if they were in London in Thailand," Addy explains.

The technology presents significant advantages over traditional photogrammetry, offering realistic parallax effects with substantially less processing work. However, challenges remain regarding resolution and real-time performance for LED wall applications.

Other companies embracing Gaussian Splats include:

• Lightcraft: Integrating Gaussian Splats into their Jetset app for phone-based camera tracking, while also capturing LiDAR data for post-production

• Volinga: Developing ACES color-managed Gaussian Splat workflows, addressing critical color space management issues for professional production

Addy emphasizes the importance of Volinga's ACES implementation: "This is a really big deal. Doesn't seem like much, but without this step, Gaussian Splats won't be taken seriously in the production world."

The hosts note that Gaussian Splats represent a practical application of AI technology focused on recreating rather than generating content, with applications extending beyond film and TV to museums, virtual experiences, and digital twins.

Sony's VENICE Extension System Mini

Sony continues to evolve its cinema camera ecosystem with the announcement of the VENICE Extension System Mini, a more compact and production-ready version of their sensor extension technology.

This system allows the sensor and lens of Sony's high-end VENICE 2 cinema camera to be separated from the main camera body via a cable, enabling placement in tight spaces where the full camera wouldn't fit.

"The VENICE 2 is about the size of your head," Addy explains. "Let's say you want to shoot Top Gun, Tom Cruise is in a cockpit... There is probably like a six-inch by six-inch space to put a camera, so you can't put a VENICE 2 there. So this cable extends just the sensor to that little spot."

The Mini version appears to be a more productized, smaller implementation of this technology compared to the earlier Rialto system, which was perceived as more specialized and boutique. Addy mentioned industry rumors of similar extension systems potentially coming to Sony's more accessible camera lines like the FX6.

Key details:

• Smaller than the original Rialto system, approximately the size of a GoPro

• Maintains the same high-quality sensor capabilities of the VENICE 2

• Connected to the main body via cable (up to 13 feet)

• Allows full-frame cinema camera quality in extremely tight spaces

OpenAI's Move to Open-Weights: Strategic Shift

In a significant development outside of NAB, OpenAI announced plans to release an open-weight model, marking a strategic shift for the company that has often been criticized for its closed approach despite its name.

"For them to just come out and release, quote-unquote, their secret sauce of a model, that's huge," Addy explains, comparing it to DeepSeek, Stable Diffusion, and Meta's Llama models, which already operate as open-weight systems.

The hosts discuss the business implications of this move, suggesting it's less about altruism and more about addressing enterprise needs. Major companies like Disney, Amazon, and Netflix require AI solutions that can operate within their secure environments rather than relying on cloud-based "black box" systems.

Open-weight models allow organizations to:

• Deploy AI locally within secure environments

• Maintain control over sensitive data

• Customize models for specific needs

• Avoid sending proprietary information to third-party services

The conversation expands to discuss the broader business challenges facing AI companies, with Addy noting: "The money is not selling $20, $200 a month subscriptions. The money is selling a $1 million a year subscription to a big B2B company."

Both hosts speculate that the future of AI monetization might follow models similar to open-source software companies like Red Hat, where the core technology is free, but support, customization, and enterprise-grade implementation generate revenue.

Conclusion

The pre-NAB announcements reveal a media technology landscape increasingly focused on practical AI applications, streamlined workflows, and making high-end tools more accessible. From Blackmagic's camera innovations to Strada's peer-to-peer media access and the rising adoption of Gaussian Splats in virtual production, these developments suggest NAB 2025 will emphasize refining existing technologies rather than introducing completely new concepts.

Meanwhile, OpenAI's strategic shift toward open-weights signals important changes in how AI companies are positioning themselves for enterprise adoption, recognizing that control, security, and customization are essential for larger organizations.