NEW PODCAST: Sony Alpha Rumors’ Andrea Pizzini’s Walk-In Camera – Focus Check ep138 →🎙️ WATCH/LISTEN Now
Watch/Listen NowSony Alpha Rumors’ Andrea Pizzini’s Walk-In Camera🎙️NEW PODCAST
Education for Filmmakers
Language
The CineD Channels
Info
New to CineD?
You are logged in as
We will send you notifications in your browser, every time a new article is published in this category.
You can change which notifications you are subscribed to in your notification settings.
Adobe and NVIDIA have announced an expanded partnership that rebuilds Photoshop and Premiere from the ground up for NVIDIA’s new RTX Spark superchip, promising up to 2x faster AI, editing, coloring, and effects across creative workflows. The work also brings native RTX Spark acceleration to Substance 3D Painter and Stager, with rollouts expected to begin later this year alongside the first RTX Spark hardware.
The announcement came out of NVIDIA GTC Taipei, where the chipmaker unveiled RTX Spark as a single superchip aimed at a new class of Windows laptops and compact desktops built around local AI. For filmmakers and image editors, the headline is not the silicon itself but what Adobe says it is doing with it: reworking the internal architecture of two of the most widely used applications to take direct advantage of the chip’s unified memory, Blackwell GPU, and TensorRT acceleration. This continues the broader strategic partnership the two companies first announced in March 2026, which paired Adobe’s creative and marketing platforms with NVIDIA’s AI technologies, open models, and accelerated computing.
RTX Spark is an Arm-based superchip that combines a 20-core NVIDIA Grace CPU with a Blackwell RTX GPU featuring 6,144 CUDA cores and fifth-generation Tensor Cores with FP4 precision, linked over NVIDIA’s NVLink-C2C chip-to-chip interconnect. NVIDIA quotes up to 1 petaflop of AI performance and up to 128GB of unified memory, the latter being the detail most relevant to anyone working with large timelines, high-resolution stills, or heavy 3D scenes. MediaTek collaborated on the custom CPU design, and NVIDIA positions the chip as roughly comparable to its RTX 5070 laptop GPU but at significantly lower power draw.
That unified memory pool is the part worth flagging for video and imaging work. Rather than shuttling data between separate CPU and GPU memory, the GPU can address the full pool directly, which is exactly the kind of bottleneck that slows down GPU-accelerated effects, AI features, and high-bitrate decode in a conventional laptop. NVIDIA’s own examples for the platform include editing 12K 4:2:2 video through the Blackwell decoder and rendering 90GB-plus 3D scenes, so the ceiling Adobe is building against is high.
Adobe says Premiere is getting a new video pipeline built specifically around RTX Spark’s unified memory, Blackwell GPU, and TensorRT software. The stated payoff is more real-time performance during editing and color correction, GPU-accelerated AI features, and more efficient rendering of complex timelines. Firefly-powered Generative Extend, which Adobe fully integrated into Premiere with version 25.2 ahead of NAB 2025, sits among the accelerated tools that the company says benefit from the new architecture.
The up to 2x figure is NVIDIA and Adobe’s own claim, measured against unspecified baselines, so it should be read as a vendor performance target rather than an independently verified benchmark. NVIDIA has not published comparative numbers against existing laptops at this stage. Still, the direction of travel is consistent with what Adobe has been doing for some time: the company has repeatedly cited work with NVIDIA on real-time playback and smoother timeline performance for Blackwell GPUs, and Premiere Pro 25.5 already added hardware acceleration for 10-bit 4:2:2 H.264 and HEVC on Blackwell silicon.
On the imaging side, Adobe describes a next-generation Photoshop engine optimized for GPU-accelerated compositing at its core. The company points to live filters, high dynamic range, and new oil and watercolor brushing that it says should feel more natural, all running through an AI-native pipeline accelerated by TensorRT. Firefly-powered Generative Fill is named as one of the headline AI tools carried into the faster architecture.
Compositing-heavy retouching and HDR work are precisely the tasks that benefit from keeping large image buffers resident in fast, unified memory, so the architectural rework is more interesting here than a simple feature bump. As with Premiere, the practical gains will only be clear once the updated builds ship and can be tested against current hardware.
Beyond the two flagship apps, Adobe confirmed that Substance 3D Painter and Stager will run natively on RTX Spark for smoother, more responsive texturing and scene creation. This follows the recent wave of Substance 3D releases and fits the same pattern of pushing real-time 3D work onto more portable hardware.
Adobe also signaled that it will extend Premiere and Photoshop to let users create, edit, and design with Windows agents, framing these as a collaborative teammate inside the apps rather than a separate tool. That language ties directly into NVIDIA’s broader RTX Spark pitch, which centers on running AI agents locally and securely on a primary device rather than in the cloud. How meaningful agentic editing turns out to be for professional post-production remains to be seen, and it is the part of this announcement we would treat with the most measured skepticism until there is something concrete to evaluate.
Updates to Premiere, Photoshop, and Substance 3D are expected to start rolling out later this year, with Adobe indicating they will arrive alongside RTX Spark availability. NVIDIA has said RTX Spark laptops and compact desktops will ship this fall from ASUS, Dell, HP, Lenovo, Microsoft Surface, and MSI, with models from Acer and GIGABYTE to follow. Adobe has not published pricing for the updated applications, which are tied to the existing Creative Cloud subscriptions, and NVIDIA has not detailed pricing for individual RTX Spark systems.
For now, this is an architecture and roadmap story rather than a benchmark one, and the real test will be how the rebuilt pipelines hold up on shipping hardware against the laptops editors and retouchers already own.
Would a unified-memory laptop change how you handle large timelines or heavy compositing work, or are you waiting for independent benchmarks before believing the 2x claims? Don’t hesitate to let us know in the comments below!
Δ
Stay current with regular CineD updates about news, reviews, how-to’s and more.
You can unsubscribe at any time via an unsubscribe link included in every newsletter. For further details, see our Privacy Policy
Want regular CineD updates about news, reviews, how-to’s and more?Sign up to our newsletter and we will give you just that.
You can unsubscribe at any time via an unsubscribe link included in every newsletter. The data provided and the newsletter opening statistics will be stored on a personal data basis until you unsubscribe. For further details, see our Privacy Policy
Johnnie Behiri is a documentary filmmaker with more than three decades behind the camera: he learned the craft in daily television news gathering, filmed for Japan's public broadcaster NHK, and spent eight years shooting and editing for BBC News. He joined cinema5D in 2011 and has co-owned and co-run CineD and MZed ever since. Johnnie writes many of CineD's camera reviews, co-hosts our weekly Focus Check podcast, and heads CineD's manufacturer relationships and product development.