"Blown away by the results. Highly recommend!"

— Jason Key, Principal VFX Artist at JangaFX

At least 5–10 times faster to render.
Up to 80–93% smaller data footprints.

"Сrazy compression,
crazy speed!"

— Ivan Safronov, Unreal Engine Cinematic Artist

Simpler workflows, faster throughput,
and
more iterations.

"The team is incredibly
responsive to feedback."

— Christian Hecht, Director & Head of Real-Time at Tendril Studio

What Is ZibraGDS

ZibraGDS (Geometry Data Structure) is the latest addition to our suite of industry-leading rendering and data optimization technologies. A high-performance, GPU-friendly, multi-attribute geometry compression/decompression/rendering solution, it is designed for storing and rendering extremely high quality simulations and animations across a broad range of applications: real-time cinematics, virtual production, location-based entertainment, 3D motion graphics, VFX, games, digital twins, and machine learning.

Our proprietary compression engine and rendering pipeline deliver stunning visual fidelity at significantly lower memory and rendering costs. VFX artists can now author film-quality effects for runtime usage, while ML pipelines can benefit from unprecedented realism that meet training demands.
ZibraGDS (Geometry Data Structure) is the latest addition to our suite of industry-leading rendering and data optimization technologies. A high-performance, GPU-friendly, multi-attribute geometry compression/decompression/rendering solution, it is designed for storing and rendering extremely high quality simulations and animations across a broad range of applications: real-time cinematics, virtual production, location-based entertainment, 3D motion graphics, VFX, games, digital twins, and machine learning.

Our proprietary compression engine and rendering pipeline deliver stunning visual fidelity at significantly lower memory and rendering costs. VFX artists can now author film-quality effects for runtime usage, while ML pipelines can benefit from unprecedented realism that meet training demands.
Simulations and assets from: The Pixel Lab, the point and prim, Carsten Baars, Ivan Safronov (incl. renders), Viswesh Rameshkumar, SideFX, Fab.

Industries & Use cases

Built for Your Pipelines

Visual Effects
Games
Digital Twins
ZibraGDS helps you deliver projects with extraordinary visual fidelity, ensuring high performance and reliability for the artists, while significantly reducing storage, asset delivery, and iteration costs.
Ray/Path Tracing
Unprecedented Resolutions
Multiple Streaming Options
Easy to Combine with ZibraVDB
Real-Time Cinematics
Virtual Production
Location-Based Experiences
Feature/Episodic VFX
Motion Graphics
Advertising
ZibraGDS presents a DCC-agnostic, simple, and unified workflow for bringing high-quality VFX assets to games. What used to require much more time and effort to export and set up, now only takes a few clicks in a fast and streamlined workflow.
60 FPS
Meshlet Culling
Motion Blur
Native Material/Lighting Features
Game Cinematics
In-Game Cutscene
Gameplay VFX
ZibraGDS extends the realism and richness of virtual worlds, accurately capturing scenarios previously difficult to incorporate in real-time rendering and training pipelines, helping to reduce data generation costs and enhance model capabilities.
Environment Simulations
Destruction Simulations
Highly Detailed Dynamics
Realistic Rendering
Simulations
Visualizations
Synthetic Data
Computer Vision
Machine Learning

Licensing

Choose Your License

Educational

Students & Academics

Free

Free for students & academics
Up to 5m vertices / frame
Full feature access
Free updates
Standard support
Commercial use

Personal

Revenue/Funding < $100K

$4.9

per seat / month
Free studio trial
Up to 2m vertices / frame
Full feature access
Free updates
Standard support
Commercial use

Indie

Revenue/Funding < $1M

$14.9

per seat / month
Free studio trial
Unlimited resolution
Full feature access
Free updates
Standard support
Commercial use

Studio / Enterprise

Revenue/Funding > $1M

Site-Wide Deal

Free studio trial
Unlimited resolution
Custom builds and tooling *
Priority build access
Priority / end-to-end support *
Commercial use
* See details below

Educational

Students & Academics

Free

Free for students & academics
Up to 5m vertices / frame
Full feature access
Free updates
Standard support
Commercial use

Personal

Revenue/Funding < $100K

$49

per seat / year
Free studio trial
Up to 2m vertices / frame
Full feature access
Free updates
Standard support
Commercial use

Indie

Revenue/Funding < $1M

$149

per seat / year
Free studio trial
Unlimited resolution
Full feature access
Free updates
Standard support
Commercial use

Studio / Enterprise

Revenue/Funding > $1M

Site-Wide Deal

Free studio trial
Unlimited resolution
Custom builds and tooling *
Priority build access
Priority / end-to-end support *
Commercial use
* See details below
Educational
Personal
Indie
Studio / Enterprise
Access
Free Studio Trial
Licensing Structure
Per-seat
Per-seat
Per-seat
Hybrid site-wide
Local License Server
Contact us
Resolution Limit
5m vertices / frame
2m vertices / frame
Unlimited
Unlimited
Full Feature Access
Free Updates
Priority Build Access
Source Code Access
Upon request
Integration Assistance
Upon request
Custom Builds
Upon request
Custom Tooling
Upon request
SEAT TYPES
Interactive Seats
View-Only Seats
Headless Seats
Support
Support Tier
Standard
Standard
Standard
Priority /
end-to-end
Email support
Discord Support
Priority Response
Private Slack Channel
Support Calls
TD/TA Support
Engineering Support
Upon request

Trusted by

What Do Artists Say

" What excites me the most personally, is how both ZibraVDB and ZibraGDS open the door to having much larger and more complex effects running real-time in Unreal Engine. Far beyond of what’s natively supported. Not having to worry about memory constraints frees the creative process and allows us to push boundaries. Having worked in close collaboration with the Zibra AI team, I was surprised at how incredibly responsive to feedback the team is and their ability to solve issues in mere days. Making the user experience smoother is always top of mind. I can’t wait to see where the team is taking real-time effects in the future! "

— Christian Hecht, Director & Head of Real-Time at Tendril Studio

" As a real-time artist, I used to feel limited by Unreal when it came to heavy simulations, but ZibraGDS gave me the spark to dive back into learning Houdini and complex simulations. Сrazy compression, crazy speed! "

— Ivan Safronov, Unreal Engine Cinematic Artist

" I stress tested the ZibraGDS beta and was blown away by the results. I was able to utilize large liquid simulation caches (40GB+) from LiquiGen that previously would not even import into Unreal. With ZibraGDS I got up to 93% file size compression, with smooth playback in-engine and no perceptible quality loss. Highly recommend! "

— Jason Key, Principal VFX Artist at JangaFX

Ready to Get Started?

ZibraGDS FAQ

Frequently
Asked Questions

How much does ZibraGDS compress my assets, and how fast is it?

ZibraGDS can support millions of polygons per frame and thousands of frames per sequence. Compression achieves an 80–92% reduction in data size, depending on the characteristics of the input mesh sequence. Assets import 3–4x faster than with existing techniques, and rendering (including decompression) is at least 5–10x faster than existing solutions.

What types of assets is ZibraGDS best suited for?

ZibraGDS is best suited for high-resolution animated mesh sequences that are traditionally too heavy or too complicated to bring into real-time engines, especially fluids, destruction, growth simulations, 3D motion graphics, and other types of artistic designs and visualizations.

Can ZibraGDS be used together with ZibraVDB?

Yes. ZibraGDS works seamlessly with ZibraVDB, Zibra AI's volumetric compression and rendering solution, allowing artists to author complex, two-way coupled simulations involving volumes and meshes and combine the effects in a real-time scene.

Can I control quality, file size, and memory usage?

Yes. ZibraGDS exposes compression quality controls for individual attributes. It also provides options for recomputing normals, computing tangents, generating motion vectors, removing duplicate faces, and applying transform adjustments during compression. For playback, artists can choose whether to load compressed frames into VRAM, stream from RAM, or stream from disk.

What input formats and geometry does ZibraGDS support?

ZibraGDS imports Alembic (.abc) files. Only meshes with triangles and quads as polygons are supported. Topology is compressed losslessly, while other attributes (positions, normals, UVs, vertex colors, and velocities) use efficient lossy compression with per-attribute quality controls.

How do I activate my license?

Get a license key at zibra.ai/zibragds, then go to Tools > ZibraGDS > Open License Settings in Unreal Engine and enter your key in the License Key field. A "License successfully verified" notification confirms activation.

How do I compress an asset?

In Unreal Engine, go to Tools > ZibraGDS > Compress Alembic File, choose your Alembic file in the File Path parameter, optionally adjust Compression Settings, and press Compress. The compressed ZibraGDS asset appears in your Content folder. We recommend keeping the default Quality values and adjusting only if you notice unacceptable quality degradation.

What streaming and playback options are available?

Three streaming methods let you balance performance against memory: “None” loads all compressed frames to VRAM for the best performance; “From RAM” keeps compressed frames in RAM and streams to VRAM; “From Disk” streams compressed frames from disk through RAM to VRAM, with configurable frames-ahead and frames-behind buffers. Playback controls include frame rate, speed multiplier, start frame, and frame interpolation for constant-topology geometry. ZibraGDS is also fully integrated with Unreal Level Sequences.

Which Unreal Engine versions and platforms are supported?

ZibraGDS for UE 0.7 supports Unreal Engine 5.3 through 5.8. On Windows, it requires D3D12 with Shader Model 6 (SM6) and the latest Visual C++ Redistributable.

Does ZibraGDS work with Unreal Engine's lighting and rendering features?

Yes. ZibraGDS output integrates into the engine's native lighting, rendering, and material pipelines, with support for Lumen, ray tracing, path tracing, ray-traced shadows, virtual shadow maps, Substrate materials, lighting channels, custom depth/stencil, and translucency sorting. Motion blur is supported via the velocity attribute.

How is ZibraGDS different from Nanite?

Nanite is primarily aimed at static geometry and certain animated mesh workflows, while ZibraGDS is focused on animated mesh sequences, especially real-time VFX where topology, attributes, streaming, and per-frame geometry changes create additional challenges.

What changed in ZibraGDS 0.7?

Version 0.7 added several workflow and rendering improvements, including a Reimport button, a custom Sequencer playback track, a ZibraGDS asset editor, support for up to 3 UV channels, velocity generation for constant-topology geometry, tangent computation during import and runtime, custom depth and stencil support, translucency sorting, speed multiplier properties, lighting channels, automatic triangulation for N-gons, several Alembic import fixes, RAM/disk streaming options, and optimized VRAM footprint.

Are compressed assets compatible across Beta, 0.7, and future 1.0+ versions?

Future 1.0+ versions of ZibraGDS will have backward compatibility among themselves. Unfortunately, compressed assets created with 0.7 will not work with future 1.0+ versions due to planned file format improvements. Please consider keeping a copy of your original geometry sequence so you can re-compress in newer versions. Please also note that 0.7 is not backward compatible with the Beta version. A step-by-step migration guide in the ZibraGDS documentation covers upgrading projects, re-importing assets, and re-linking actors and Level Sequences.

AI Usage Disclosure

AI Usage Disclosure

ZibraGDS does not utilize or depend on any generative AI or neural networks. We do not pre-train any AI models or collect any user data for AI training in the development of this product. The product also does not embed any third-party AI tools or contents.