FAQ

Frequently asked questions about 360 Hextile

Last updated: 2026-10-27

General Questions

What is 360 Hextile?

360 Hextile is a desktop application that builds production 360° worlds using open-weight AI diffusion models on your own NVIDIA GPU. Build a brand-new world from a text prompt with the built-in 360° LoRA, or transform your own equirectangular image. Sphere-space hexagonal tiling on a polyhedral mesh is designed to reduce boundary artifacts across the full sphere.

Do I need my own 360° image to use 360 Hextile?

No. You can generate a panorama entirely from a text prompt using the built-in 360° LoRA — no source image required. If you do have an equirectangular image (from a 360° camera, a 3D render, Street View, etc.), you can bring it in and transform it instead. Both paths are first-class.

How much does it cost?

360 Hextile is a one-time purchase — no app subscription or per-render cloud processing fee for local generation. Optional third-party AI provider charges may apply; see Copilot. Standard is what you can buy at launch (full list $360; public beta list $180 when beta is active). Pro is not for sale at launch. See our pricing page for current pricing.

What hardware do I need?

Minimum: - Windows 10 (64-bit, 21H2+) or Windows 11 - CUDA-capable NVIDIA GPU with a driver compatible with the CUDA runtime shipped in the current release - VRAM depends on the pipeline — there is no single VRAM floor for every model. SDXL (the default) needs about 12 GB; lighter pipelines can run with less. - System RAM needs depend on the selected model and placement. Qwen-Image and CPU offload have no measured host-RAM minimum. - Enough free storage for the selected model downloads, installation, and outputs

Recommended: - Check the in-app model compatibility guidance for your GPU and chosen pipeline. - Extra system RAM may help when using CPU offload. - SSD storage with room for the models and projects you choose

Do I need an internet connection?

Local image generation uses your installed models and GPU. Internet is needed for initial online license verification, model downloads, updates and the online features you choose.

After a successful online license check, rendering can continue offline for seven days after the last successful check. Reconnect before that window expires. A saved key alone does not establish first-time offline access. See License Activation.

Copilot needs a reachable AI provider even when its documentation is cached. Gallery publishing, online prompt assistance and sending feedback also need a connection. See Privacy and network use.

What operating systems are supported?

Currently Windows 10 and Windows 11 (64-bit) with an NVIDIA GPU. CUDA is required.

Processing Questions

How long does processing take?

Depends on template size, pipeline, and GPU. Smaller templates finish sooner; heavier models and higher tile counts take longer. We do not publish wall-clock minute estimates — measure on your hardware.

What image formats are supported?

Input: JPG, PNG (equirectangular 2:1 aspect ratio) Output: PNG, JPG, Radiance HDR, OpenEXR, cubemap, and fulldome

What is the maximum output resolution?

Any valid license gets full output and basic sequences — no resolution gate by licence tier. The default equirectangular canvas is often 8192×4096; tile templates and Real-ESRGAN Alpha upscaling can reach the application cap of 16384×8192, subject to GPU/VRAM. At launch only Standard is for sale (2 machine activations). Pro (planned later) adds advanced animation workflows and 4 activations — not a higher resolution cap.

What AI models are included?

The current production diffusion families are SDXL, SD 3.5, SD 2.1, SD 1.5 and FLUX.2 Klein. FLUX.1 Schnell, Qwen-Image, Chroma and Z-Image are Alpha; availability depends on your build and enabled features. TeleStyle style transfer and Real-ESRGAN upscaling are separate Alpha tools.

Check the pipeline guide and the app's model catalogue before downloading. Alpha features are available for testing and feedback; they are not a production-readiness promise.

Technical Questions

Why hexagonal tiles?

Hexagons provide better sphere coverage, fewer hard boundaries than square tiles, and more even distribution across the sphere. Tiles are processed in sphere-space on a polyhedral mesh (up to a 92-face Goldberg sphere) so neighbours share context. Results depend on your input and settings; inspect tile boundaries and poles before delivery.

Does it work with AMD GPUs?

Not currently. 360 Hextile requires NVIDIA CUDA for GPU processing.

Can I use custom LoRA models?

Yes. You can install and swap LoRA models to unlock specific artistic styles and aesthetics for environments and worlds.

Can I use ControlNet?

Yes. Managed ControlNet models are available for SDXL, SD 2.1, SD 1.5, and Qwen-Image Alpha. Each pipeline supports a different set of Canny, Depth, OpenPose, Scribble, Soft Edge, MLSD, Segmentation, Normal, and Lineart modes. See the ControlNet compatibility guide.

Purchase & Licensing

Can I use it commercially?

Yes. Your license permits both personal and commercial use. All content you create with 360 Hextile is yours.

How many machines can I install it on?

Standard licenses include 2 machine activations. Pro (not for sale at launch) is planned with 4. You can deactivate a machine to free up an activation slot.

Is there a money-back guarantee?

Yes — 30-day money-back guarantee, no questions asked. See our refund policy for details.

How do I activate my license?

After purchasing, you'll receive a license key. Launch 360 Hextile, enter your key, and activation is automatic. See the Getting Started guide for step-by-step instructions.

Troubleshooting

"CUDA out of memory" error?

Solutions: - Use a coarser factory template (Cube or Dodecahedron) - Close other GPU-intensive applications - Check VRAM headroom for the active pipeline; SDXL (the default) needs about 12 GB

Processing is very slow?

Check: - GPU temperature (thermal throttling?) - Other applications using your GPU - Try a smaller template first - Ensure you have sufficient VRAM

Where can I get help?


Didn't find your answer? Try searching the docs, ask in our Discord, or contact us.

Esc
Searching...
No results found.
Type to search the documentation
↑↓ Navigate ↵ Select Esc Close

We use cookies for analytics to understand how visitors use our site. Privacy Policy