Supported OS distributions for MistikaVR:
macOS: (native support for Apple Silicon)
Windows: Windows 10 (x86_64). Windows 11 (x86_64).
Maximum render (output) resolution for MistikaVR:
Mistika does not enforce any resolution limits by itself, but each platform stablish its own limits (typically due to openGL limitations). At the date of this document:
macOS: 16k x 8k
Windows: 32k x 16k
Supported hardware for MistikaVR:
- Minimum number of CPU cores: 4 cores. (32 cores recommended for high performance)
- RAM: Minimum 16 GB. 32GB recommended.
- Display resolution: 1920x1080
- Graphics card model:
Windows:
- An NVIDIA GPU is required on Windows (Pascal generation from 2016 or later). Minimum 8GB of graphics memory, 16GB VRAM recommended. It is important to upgrade the nvidia driver to a recent version.
macOS:
- Mac Apple Silicon (ARM) models (M1/M2/M3/M4...): They are natively supported. It is recommended to have 16GB RAM or more.
- Mac Intel (Legacy) models: Requires a discrete GPU from AMD. Mac models from "late 2013" are supported.
Recommended hardware configuration for Mistika VR
Requirements for working at high resolutions (8K or more) or with many source cameras (6 or more):
- Most of the internal processing inside Mistika VR is made by GPU: Geometry adjustments, Color adjustments, Optical Flow (which is typically the bottleneck), etc. So the GPU is normally the most important component (but not always, as explained below).
- In the other hand, the processes that are let to CPU processing are the decoding of camera codecs and the encoding of the render codec. With these exceptions on Windows:
* NVIDIA nvenc h264 / nvenc h265 codecs are rendered by GPU. Decoding H264/H265 camera rushes may use the NVIDIA GPU as well, but not in all cases (this is work in progress). RED R3D is also decoded by GPU.
Note: The NVidia encoder uses dedicated hardware for render encoding that is not used for any other task. This means that using those H264/H265 codecs will not reduce the render speed. If you want to take advantage of this capability we recommend a NVIDIA model from the Pascal generation or later, as previous models are discontinued and not supported for this (and can not do hardware encoding at 8K resolution anyway. For more details about supported encoding formats: https://developer.nvidia.com/video-encode-and-decode-gpu-support-matrix-new#Encoder )
Decoding camera rushes and rendering may require a lot of CPU power, depending on the number of cameras, their codecs, and their resolution. Up to the point that In many cases the bottleneck is the CPU and not the GPU. In these cases Mistika VR will take advantage of as many CPU threads as possible (up to 32 core / 64 Threads, as past those numbers there are diminishing returns). To study your particular case check your CPU usage in the task manager / activity monitor.
- Meanwhile, the decoding and encoding of uncompressed formats do not use significant GPU or CPU resources, what they need is high storage speed (but a lot of it!).
So it depends. Projects using many camera rushes based on compressed codecs or requiring heavy CPU encoding to CPU intensive formats such as ProRes and EXR can take advantage of having many CPU cores at high clock speeds. Meanwhile, stitching smaller camera rigs and then rendering to NVIDIA h264/h265 deliveries will not take advantage of high end CPUs at all. We recommend to watch the Windows Task Manager or macOS Activity Monitor while playback and rendering to find where are the hardware bottlenecks for your particular case.
- Mistika VR only uses one GPU, the one where it is launched. Do not expect extra speed by having two GPUs.
Another recommendation could be this: In general, the highest end models of CPUs and GPUs only provide a bit more speed than models that are a bit inferior, but they can cost much more. In the worst cases you may need to pay 100% more for just 15% additional speed, compared with cost efficient hardware in the sweet spot.
For this reason, it can be better to use two (or more) computers with just "good" specs rather than one only super workstation with top specs..
- Regarding RAM: 8GB is the absolute minimum for very simple cases (few source cameras and rendering up to 4K). At least 16GB are recommended, and 64 GB is a good reference for complex cases. It mainly depends on:
* The number of cameras and their resolution
* The number of CPU cores. Each CPU core can be working on a different frame in parallel, so using many cores requires a lot of extra RAM to take full advantage of them.
* The render resolution (obviously).
* When using network storage, the OS will normally use the unused RAM as a read / write cache, so even if Mistika is not using it directly it can help in these cases.
* The memory type is also important. For example, systems with DDR5 or LPDDR5 will render faster than systems with DDR4 (with everything else similar)
- Display resolution: There are no current limitations. In 4K and higher resolution displays the GUI elements may appear too small or big initially (depending on the monitor size), but you will find scaling options in the File -> Preferences -> Interface menu.
- Storage speed: Ideally, as fast as needed to read all the camera shots simultaneously while writing the render results at the same time. Also important when rendering to uncompressed formats. Please note that rendering involves both reading and writing at the same time, so if the storage is not fast enough and the destination format is specially big (like DPX or EXR) then it can help to put the original files in one storage volume and rendering to a different one (or at least for testing if your render bottleneck is in the storage access). If you work on network drives (NAS), we recommend to test your reference project on local NVMe disks first (which is the fastest way), so that you will know how much performance is lost when using network drives.