HDCOLORS / 8K.ART / Open Studio

Hardware and Storage Plan

The performance-first studio/server architecture developed in the chat.

Performance-first architecture

The home studio is not a single computer. It is a distributed media lab: capture, storage, CUDA processing, network backbone, and publication should each have a clean role.

Canon R5C / dual fisheye / YoloBox / Blackmagic capture
  → ingest
  → office server/workstation
  → Thunderbolt DAS
  → 10GbE / 100GbE backbone
  → RTX 5090 laptop + RTX 5070 Ti workstation + RTX 4070 worker
  → archive, publish, multicast, and iterate

Hardware already identified in the conversation

Canon R5C for 8K capture.
Dual fisheye lens arriving for stereoscopic / VR180 development.
YoloBox workflow for rapid record / test / upload iteration.
RTX 5090 laptop for mobile CUDA and development work.
RTX 5070 Ti desktop workstation for main office AI/video work.
Spare RTX 4070 to become a CUDA / NVENC worker node.
Existing 4 × 16TB mechanical drives that should be moved away from the GPU workstation because of heat.
Large 10GbE / 2.5GbE switch already in the box.

Why the unopened gaming mini PC should go back

The G1 Pro gaming mini PC was powerful but wrong for the central-server role. The real requirements are:

Thunderbolt 4 10GbE minimum PCIe x16 expansion 100GbE path good airflow DAS compatibility

Storage rules

The AI/GPU workstation should not be a storage furnace.
Hot mechanical archive drives should move into a dedicated DAS/server zone with better airflow.
Mother’s damaged S: drive is evidence: do not boot it, do not repair it, do not write to it.
The future S: should be a mapped network share backed by server storage and duplicated to separate backup media.

CUDA home-lab model

RTX 5090 laptop       = fastest mobile CUDA / NVENC / test node
RTX 5070 Ti desktop   = main office production workstation
RTX 4070 spare        = dedicated worker / encoder / batch node
Central server        = storage, DAS, network, archive, automation