VFX Compositor on High-Res Depth Estimation: DepthCrafter vs. Video-Depth-Anything
Legitimate_Age884 · reddit · 2026-08-16
An 8-year VFX compositor shares practical experience using depth estimation models in high-res workflows. He switched from DepthCrafter to Video-Depth-Anything (VDA), which significantly improved depth extraction and integration. However, processing 4K scans directly remains a heavy load even for consumer GPUs like the RTX 5090. His current workaround involves cropping regions or downscaling plates. He uses depth maps for keying, atmospheric effects, and feeding ControlNet pipelines. He asks the community for recommendations on better depth models or optimization pipelines (tiling/upscaling tricks) for handling 4K production footage efficiently.
More from Embodied
- Vision-powered robotic laundry folder worth $3k? — HamsterUnfair6313 · 2026-08-16
- RTX 5090 turns full manga chapter into animated book in ~3 hours, fully automated and local — Ill-Ant-9489 · 2026-08-16
- Seeking budget GPU upgrade to run Qwen 2.5 72B with larger context — pulse77 · 2026-08-16
- Humanoid Use Case: Finding Four Repeatable Tasks — Scobleizer · 2026-08-16
- Berkeley Humanoid Lite Cuts Robot Cost to Under $5k — capetorch · 2026-08-16
- Post-Mortem Brain Tissue Trained to Mimic Piano Playing — 量子位 · 2026-08-16