4K AI video on 6GB VRAM: MiniMax H3 in ComfyUI with SelfLift sampling and DLSS5 upscaling
cgpixel23 · reddit · 2026-10-25
A Reddit user shared a low-VRAM ComfyUI workflow that generates 4K AI videos with MiniMax H3 by combining SelfLift progressive-resolution sampling with DLSS5 upscaling nodes — no additional model training required.
Workflow highlights:
- Covers MiniMax H3 generation, SelfLift progressive sampling, and DLSS5 upscaling, with 2K and 4K output options;
- Optimized for low-VRAM GPUs like the RTX 3060 6GB;
- On a 6GB card, generating and upscaling the final 4K result took 35 minutes at 8 steps and 0.6 megapixel.
The full workflow file, a Civitai tutorial article, and the DLSS5 DLL are all shared for direct reproduction.
More from Multimodal
- One prompt, four approvals: Claude produces a 78-second cartoon explainer video — alex_verem · 2026-10-25
- Canva's new Craft image model plans layouts first and splits images into editable layers — alex_verem · 2026-10-25
- Local RTX 5090 test: FLUX.1 Dev keyframes + Wan 2.1 14B I2V nail neck folds and wet-scale physics — LatentForge-Dev · 2026-10-25
- Credit: the Qwen3.5-4B 3D generation model is by community dev ccore2 — victormustar · 2026-10-25
- Lure-TTS: 600M voice models trained for under $200, running 4x faster than real time — smirkishere · 2026-10-25
- MCP turns AI podcast production into a workflow — the hard part is listener retention — aziz4ai · 2026-10-25