We're giving away a Hogwarts Legacy custom wrapped NVIDIA GeForce RTX 5090 to celebrate its official release with DLSS 5 πͺ
To enter:
π’ Share this post
π’ Comment #RTXON
We're giving away a Hogwarts Legacy custom wrapped NVIDIA GeForce RTX 5090 to celebrate its official release with DLSS 5 πͺ
To enter:
π’ Share this post
π’ Comment #RTXON
Terms and Conditions: https://t.co/45T3R732hf
Speed on an RTX PRO 5000, 5 s clip at 2K: refine 474 s β 122 s with Sol-Attn, decode 50 s β 22 s with the fast decoder.
Credit: NVIDIA's Open-H3-Regenerate-2K team, MiniMax H3, LBH's latent upscaler, LightX2V, Alissonerdx.
https://t.co/iJRZk8qfa9
Open-H3-Regenerate-2K, now in ComfyUI.
NVIDIA's open take on H3-Regenerate-2K: redraw any low-res video draft (H3, LTX, Wanβ¦) at 768Pβ4K in 3 steps on native MiniMax H3. 5 s β 2K in ~3 min on an RTX PRO 5000.
Left: 768P draft. Right: 2K regenerate.
Links in the replies π
Nodes + workflows: https://t.co/UdY1l2ST6V
NVIDIA models: https://t.co/oPZoV3t2z5
β’ One loader for the refiner + Turbo + Sharp LoRA
β’ Refined latent saved before decoding: a failed decode costs a minute, not a rerender
β’ Video β H3 Prompt node for your own vision server