The comfy-native cuts of the MiniMax-H3 × Z-Image graft: Z-Image's spatial-attention profile on H3's engine — richer sets and textures, same identity, no per-shot sharpening creep. Full story, demos and verification on the GGUF page.

Load with the plain Load Diffusion Model node, ComfyUI 0.32+. Files are the pruned H3 builds with the graft baked in (zs05 = late-block gains, dose 0.5):

bf16 — the master (ref2va)
comfy-fp8 / fp8e5m2 — fp8 scaled
comfy-int8 / int8_convrot — the fast pick on RTX 50
comfy-w4a8 / w4a4 / nvfp4 — 4-bit family for 16 GB cards (w4a8 is the quality pick; nvfp4 is Blackwell-native, emulated elsewhere)
comfy-mxfp8 — 8-bit microscaling, Blackwell-specialized; little benefit on older architectures
fl2va and ref2va variants where both exist; picking rules are identical to stock H3

RTX 30/40: the GGUF repo is 4–8× faster than any 4-bit comfy-native arm on Ampere.

Made by joeygambino. Questions: open a discussion — I answer.

TechnoBaptist/MiniMax-H3-x-Z-Image-native

MiniMax-H3 × Z-Image — spatial detail graft (comfy-native)

TechnoBaptist/
MiniMax-H3-x-Z-Image-native