MiniMax H3 — 10Eros + Semantic Bridge + Ultimate Upscale (8GB T2V/I2V)

A two-stage MiniMax H3 text-to-video graph for consumer 8GB cards.

1. Stage 1 — 10Eros TURBO-hybrid generates a short AV clip at 0.3–0.4 MP in 6 steps (euler / simple). Turbo is already baked into 10Eros. Do not add LightX2V / ModelTC / other speed LoRAs.

2. Semantic Bridge — an ~11 MB conditioning adapter (not a second video model). It blends a learned semantic residual into H3 text cond. Goal is tighter prompt structure, not extra VRAM. SenseNova is not needed at runtime.

3. Stage 2 — MMH3 Ultimate Upscale with the official H3 3D latent upscaler, 8GB tiles, denoise 0.18.

H3 is CFG-distilled. This graph uses BasicGuider — no negative prompt, no CFG. Put exclusions in the positive prompt.

Example shot in the JSON: SFW golden-hour causeway, native stereo audio.

More graphs: artrealmai.com (https://artrealmai.com)

Custom nodes (2 extra packs)

Everything else is ComfyUI 0.33+ native MiniMax H3.

1. Comfyui-MMH3-UltimateUpscale (https://github.com/bbaudio-2025/Comfyui-MMH3-UltimateUpscale) → custom_nodes/

2. MiniMax H3 Semantic Bridge (https://huggingface.co/speach1sdef178/MiniMax-H3-Semantic-Bridge) — extract MiniMax_H3_Semantic_Bridge_v1.0.zip into custom_nodes/ (not in ComfyUI-Manager)

Restart ComfyUI after both are in place.

Files

diffusion_models

10Eros skip-edges INT8 (https://huggingface.co/cicalooo/10Eros-Max-h3-int8-convrot/resolve/main/10Eros_Max_h3_TURBO-hybrid_beta3_int8_convrot_skip_edges.safetensors)

text_encoders

qwen3vl NVFP4 (https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/text_encoders/qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors)

vae

video fp16 (https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/vae/minimax_h3_video_vae_fp16.safetensors) · audio fp32 (https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/vae/minimax_h3_audio_vae_fp32.safetensors)

latent_upscale_models

H3 3D upscaler fp16 (https://huggingface.co/LBH-123-AI/Minimax_h3_latent_Upscaler/resolve/main/minimax_h3_latent_upscaler_3d_fp16.safetensors)

semantic_bridge (models/semantic_bridge/)

MiniMaxH3_SemanticBridge_v1.safetensors (https://huggingface.co/speach1sdef178/MiniMax-H3-Semantic-Bridge/resolve/main/MiniMaxH3_SemanticBridge_v1.safetensors)

CLIPLoader type = minimax.

8GB defaults (as shipped)

• Stage 1: 9:16 @ 0.4 MP, euler / simple / 6

• Stage 2: 9:16 @ 0.9 MP (1280×736), euler / simple / 6, denoise 0.18, tiles 512×384

• Semantic Bridge alpha 0.10, per_token

• Launch: --lowvram --force-fp16 --use-ck-attention

• Load Image bypassed = T2V

• Saves video/MiniMax_H3_Original and video/MiniMax_H3_UltimateUpscale

Start alpha at 0.10. 0.15 is stronger (the adapter author’s A/B setting). Experimental on 10Eros — the adapter was distilled on native H3 FL2VA conditioning, which this graph still uses. Do not use it on Ref2VA.

Do not add turbo LoRAs. Do not connect a negative prompt. Extra style LoRAs: stock Load LoRA (Model Only) after UNET.

Credits

MiniMax / Comfy-Org · 10Eros TenStrip / cicalooo · Semantic Bridge speach1sdef178 (https://huggingface.co/speach1sdef178/MiniMax-H3-Semantic-Bridge) · MMH3 Ultimate Upscale bbaudio-2025 (https://github.com/bbaudio-2025/Comfyui-MMH3-UltimateUpscale) · graph ArtRealmAI

License

Free to use for personal and commercial generation. Videos you create with this graph are yours.

You may not sell, rebundle, paywall, or republish this workflow JSON (or a trivial copy) without permission from ArtRealmAI. Ask at artrealmai.com (https://artrealmai.com).

Upstream MiniMax H3 Community License still applies to the H3 weights and to Semantic Bridge as an H3 model derivative.

Tags: minimax h3, 10eros, t2v, workflow, semantic bridge, ultimate upscale, 8gb, comfyui