Views
No views yet
.pt files. TurboVAE, the
T5 text encoder, and the video/audio VAEs still come from the official
releases.540p_sr/ / 1080p_sr/ config directories are not required.Want to try it quickly? Open RunningHub to explore runnable AI workflows. Want to integrate digital human / talking-head video generation into your product? Start with the RunningHub API Documentation.
Submit task -> get taskId -> check task status -> retrieve generated result| File | Size | Role |
|---|---|---|
base_int8.pt | 14.25 GiB | INT8 Base DiT (32 steps, higher quality) |
distill_int8.pt | 14.25 GiB | INT8 Distill DiT (8 steps, faster) |
sr_540p_sr_int8.pt | 14.25 GiB | Optional INT8 540p super-resolution |
sr_1080p_sr_int8.pt | 14.25 GiB | Optional INT8 1080p super-resolution |
base_int8.pt / distill_int8.pt. Add an SR file only
when the loader's sr_model is 540p_sr or 1080p_sr.1ComfyUI/models/MagiHuman/
2├── base_int8.pt
3├── distill_int8.pt
4├── sr_540p_sr_int8.pt # optional
5└── sr_1080p_sr_int8.pt # optional1cd /path/to/ComfyUI
2python3 -m pip install -U huggingface_hub
3
4hf download Gluttony10/ComfyUI-RH-daVinci-MagiHuman base_int8.pt \
5 --local-dir ./models/MagiHuman
6hf download Gluttony10/ComfyUI-RH-daVinci-MagiHuman distill_int8.pt \
7 --local-dir ./models/MagiHuman
8hf download Gluttony10/ComfyUI-RH-daVinci-MagiHuman sr_540p_sr_int8.pt \
9 --local-dir ./models/MagiHuman
10hf download Gluttony10/ComfyUI-RH-daVinci-MagiHuman sr_1080p_sr_int8.pt \
11 --local-dir ./models/MagiHuman1pip install modelscope
2cd /path/to/ComfyUI
3modelscope download --model Gluttony10/ComfyUI-RH-daVinci-MagiHuman base_int8.pt \
4 --local_dir ./models/MagiHuman
5modelscope download --model Gluttony10/ComfyUI-RH-daVinci-MagiHuman distill_int8.pt \
6 --local_dir ./models/MagiHuman
7modelscope download --model Gluttony10/ComfyUI-RH-daVinci-MagiHuman sr_540p_sr_int8.pt \
8 --local_dir ./models/MagiHuman
9modelscope download --model Gluttony10/ComfyUI-RH-daVinci-MagiHuman sr_1080p_sr_int8.pt \
10 --local_dir ./models/MagiHuman1# Official MagiHuman assets (TurboVAE, T5 text encoder)
2hf download GAIR/daVinci-MagiHuman --local-dir ./models/MagiHuman
3pip install modelscope
4modelscope download --model GAIR/daVinci-MagiHuman --local_dir ./models/MagiHuman
5
6# External VAEs
7hf download stabilityai/stable-audio-open-1.0 \
8 --local-dir ./models/audio_checkpoints/stable-audio-open-1.0
9hf download Wan-AI/Wan2.2-TI2V-5B \
10 --local-dir ./models/Ovi/Wan2.2-TI2V-5B1ComfyUI/models/
2├── MagiHuman/
3│ ├── base_int8.pt
4│ ├── distill_int8.pt
5│ ├── sr_540p_sr_int8.pt
6│ ├── sr_1080p_sr_int8.pt
7│ ├── t5gemma-9b-9b-ul2/
8│ └── turbo_vae/
9├── audio_checkpoints/stable-audio-open-1.0/
10└── Ovi/Wan2.2-TI2V-5B/| Node | Purpose |
|---|---|
RH MagiHuman Model Loader (RH_MagiHumanModelLoader) | Load INT8 DiT, VAEs, and optional SR |
RH MagiHuman Generate (RH_MagiHumanGenerate) | Talking-head video from a reference image and text prompt |
model_type: base (32 steps) or distill (8 steps)vram_mode: mid_vram (~16 GB) or low_vram (~6.5 GB)sr_model: none / 540p_sr / 1080p_srmid_vram + 540p SR needs about 24 GB. mid_vram + 1080p SR needs about 48 GB.