From 5d7594b4ebb6bc45d2a58c5849595bed46b46bb9 Mon Sep 17 00:00:00 2001
From: Shmuel Ronen <80190186+ShmuelRonen@users.noreply.github.com>
Date: Thu, 2 Jan 2025 09:33:37 +0200
Subject: [PATCH] README.md

---
 README.md | 19 +++++++++++++------
 1 file changed, 13 insertions(+), 6 deletions(-)

diff --git a/README.md b/README.md
index c2f4f27..ac6079d 100644
--- a/README.md
+++ b/README.md
@@ -53,6 +53,19 @@ ffmpeg-python
 safetensors
 soundfile
 ```
+## Note on Model Downloads
+
+On first use, the node will automatically download required model files from HuggingFace:
+- LatentSync UNet model
+- Whisper model for audio processing
+- You can also manualy download the models from HuggingFace repo: https://huggingface.co/chunyu-li/LatentSync, the checkpoints should appear as follows:
+
+```
+./checkpoints/
+|-- latentsync_unet.pt
+|-- whisper
+|   `-- tiny.pt
+```
 
 ## Usage
 
@@ -70,12 +83,6 @@ The processed video will be saved in ComfyUI's output directory.
 - `seed`: Random seed for reproducible results (default: 1247)
 
 
-## Note on Model Downloads
-
-On first use, the node will automatically download required model files from HuggingFace:
-- LatentSync UNet model
-- Whisper model for audio processing
-
 ## Known Limitations
 
 - Works best with clear, frontal face videos