Is there an existing issue for this?

• I have searched the existing issues

Operating System

Windows 11 Pro

DeepLabCut version

3.0.0 rc6

What engine are you using?

pytorch

DeepLabCut mode

single animal

Device type

RTX 4060

Bug description 🐛

Hi! I used jupyter notebook to generate high quality video for an analysed data. The generated frames don't look correctly labeled. However, when I check the frames using labeling GUI, the labels look great. I attached an example of same frame, one is saved by video generation and the other is what it looks in the labeling GUI. I used the same model and I didn't crop the video when I analysed it.

Steps To Reproduce

import deeplabcut

config_path = "F:/Deeplabcut/projects/CSA_tongue-Jiachen-2025-09-26/config.yaml"

videos = [
"E:/video/135/135_tongue_day2.mp4"]

deeplabcut.create_labeled_video(
config_path,
videos,
shuffle = 3,
filtered = True,
pcutoff = 0.75,
save_frames = True)

Relevant log output

Loading DLC 3.0.0rc6...
C:\Users\Limlab\anaconda3\envs\deeplabcut\lib\site-packages\tqdm\auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html
  from .autonotebook import tqdm as notebook_tqdm
Starting to process video: E:/video/135/135_tongue_day2.mp4
Loading E:/video/135/135_tongue_day2.mp4 and data.
Duration of video [s]: 3959.55, recorded with 30.01 fps!
Overall # of frames: 118843 with cropped frame dimensions: 1200 1200
Generating frames and creating video.
100%|██████████| 118843/118843 [14:53:23<00:00,  2.22it/s]  
Labeled video E:\video\135\135_tongue_day2DLC_HrnetW32_CSA_tongueSep26shuffle3_snapshot_280_filtered_p75_labeled.mp4 successfully created.

Anything else?

Code of Conduct

• I agree to follow this project's Code of Conduct