offline_render.md

Offline Rendering

python render.py -m <path to trained model> # Generate renderings after training

Command Line Arguments

• --model_path / -m

  Path to the trained model directory you want to create renderings for.

• --skip_train

  Flag to skip rendering the training set.

• --skip_val

  Flag to skip rendering the test set.

• --skip_test

  Flag to skip rendering the validation set.

• --select_camera_id

  Only render from a specific camera id.

• --target_path / -t

  Path to the target directory containing a motion sequence for reenactment.

NOTE: The below parameters will be read automatically from the model path, based on what was used for training. However, you may override them by providing them explicitly on the command line.

• --source_path / -s

  Path to the source directory containing a COLMAP or Synthetic NeRF data set.

• --eval

  Add this flag to use a MipNeRF360-style training/test split for evaluation.

• --resolution / -r

  Changes the resolution of the loaded images before training. If provided 1, 2, 4 or 8, uses original, 1/2, 1/4 or 1/8 resolution, respectively. For all other values, rescales the width to the given number while maintaining image aspect. 1 by default.

• --white_background / -w

  Add this flag to use white background instead of black (default), e.g., for evaluation of NeRF Synthetic dataset.

Novel-View Synthesis

Render the validation set:

SUBJECT=306

python render.py \
-m output/UNION10EMOEXP_${SUBJECT}_eval_600k \
--skip_train --skip_test

Self-Reenactment

Render the test set:

SUBJECT=306

python render.py \
-m output/UNION10EMOEXP_${SUBJECT}_eval_600k \
--skip_train --skip_val

Render the test set only in a front view:

SUBJECT=306

python render.py \
-m output/UNION10EMOEXP_${SUBJECT}_eval_600k \
--skip_train --skip_val \
--select_camera_id 8  # front view

Cross-Identity Reenactment

Cross-identity reenactment with the FREE sequence of TGT_SUBJECT:

SUBJECT=306  # the subject of a trained avatar
TGT_SUBJECT=218  # the subject of a target motion

python render.py \
-m output/UNION10EMOEXP_${SUBJECT}_eval_600k \
-t data/${TGT_SUBJECT}_FREE_v16_DS2-0.5x_lmkSTAR_teethV3_SMOOTH_offsetS_whiteBg_maskBelowLine \
--select_camera_id 8  # front view

Cross-identity reenactment with 10 prescribed motion sequences of TGT_SUBJECT:

SUBJECT=306  # the subject of a trained avatar
TGT_SUBJECT=218  # the subject of a target motion

python render.py \
-m output/UNION10EMOEXP_${SUBJECT}_eval_600k \
-t data/UNION10_${TGT_SUBJECT}_EMO1234EXP234589_v16_DS2-0.5x_lmkSTAR_teethV3_SMOOTH_offsetS_whiteBg_maskBelowLine \
--select_camera_id 8  # front view

FPS Benchmark

To benchmark rendering FPS directly with our demo avatar, run

SUBJECT=306

python fps_benchmark_demo.py --point_path media/306/point_cloud.ply \
--height 802 --width 550 --n_iter 500 --vis

To benchmark rendering FPS with the original dataset, run

SUBJECT=306

python fps_benchmark_dataset.py -m output/UNION10EMOEXP_${SUBJECT}_eval_600k \
--skip_val --skip_test --n_iter 500 --vis

NOTE: To avoid the influence of I/O, we only read the first view of each split and repeatedly render the same view for n_iter times.