We provide detailed instructions on how to construct the kinetic typography dataset. Due to IP licensing issues with the templates, we also provide source code for post-processing and video rendering. After downloading the original template files from the official webpage and following these instructions, you can easily create the kinetic typography dataset. Details are as follows.
Examples of our KineTy dataset. You can download them with this link.
To construct the dataset, please ensure you have Adobe After Effects installed, as our dataset is created using Adobe After Effects templates. Additionally, you will need other programs related to Adobe After Effects.
You can download the software from here: Adobe After Effect.
We use only 'nexrender-cli-win64.exe' from Nexrender Installation link. This is for Windows and we use version 1.46.6. This program enables the use of Adobe After Effects through programming. For more information about this, you can find it here: Nexrender
We will provide you with templates, but you can download more Adobe After Effects templates from the site below.
Write down the words you want to print out in the video to word.csv
Edit rendering information such as resolution and frame number in basic_rendering_info.json
. You can add or change the other elements in reference to Nexrender.
Record the below information in template_info.csv
.
id
: just a number of the templatetemplate_name
: template file namecomposition_name
: a composition name what you want to rendercomposition_text
: the composition name where the text layer islayer_text
: a text layer name which has a 'source text'
For using nexrender:
pip install ffmpeg
Then, set your path to dataset construction folder in render_set.py
:
# Path to your project folder
basedir = "~~~/dataset_construction"
Lastly, designate your path to nexrender software and aerender software in render.py
:
# Path to your nexrender-cli-win64.exe
nexrender_cli_path = "~~~\\nexrender-cli-win64.exe"
# Path to your Adobe After Effects aerender.exe
aerender_path = "~~~\\aerender.exe"
Prepare rendering information with each video and render the template:
python render_set.py # Save the rendering information
python render.py # Render the template into a video