README

Solution Description

This is a Python-based solution that processes customer-provided text and image inputs to convert them into short-form video formats. It generates an AI summary from the provided text content, an audio track from the summary, and finally produces a video file (.mp4) that is a concatenation of the provided images with some visual effects and the audio track as voiceover.

This solution is still under development and should be considered experimental.

Colab

Try Article to Video in Colab!

Requirements

• Google Cloud project with Google Cloud Storage, Vertex AI, and Text-to-Speech enabled APIs.
• Valid article + related images to convert to a video.

Environment setup

• Install the requirement.txt libraries.
• Install the gCloud CLI Link
• Set up GCP default client credentials Link
• Install the ffmpeg CLI: Link
• Install ImageMagick – Mac users can use homebrew: homebrew install imagemagick Link
• Update any placeholder values in config.yml

Execution

Standalone (CLI)

usage: video_generator_execution.py [-h] [--config CONFIG] [--video_id VIDEO_ID] --article_path ARTICLE_PATH --image_dir IMAGE_DIR [--disable_text_overlays]
                                    [--multi_text] [--splash_image SPLASH_IMAGE] [--burn_in_subtitles] [--language {en-US,en-GB,fr-FR,de-DE,es-ES,pt-BR}] [--step {audio,storyboard,video}] [--debug]

options:
  -h, --help            show this help message and exit
  --config CONFIG, -c CONFIG
                        Path to the configuration YAML file. Defaults to 'config.yml'.
  --video_id VIDEO_ID   Unique ID for the generated video. If not provided, a UUID will be generated.
  --article_path ARTICLE_PATH, -a ARTICLE_PATH
                        Path to the article text file.
  --image_dir IMAGE_DIR, -i IMAGE_DIR
                        Path to the directory containing images.
  --multi_text MULTI_TEXT
                        Specify that the LLM is to summarize multiple articles, not just 1.
                        Changes prompt accordingly.
  --splash_image SPLASH_IMAGE
                        Specifies the file name of the splash image (do not
                        include filename extension)
                        indicating the image to optionally use at the beginning
                        of the video
  --disable_text_overlays
                        Disables text overlay generation.
  --burn_in_subtitles   When provided, subtitles will be burned into the content.
  --language {en-US,en-GB,fr-FR,de-DE,es-ES,pt-BR}, -l {en-US,en-GB,fr-FR,de-DE,es-ES,pt-BR}
                        The language for the output video. Defaults to 'en-US'.
  --step {audio,storyboard,video}
                        Run a discrete step in the video generation flow.
  --debug, -d           Enable debug logging.

To generate a video:

1. Save article contents a text file.
2. Put images into one directory (supported extensions are .jpg, .jpeg, .png, .gif, .bmp)
3. Provide a human readable video ID (you can also omit this and a UUID will be generated for the video instead).
4. Run:

python video_generator_execution.py --article_path path/to/article.txt --image_dir path/to/images/dir --video_id id_for_your_video

The video and any other generated assets will be available in the output directory (configured in config.yml) under a subfolder named for the provided video ID.

To make edits without regenerating totally from scratch, you can make small edits to things like timing, animations, and text overlay content in the storyboard.json file and then regenerate the video based on the updated storyboard using the --step video option with the CLI.

Local Docker

[sudo] docker build -t gtech/video-generation-from-images .
[sudo] docker run -e PORT=5000 -p 5000:5000 -t gtech/video-generation-from-images

The docker image with build and deploy the UI, but the UI is still experimental and will not work correctly at this time. However, the CLI can be accessed with the docker by running the following command (you'll also need to configure your Google Cloud ADC):

docker exec -it <CONTAINER_ID> /bin/bash

Local Server

cd ui
ng build
cd ..
python3 -m flask --app video_generation_from_articles run

Cloud Deploy

gcloud builds submit . --substitutions _IMAGE_NAME=atv,_SERVICE_NAME=atv,REPO_NAME=test-repo

Running tests

From the base folder, run

python3 -m unittest discover -p "*_test.py"

Upgrading Dependencies

pip-compile is used to generate the requirements.txt file. To update dependencies, make changes to requirements.in and then use the following command:

pip-compile requirements.in --generate-hashes --upgrade

Disclaimer:

Copyright 2025 Google LLC. This solution, including any related sample code or data, is made available on an “as is,” “as available,” and “with all faults” basis, solely for illustrative purposes, and without warranty or representation of any kind. This solution is experimental, unsupported and provided solely for your convenience. Your use of it is subject to your agreements with Google, as applicable, and may constitute a beta feature as defined under those agreements. To the extent that you make any data available to Google in connection with your use of the solution, you represent and warrant that you have all necessary and appropriate rights, consents and permissions to permit Google to use and process that data. By using any portion of this solution, you acknowledge, assume and accept all risks, known and unknown, associated with its usage, including with respect to your deployment of any portion of this solution in your systems, or usage in connection with your business, if at all.