This software allows for drawings to be made and played back in a time-based format. Additionally, it contains tools to create excerpts of the drawings and browse these segments by annotation.
This software was developed in light of a research project in which we used it in an interview setting. Asking speakers not just to speak, but to draw the entities they mention and the relations between them. This adds a visual layer on top of the audio.
For more information on the rationale behind developing this tool, see the [citation](#citation) section.
These instructions will give you a copy of the project up and running on your local machine for development and testing purposes. See deployment for notes on deploying the project on a live system.
### Prerequisites
The easiest way to run svganim is with docker-compose. Alternatively you can choose set up your own python environment. If you do so, ffmpeg is required.
### Installation
*`git clone` this repository.
*`cd` into the folder and run `docker-compose build`
*`docker-compose up -d`
* Docker should now expose the service at port 7890. Edit docker-compose.yml to change this.
* rerun `docker-compose build` to rebuild the image
* rerun `docker-compose up -d` to restart the container
### Deployment
The server is best exposed through a proxy. An [example Apache configuration file](./diagramming-apache-site.example.conf) is included.
## Usage
When the service runs, open it using your browser. The first page shows three options:
* Tags -- Manage a set of tags for annotating.
* Drawings -- All recorded drawings to be played back and annotated (or removed).
* Draw -- start drawing a diagram.
### Drawing
* By default, the drawing page gives four colour options. Select the desired colour on the left.
* The big 'Fullscreen' button puts the canvas in fullscreen. To exit, use the escape key.
* The drawing will be recorded from the moment the pen touches the canvas.
* When you stop drawing (ie. close the tab or browser) you can recover the drawing by going to 'Drawings' and select 'Draw' at the drawing you want to continue.
* **NOTE** Please note that when continuing a drawing, the intermediate time _is not captured_. Ie. the last stroke of the first drawing, will be immediately followed by the first stroke from the second drawing.
* A pen display such as a Wacom or Huion is recommended but not at all required.
### Annotating
* Before annotating, make sure to configure your initial tags on the tags page (`URL/index`)
* [optional] Add audio
* add an audio recording of the conversation to `files/audio`. Note that a wav file can be quite slow as everything goes through a browser. So possibly convert wav to mp3 or ogg.
* Open the drawing in the interface URL/annotate.html. Hover the title on the top left, and select the audio file from the pull down.
* Use the _Offset_ parameter to align the audio to the drawing (number in seconds).
With the tags interface you can manage a nested set of tags and manage the annotated segments. In our workflow, we had annotated the interviews, but tweaked the terms to create interesting subsets of drawings.
* To create a nested tag, select a tag, and an 'add tag' option will appear.
* Double-click the title to rename.
* Use the bucket icon to remove a tag.
* Click the coloured square to tweak the colour.
* In the overview of excerpts, use the checkboxes to change the tags in batch.
The tag index contains a download button which creates a zip file with all the required files for an embeddable player.
There is however no integrated a way to export the excerpts to a video file. The easiest way to do so however is to make a screen capture of a part of your display.
E.g. on Linux: set the pulse default input device to the monitor of the speakers. Then use wf-recorder: `wf-recorder -g"$(slurp)" -a -f recording.mp4`
### Notes
The `parse_offsets.py` script can be used to pad the diagram in order to sync it with the audio. This is necessary eg. after a network failure. It works by adding a line with the required offset to the `.json_appendable`-file. It dumps the output to the terminal. So use it with a redirect. E.g. `python parse_offsets.py -i faulty_file.json_appendable > new_file.json_appendable`.
This software uses (and for ease of use, redistributes) the [noUiSlider](https://refreshless.com/nouislider/) and [wnumb](https://refreshless.com/wnumb/) javascript libraries, developed by [Léon Gersen](https://refreshless.com/). Also using the MIT License.
