Our first idea for this project was to form movie sequence with produced images from text prompt Ai.
Since
people and Ai interpret things differently, it would be interesting to see how an Ai views famous movie
scenes in forms of text. We scrapped this as we did not have a reason on why we focused on movie
sequences
and felt that it may be too challenging to follow through in the next couple of weeks.
Still sticking with a text prompt Ai to see how a bot’s interpretation differs from ours, we decided to
get
it to interpret something that is not tangible. Inspired by Spotify Wrapped, an annual compilation of a
user’s data that many people look forward to at the end of each year, we wanted to come up with
something
that was personalised to an individual. An aura was a unique, non-tangible thing that represents a
person to
us and using an Ai to produce an outcome for someone to be able to see would be intriguing.
With a survey, we collected personal details from many individuals describing their physical and
personality
traits. These were used as the text prompts as we felt it embodies a person the most.
We did not want to just stop with the generated images from the Ai, so we took things further by seeing
how
much we could manipulate the data to produce different outcomes and place these outcomes together to
form
one single profile representing the person.
Various explorations were done to see what fits best and you can find out more in the descriptions
below!
Description
From Buzzfeed quizzes to personality tests, we realised people like finding out what represents them as
a
person and their nature. We decided to focus on people's aura as a medium to visualise what embodies a
person’s energy and innate nature.
What’s your aura? is a project that makes use of a text prompt Ai, Dream.Ai by Wombo, to create a series
of
personalised profiles with the images generated.
We chose to work with a text prompt Ai to understand how it would interpret a set of data and use this
produced outcome to embody something that isn’t tangible. These generated images are put through various
steps of manipulation where we explore different software and produce a result which each one. These
software include Adobe Illustrator, Photoshop, After Effects, Premiere Pro and Blender.
Findings & Insights
These auras are represented by a mixture of colours derived from their descriptions to represent each
individual. Different colours can evoke certain moods and emotions. For example, red is often associated
with passion while blue with mellowness. A person’s aura can be an interesting blend of various colours,
with each shade representing a different emotion.
There were some findings that we discovered throughout the course of this project. One of the findings
was
that while generating the images using the text prompt Ai, we realised that details describing physical
traits affect the composition in the image whereas the details on personality traits affect the colours.
We
also realised that the animated colour gradients produced for each individual will always show the
colour
pink no matter how different each palette was.
Another finding we found interesting was that two individuals that shared the same birthday, shared very
similar colour palettes as well. Although the two shapes are different, they both have the colours: red,
yellow, green, blue and purple on them.
Design Decisions
Since most of the design aspect, such as colour palette, was decided by the Ai, we had to use a neutral
colour so that it will match all the auras. A black background was chosen to bring out the colours in
the
aura and white text is used to compliment it.
The design of the profiles has went through several stages of modification. Initially the mock up we did
showed the details in smaller texts, positioned at the corners but we realised it did not felt like a
person’s profile. We wanted to highlight the details more so it was changed to a clearer font and the
text
sizes increased too. Next, we tried out the different arrangements of the profiles by experimenting with
the
font and arrangement of the elements
A whimsical font (BRANIC) was picked over a san serif as it gave the profile a less stiff look to it. A
san
serif (M PLUS 1) was used for the body text instead as to not overcomplicate the whole profile because
the
focus should still be on the outcomes.
Challenges
The first challenge we faced was while using the text prompt Ai as an Ai could really generate all sorts
of
images. When running our text prompts for the first time, the Ai produced a lot of realistic looking
images
when we were aiming for something more abstract. To solve this, we added the word ‘painting’ at the back
of
every prompt to get rid of the realistic looking images.
Blender was also a hurdle as it was a relatively new software for the both of us. With little
experience, we
had to find various tutorials to follow. None of the tutorials we found taught us how to create an
abstract
shape and adding a colour gradient on it, so we had to combine both tutorials for the outcome. To
overcome
this, we also seek help from classmates who were more familiar with Blender and assisted each other as
two
heads is always better than one right?
Considering the number of softwares we had to use to produce an outcome, it was tedious for the two of
us.
We shared a Google Drive folder to keep everything organised and labelled so when each step was
finished,
the other person could easily access and continue from there as well.
Feedbacks
At the beginning stages of the project, we shared on our worries about Blender as we were also unsure if
our
computers would be able to handle all the rendering. We thought of changing the profile to only show the
animated colour gradient, however, our friend told us that we should just not be scared and attempt
Blender
first as the 3D rendered model would definitely pay off. He was right! It was a scary move to us as
everything felt so uncertain but we were glad that we followed through. (Ps. Our computers could
handle
the rendering as well!)
When sharing the outcomes with the individuals collected data from, one of them shared that it looked
like
it could be a wallpaper. That was where the idea for our artefact came from: a downloadable file
containing
wallpapers for each individual. It was a nice idea as the everyone we have collected data from gets
something in return as well and that would make them more inclined to do the survey too.
Instead of just displaying the outcomes on the website, we were also advised to showcase more on the
process
of our steps too. Our lecturers thought it would be interesting to see certain steps clearer, like the
generated images from the Ai.
Achievements
We did great, yay! Especially considering the amount of effort the two of us put in for this project.
There were many things that we learnt as well, like positioning the camera on Blender and exporting a
transparent video. We also taught each other different things and shared skills along the way.
When presenting our project again in week 14, the feedbacks given were pleasant and we felt happy that
we
were able to produce such nice outcomes with just texts and datas. People who have their auras produced
also
liked the outcomes when shown. Overall, it was a very feel-good experience!
