about

Dear Reader,

Thank you for taking the time to look through my work. It means a lot to me.
Below you will find these selected works:

1. I knew that if I walked in your footsteps, it would become a ritual
2. Ancestral Apparitions
3. After Goya
4. Kolam Series
5. After Image
6. Cloud9.garden
7. Recollections in Haldi & Light

In the top right corner you will also find a link to an about page with a short bio and additional links (i.e. CV, blog, Instagram).
If you are looking for my web development/design work, you may want to check out aarati.services

Sending you my best,

Aarati Akkapeddi

1. I knew that if I walked in your footsteps, it would become a ritual

May 2021 - Ongoing, Code, Video

In this project, I am thinking about memory as a generative practice. I'm interested in retracing the past through talking about family photos, but in a way that acknowledges that our understanding of the past is always an approximation. Names are forgotten, moments are conflated with other ones. Collective and individual memory is something that mutates over time and over generations. And I think of this as less of a fault and more of a healing fluidity that allows us to process the past in a way that's enmeshed with the context of the present.

This idea of approximation and mutation is something that has drawn me to use machine learning in this work. Specifically, I have been using family photographs as training data for a Generative Adversarial Network, which is a type of machine learning model that in this case, tries to look at the images you show it and create its own approximations of what it sees as best it can.

In this work, I've also been interviewing family members about specific photographs and incorporating those interviews back into the work itself through animation and video.

You can read more about my process in this blog post.

A special thanks to Ada X for it was during my residency with them that I developed this work.

animated still from video

generated images of my mother

generated image of my mother

animated still from video

2. Ancestral Apparitions

September 2020, Code, Lumen Print, Video

Ancestral Apparitions uses machine learning to evoke simultaneous feelings of familiarity and distance in a meditation on diasporic identity and more broadly, on the pendulum between collective and individual memory. Thus far, a series of 35 small photographs have been made. For these images, a Generative Adversarial Network is trained on my own family photographs and images from The Studies in Tamil Studio Archives and Society. The ML model is then able to produce its own ghostly "family photographs" which are used to create digital negatives and print images on expired photo paper, returning the images to the original medium of the training data.

exhibition view

a distorted generated photograph of a man

single generated photograph (appears to be a couple)

a distorted generated photograph of three men standing

a distorted generated photograph of a man

3. After Goya

January 2020, Code, Intaglio

This work was created with the help of El Museo Goya Fundación Ibercaja as part of a residency with Etopia Center for Art & Technology in Zaragoza, Spain. A generative adversarial network is trained on Goya's etchings from Los Caprichos, Los Desastres de la Guerra, & Los Disparates. The outputs are used to then laser etch plates and create intaglio prints. In training a machine learning model on Goya’s etchings, I was interested in generating images that would still reflect an aesthetic essence from the originals. While the generated images are quite abstract, features such the use of chiaroscuro and the compositional structure still carry through.

etching

gallery view (photo taken by Julian Fallas)

comparison with original etchings

4. Kolam Series

code, embossing on cyanotype, 2020, ongoing experiments.

Based on typical South Indian Pulli (dot) Kolam patterns, kolam designs were produced based on facial recognition data from family photographs. I took one photograph of my maternal grandmother, one of my mother and one of my paternal grandmother and translated each number in each corresponding facial recognition vector into a component of the kolam pattern, producing three distinct kolam designs. I then embossed these patterns onto cyanotype prints of the corresponding photographs.

Detail photograph of embossing on an image of my paternal grandmother

my maternal grandmother

In addition to working with facial recognition data, I have also been developing systems of translating text to Pulli Kolams using binary codes as an intermediary step. Below you will see some examples of this as well and diagrams explaining the translation process.

translating my late uncle's nickname into a kolam pattern

Sunprinting this pattern onto a Paan leaf to make an ephemeral memorial.

An example of an alternative spiralized translation method.

5. After Image

Code, Digital Print, Wood, Video, 2019.

After Image is an art installation showcasing a series of experiments done with an archive of my own family photographs in relation to a larger archive of South Indian studio photographs from the Studies in Tamil Studio Archives and Society. Each of these experiments uses machine learning and computational techniques to sort, average, and analyze the images in order to surface semantic and visual patterns across the hundreds of images. With my experiments, I performatively question the notion of collective and individual identity, and highlight the complexity of the image as a data point. The images were first grouped by subject matter using a subject detection machine learning model and a clustering algorithm. The groupings created by subject matter are represented around the central video component in the form of an intricate mapping of images. Each group is represented by a cluster of its individual images around a central node image, which is a computational averaging of all the images within the cluster. I also created risograph prints of these averaging images

On a second layer above the images, representations of "machine vision" are overlaid onto the individual images in each cluster. These representations show lines indicating pose, circles indicating areas of high exposure and contrast, and finally rectangular boxes indicating objects that were detected. Boundaries between clusters are represented by white lines that can be traced if one chooses In the center on loop, a video shows me placing my own family photographs centrally, triggering the flashing of similar images from the S.T.A.R.S. archive on the left, and on the right, their corresponding machine overlays indicating pose and object detection.

If you would like to read more about the specific findings and methodology of this project, please feel free to read my paper.

top view

Risograph Print of averaged photos of couples where one person is sitting and one is standing

Risograph Print of averaged photos of couples where both people are standing

6. Cloud9.garden (*Group Project)

Website, 2020.

The CLOUD9 Memorial Garden is a space for collective remembrance with care, gentleness and respect. Pandemic times limit how we can gather to mourn and remember. During periods of mass loss and isolation, creating intentional spaces to enact and celebrate collective memory is vital. Our community continues to plant new seeds for all of ours that we have lost in this time, due to state violence, due to Covid and due to the continued systems centered on harming Black people, Indigenous people, brown people, low income people, trans people, undocumented people, and marginalized people. We invite you to transform personal grief into collective healing, growth and liberation.

The Cloud9.garden virtual memorial is a collaborative effort between myself, BUFU, Chiara Marcial Martinez, Zai Aliyu, and Melanie Hoff.
*I was primarily responsible for developing a system by which unique memorial flowers could be generated for each dedication. This flower system uses p5js. You can visit the garden at cloud9.garden.

flower generative design iterations

screenshot of dedication submission form

7. Recollections in Haldi & Light

September 2020 - Code, Turmeric Anthotype, Lumen Print, Blind Embossing

Hand opening a photo album of lumen print silhouettes

Recollections in Haldi & Light uses a pre-trained machine learning depth detection model to extract silhouettes from the artist's family photographs. These silhouettes are then used to produce digital negatives and printed on expired photo paper and turmeric anthotypes. The silhouettes are also used to 3D print embossing plates to create blind embossing prints. The prints are meant to evoke a sense of intergenerational memory and loss. The use of the turmeric anthotype brings a ritual dimension to the work as turmeric is considered an auspicious material in Hinduism and is also a familiar element in South Asian cuisine and culture.

turmeric print: uncles sitting

lumen print: my grandparents and my mother and her siblings

exhibition view

lumen print: group photo

turmeric print: my uncle as a toddler