top of page

Independent Research Experiment
100 Days of Generative AI
16th Dec 2023 - 24th March 2024

I had been using generative AI since mid 2022, but with the almost daily advancements in the field, my fascination with AI kept on increasing.

To stay abreast with the field and be creative, I used AI for the recurring next 100 days. I posted my explorations with AI daily, which were not restricted to any tool, any design language, or any output format.

It was a way for me to be creative and have fun with AI without the pressure of achieving the best outcome or being original. I called this - 100 days of generative AI. 
 
P.S - Give the page a minute to load. 

Day 1 

16th Dec 2023

Tools: ComfyUI, SDXL Turbo

Day 2

17th Dec 2023 | ShiftingCurves

Tools: ComfyUI, Animate Diff

Day 3 

18th Dec 2023 | DoodlingParametricCities

Tools: Krea AI

Super fast one-second 1024x1024 image generation with SDXL Turbo, great for ideation and scrapping unwanted ideas. 

Transforming space with Animate Diff and LoRA models.  First experience with open source video models. 

Real-time editing in krea.ai by quick doodling to guide interior, exterior, and site planning. 

Day 4 

19th Dec 2023 | EnhancingParametricCities

Tools: Magnific AI

Day 5 

20th Dec 2023 | TweakingParametricCities

Tools: Photoshop AI

Day 6 

21th Dec 2023 | AnimatingParametricCities

Tools: Kaiber AI

Tested 8-10 upscaler from cutout.pro, upscale.media, picsart, vanceai and finally landed with Magnific AI. 

Adding context to images with Photoshop's generative AI fill. Quick and cost-effective tool to add design detail. 

While animate diff was good for video generation, it lacked precision, so shifted to Kaiber AI for design consistency. 

Day 7 

22nd Dec 2023 | VibrantDieties

Tools: ComfyUI, SDXL 4K Upscale

Day 8 

23rd Dec 2023 | MachinicOrganisms

Tools: ComfyUI, SDXL

Day 9 

24th Dec 2023 | DosaWorld

Tools: Midjourney

First exploration of generating and upscaling images to 4K inside ComfyUI. 

Exploring the previous upscaling workflow with LoRA's. 

Time when midjourney v6 launched and influence the output through the input image. 

Day 10 

25th Dec 2023 | Christmas

Tools: Midjourney, Magnific AI

Day 11 

26th Dec 2023 | IslandGlomes

Tools: Midjourney, Magnific AI

Day 12 

27th Dec 2023 | BubbleInstallations

Tools: Midjourney, Magnific AI

Tested how well MidJourney V6 can handle parametric and intricately detailed prompts. 

Pushing MidJourney further by uploading different images. 

Abstract interior silhouette plus floating bubbles exploration in MidJourney. 

Day 13 

28th Dec 2023 | OutGrown

Tools: Midjourney, Magnific AI

Day 14 

29th Dec 2023 | WhereToHangClothes?

Tools: Dall E, Magnific AI

Day 15 

30th Dec 2023 | AiTravelDiaries

Tools: Kaiber AI

Architecture that is overgrown and untouched. 

Inspired by dhobi ghats, these images depict architecture that is overwhelmed with clothes, everywhere.

Animated real images of my year end trip to see how AI can be used in our daily lives. 

Day 16 

31st Dec 2023 | 2024

Tools: Midjourney, Magnific AI

Day 17 

1st Jan 2024 | tambak(Tobacco)

Tools: ComfyUI, Magnific AI

A new year post to burn the old and be ready to embrace the new. 

Spaces dedicated to spitting tobacco! An all-you-want spitting zone with no one to stop or complain. 

Day 18 

2nd Jan 2024 | Operas

Tools: ComfyUI, ControlNet

Imagining the Sydney Opera House in different materials, contexts, and designs with controlNet inside ComfyUI. 

Day 19 

3rd Jan 2024 | LivingStructures

Tools: DallE, Magnific AI

Day 20 

4th Jan 2024 | WahTaj

Tools: ComfyUI, Masking

Day 21 

5th Jan 2024 | ComfyShop

Tools: ComfyUI + Photoshop

When architecture gets acquired by living organism. 

What if the dome of the Taj Mahal was designed by different architects? Explored with LoRA's and masking in ComfyUI. 

Controlling the gen AI output of ComfyUI from Photoshop with real-time doodling. 

Day 22 

6th Jan 2024 | Crispy

Tools: Midjourney, Magnific AI

Day 23 

7th Jan 2024 | AiPlots

Tools: ComfyUI, SDXL

Day 24 

8th Jan 2024 | MagicalTemples

Tools: DallE, Magnific AI

What if we had a restaurant themed around fried chicken? Crispy details!

Tested various checkpoint models, LoRA’s, and samplers inside ComfyUI by creating a plotting script for choosing the right model. 

Fairy tale workship places inspired by Indian temples, mosques, and their ornamentation. 

Day 25 

9th Jan 2024 | AssortedIndianArchElements

Tools: Midjourney, Magnific AI

Day 26 

10th Jan 2024 | IllustriveSpaces

Tools: ChatGPT, Runway, Pika, Magnific

Day 27 

11th Jan 2024 | RawCurved

Tools: ComfyUI

Indian architectural elements through the lens of gen AI models. 

Animating different parts of a central atrium in 4k-upscaled images. The process was more manual by stitching parts of the video.  

Concrete facades that have been aged. 

Day 28 

12th Jan 2024 | AroundWells

Tools: Midjourney, Magnific AI

Day 29 

13th Jan 2024 | MorphingFacades

Tools: Midjourney, RunwayML

Day 30

14th Jan 2024 | FallenKites

Tools: ComfyUI, Magnific AI

Imagining cities surrounded around wells to combat the water shortage and climate crisis. 

Morphing facades with masking. 

Expressing the kite-flying festival in India and converting it into architectural spaces. 

Day 31 

15th Jan 2024 | MonoTypologies

Tools: ComfyUI, Magnific AI

Day 32 

16th Jan 2024 | AiFails

Tools: Midjourney, ComfyUI

Day 33

17th Jan 2024 | ConsistentStyle

Tools: ComfyUI, Magnific AI

Single prompt to multiple design options of the same architectural typologies maintaining the same color scheme. 

AI fails from the past month of gen AI explorations. 

Working with Style Aligned custom nodes to create consistent and controlled interior spaces to tackle the inconsistency of gen AI. 

Day 34 

18th Jan 2024 | ToonSplash

Tools: ComfyUI, Magnific AI

Day 35 

19th Jan 2024 | JoyfullInteriors

Tools: ComfyUI, RunwayML

Day 36

20th Jan 2024 | StretchingConcrete

Tools: ComfyUI

Infusing a cityscape with whimsical and playful cartoon elements using comfyui to bring a lighthearted touch.

Whimsical and playful interior spaces animated using the motion brushes feature of RunwayML. 

Moving frames in different directions using a guiding image with the help of a control-net node within comfyUI.

Day 37 

21st Jan 2024 | BlingBling

Tools: Midjourney, Magnific AI

Day 38 

12nd Jan 2024 | mouldedFacades

Tools: ComfyUI

Day 39

23rd Jan 2024 | JyotiBuildingSchemes

Tools: ComfyUI, ControlNet, Masking

Speculative automobiles in different materials and contexts

Design options from the stretching concrete workflow. 

Streamlining the retrofitting process using controlNet and LoRA for idea exploration, better communication, and cost estimation.