Integrated AI Case Study
Case Study
Office Desk
Designed with functionality, ergonomics, and productivity in mind, the new Desk allows for a personalized work experience:
-
Sit-to-Stand Function: Adjusting beyond 27" short and 45" tall, the height adjustability of the desk surface supports an unimaginably wide range of users.
-
Heat/Cool Desk Surface: Comfort at the tip of your fingers (literally), no matter the weather.
-
Powered Footrest: Working in tandem with the height of the desk surface and the adjustable chair, the desk system fits to each and every unique body.
Using AI within the Product Design process is no different than with traditional design tools when it comes to the phases of creative expansion and contraction. AI may be used at different point in the process, depending on the project at hand. When designing the Desk, artificial intelligence best served the project within the first half of the process. This method allowed for massive initial expansion in the form of brainstorming and ideating followed by another smaller cycle of divergence and convergence. These phases are most useful when they are allowed to overlap and inform one another.
Product Design Methodology
AI Order of Operations:
Expansion & Contraction
Product Development Cycle
MIDJOURNEY
Brainstorm
Generated from a high level language prompt, MidJourney kicks out high quality visuals aiming to inspire form and material choices. Not unlike a tournament bracket, only the most interesting and promising concepts and features move forward into the next phase.
SKETCHING PH.01
Steer and pace
After an initial round of AI generated inspiration, ideas may be selected and iterated upon in the form of rough sketching. This is the time to further explore overall product form as well as feature details.
VIZCOM KREA
Explore concepts
Vizcom and Krea are both powerful and fast programs that can iterate upon super rough sketching as well as polished renders and images. Partially fleshed out ideas can be expanded upon once more in this round of concepting.
SKETCHING PH.02
Refine and design
Once a true direction has been chosen, more sketching can be done in preparation for final design. Furthermore, a final sketch can be rendered in any of these AI programs for a proof of concept rendering.
CAD
Engineer and test
Following final sketching and proof of concept, the design will move into CAD where it will be modeled for final renders, and ultimately production.
In order to begin this process, and to cast a wide net, a high-level prompt was devised to feed into MidJourney for image generation. By keeping an initial prompt as general as possible, a more vast array of ideas and concepts may come to the surface rather than if a super specific image generation request was made right off the bat.
In this case, the prompt consisted of only a few component: A reference image, request for desk furniture, and strong variation in image output.
"https://s.mj.run/fcElyQJl1Bs with desk future - Variations (Strong) by @cnwrk (fast)"
When designing a new product using AI tools, it is important to remember that there are no rules. While our team began this specific project using MidJourney, we have found that it can also be helpful to begin by sketching the old fashioned way. It can also be incredibly productive to run these processes simultaneously, allowing them to inform and feed off of one another. There are no rules to break when these tools have never been used before.
MIDJOURNEY
Brainstorm
Defining A Prompt
Steer and Pace
Finding Direction
After certain features and forms begin to rise to the top of the pool of AI generated images, sketches begin to sculpt the direction forward. In this case, we are starting to explore a simple electric footrest form. Our goal here is not to flesh out perfect, polished sketches, but rather to create rough images that we can interate upon in another round of AI processing.
SKETCHING PH.01
Dialing in a sketch for production after several rounds of iteration in AI image generation tools is crucial for achieving the desired outcome. Each iteration allows for refinement and optimization, ensuring that the final sketch accurately represents the vision and meets production requirements. By fine-tuning details and addressing any inconsistencies, the final sketch becomes a reliable blueprint for executing the project efficiently. This iterative process enhances quality control, minimizes errors, and maximizes the potential for success, ultimately leading to a polished and professional end result that aligns with the initial creative concept.
Refine and Design
Detailed Design
SKETCHING PH.02
The sketches in this phase are intended to detail out our showcase functionality features; Sit-to-stand, climate control desktop, and electric footrest.
Krea and Vizcomare lightning fast rendering programs that allow for rapid iteration on rudimentary forms. Here we are exploring potential tweaks to form and finish using simple sketches paired with a more specific language prompt.
VIZCOM
KREA
Language prompts should become more intricate as the design process progresses. By seeking something specific and targeted, the images generated stay in line with the degree of creative divergence sought in each round. For example, the prompt aimed to generate lively, inspiring, and unconventional images, compared to the simple prompt used for MidJourney.
Exploring Concepts
Expansive Rendering
Krea, an AI program developed by Autodesk, is particularly well-suited for product design tasks that involve ideation, conceptualization, and generating innovative design concepts. Its strength lies in its ability to understand design intent and preferences, allowing it to assist designers in generating a wide range of creative ideas quickly.
Vizcom, an AI program developed by OpenAI, is particularly well-suited for product design tasks that involve visualizing and rendering design concepts. Its primary strengths lie in its ability to generate high-quality, photorealistic images based on textual descriptions or prompts provided by users.