Ruby Goals

Introduction:

This was my first design project. For some reason in my head I wanted to create a centre diamond with rubies down the side of the shank. I had managed to pick up two 18ct rings from an auction site and thought I could make use of them by creating my own designs.

I started off by using Gemini, asking it for suggestions – if it had these rings, what would you create?

It gave me three design options, which I could then begin to grow on.

Contents:


The Donor stones


These are the original pieces which I was considering to have redesigned. The amount of 18ct gold could be recycled and the stones could be relieved from their settings.


Exploring Design styles


I was kind of hooked, but was realising the limitations of the Gemini renditions. I wasn’t able to find the words to quite get it to do what I wanted it to do. It would either put too many stones in, put extra lines where there shouldn’t be, it would create some pieces which had absolutely nothing to do with the design brief. I then began to notice the renditions starting to break, being put through the AI machine so many times. I gradually got to see that text which was initially legible and made sense in the beginning, began to produce spelling mistakes and created inconsistent grammar. I was beginning to see the limitations in which can be done.


Refining the design for limitations


Specs:

18ct Yellow gold ring, with three side stones on the shoulder shank. The centre stone is a diamond, which is a bezel set, teardrop apertures for the most light for the stone. The design was still playing with a detailed under gallery on the shoulders to allow light into the rubies, and the rubies would be set in a channel setting for security. The middle bezel set diamond would be low to the finger to help reduce being caught on things.

After many renditions, I realised I needed to know what kind of budget it could cost to have something like this created.


Costs and budgets



Simplifying the design


I realised that the costs of the design far surpass what the ring would be worth and my markup was beginning to look closer to a retail costs, despite me being able to provide majority of the materials for the project.

The main costs were in the design details: Having the custom teardrop apertures and the undercut shoulder shanks were causing the price to inflate beyond what I could imagine. I had no skill in CAD design, and didn’t really understand what that meant. I just knew it involved creating the design on a computer. I had no idea how to reduce the costs unless I simplified the aspects of the design which were causing the pricing to inflate.


Refining AI design aspects


As already discovered, when trying to add specialist requests to Gemini to alter specific parts of a design, it would not know what to do, or make elaborate changes to the design. I needed to find a way to try and overcome this issue to help streamline the amount of renders I have to make. This is not only because it does take time to create them, but my plan with Gemini is included as part of an offer on my phone contract. It eventually limits how many images can be produced and if I hit my limit, I have a timer to wait till I can continue.

I know AI doesn’t actually know what it is creating. It can give a very good guess, but it is limited because it doesn’t recognise what an object actually is. You would think this is a problem when designing such an intricate thing like a ring, but I needed a logical solution for a creative problem:

I began to highlight the parts which I wanted to change:

With this new knowledge that I could define certain aspects of a design by highlighting what needed to be fixed, I knew that I would be able to create more designs much more easily. I needed to remember that the technology needed to know where, how and what was changing before it could create the parameters to fix. I was now going to hopefully waste less renders on my limit and get less frustrated with the technology when it spits out the same image every time.


The Final Design


I had finally arrived at a design I loved, it was low profile, it had the correct amount of stone, a more cost effective gallery aperture and a safe stone bezel design. One last edit which I created was a sapphire rendition. This is mainly because I had originally settled on the ruby stone, but now as I have been able to understand a little bit about how to get prompts working well, I could now visualise what both options could be look like. I requested it to be made as if it was a drawing as well as a realistic version:

Querying CAD design