I'm excited to share with you the process of how I created this animation: A 3D key for LoveKravt to be given as away as an NFT.

The final result
Here is how I made it:
The idea of the video was very clear from the start: Making a NFT Key that would give access to higher ranks within the LOVEKRAVT community. A token given by the Digital Blacksmith (me).
That was clear, but the art style was not. Usually I would use boards in Pinterest to come up with a style, but we live in the future, now we have text to image!

Comparison of AI Text2Image

Using Midjourney, Stable Diffusion and DALLE 2, I spent hours creating dozens of images, carefully curating the elements I needed.

I tried prompts alone at first, but the style was just not very attractive.


Dall-e 2 was underwhelming at first, but slowly I started to realize that this part of the process was just as important as finding what I was looking for. These images help me realize quickly the way I did NOT want to go.It’s very similar to having a concept artist staying working late in order to finish all these different concept art pieces to then have me rejecting all of them in the morning. The big plus here is that there are no feelings to crush! 
Stable Diffusion

Oh, Stable Diffusion, you beautiful beast with deformed hands.

This fantastic tool (or set of tools) helped me further explore the idea of a key with a key chain. I even tried things like adding a box to hold the key while I was in the search for something that would look like it is a collectible.
Eventually I saw that none of this was looking outstanding so I kept on searching and testing.


With midjourney I decided to stop wasting time trying to make the AI understand me through words (Prompting). Instead, I very quickly modelled what I had in mind inside Blender. I rendered that and I fed it to Midjourney as an image reference. This was in November 2022, so this option in Midjourney was brand new.

My First 3D Render

Midjourney’s Output

This right here was my “Eureka” moment. This is when I realized this AI assistant can improve on my simple ideas and make me see the desired final result in a matter or seconds. 

I continued giving 3D renders to Midjourney and it quickly became a feedback loop in which I would model and render, the AI would improve it, then I would return to my 3D model to improve further.
My second 3D Render

My Third 3D render

Midjourney’s Output

Midjourney’s Output
That is how I learned to stop worrying and love the AI Image2Image algorithms. 

I know what you're thinking: "AI is taking our jobs!" But don't worry, there's still plenty of room for human creativity and ingenuity in this process. After all, the idea is mine and I spent the hours curating the images. This process basically just replaced the Pinterest picture-finding process.

A great plus of this workflow is that I was able to see the outcome of each idea I had in a matter of seconds, and then quickly steer away from what I didn’t like.

On top of that, looking at the awesome quality of those images, it made push myself to reach that level with my 3D skills.

I’m super proud of the final piece! So, whether you're a fan of LOVEKRAVT or just appreciate the power of AI, I hope you enjoy my collab with AI. Thanks for watching!
Amsterdam, NL