.svg)
1,400 SKUs, one visual identity, run by their team
Rixus had already tried AI and it had failed. I was not brought in to make the pictures. I was brought in to set the standard they had to hit, architect the system that makes them, and train their in house team to run it without me.
Make 1,400 products look like one brand, on a real budget
Rixus makes consumer electronics accessories: chargers, power banks, cables, hubs and cases. Around 1,400 SKUs, sold for years B2B through distributors. In 2026 a new Amsterdam team set out to refresh the brand, launch a Shopify store and move to a combined B2B and D2C model.
Shooting or hand rendering 1,400 SKUs across studio, lifestyle and infographic formats was never going to happen. They had already tried AI in house on ComfyUI and Stable Diffusion, and it produced products that melted, lost hardware accuracy and lit inconsistently. Too fragile to batch.
The problem was never the model. It was that nothing was being held still.
Stop guessing. Start constraining.
Every attempt before mine treated the model as the variable. Write a better prompt, roll again, hope. That is a slot machine, not a pipeline.
So I did not hand them pictures. I handed them a method, a standard to hit and a look to hit it in, then trained the person who would run it. Minor Photoshop retouching still tidies some compositions, and the playbook says so rather than hiding it.
The nodes that never move
Close ups of the locked blocks as they sit in Figma Weave.
Sample Rixus workflows
The style and studio nodes stay fixed. The product is what changes.
Same rig, different product
Studio output from one locked lighting and environment setup, and lifestyle output from the same system, both built off the plain white background shots Rixus already had. Some compositions still took minor Photoshop retouching before release, which is in the handover documentation that was provided to them.
I taught it rather than kept it
Three two hour workshops with build sprints in between, and a written production playbook at the end. Their marketing lead and senior designer build and run it now. That was the brief. The engagement was knowledge transfer, not production.