
The problem
Producing a first image with AI can be straightforward. Changing one detail while keeping everything else intact is often much harder. That gap becomes obvious when an illustration has already been approved and only a small correction is needed.
What I built
I have worked on OIM through practical image-creation and editing tasks: producing visual assets, refining details, comparing results, and working toward more dependable changes to existing artwork. The work includes learning from unsuccessful edits rather than disguising them as successful results.
The difference
The useful challenge is no longer just “Can AI generate this?” It is “Can the result be corrected, controlled, and used?” OIM puts that real creative-work requirement at the center of the work.