Hire creatives › Ramy Saboungui › Image Generation 2.0
Artificial Intelligence · by Ramy Saboungui
Image Generation 2.0
Passing the Turing
We've gotten to time where AI has finally gotten efficient at image generation with minimal fallacies and plenty of accuracies.
I started this to explore the new image generator by OpenAi and to find out how consistent and accurate it has become. The results are quite surprising.
Consistency entails good prompting and defining creative styles but also active rules to be applied when generating. Surely everyone can create, but generators do have limitations and are still not that "creatively" smart, although to an untrained eye, they can appear to be extremely fascinating.
You want to make sure tokens used in your prompts do not exceed your allocation for a single prompt but on the other side, you need to make sure you're not using minimal tokens per prompt. You'll face hallucinations via minimal dataset reasoning that don't make sense to the generator through time.
Working consistency entails explaining the continuity of styles you'd like to see but also the avoidances. A set of numbered rules help. Adding text in-between speech marks helps work string logic and generators are able to apply them better and consistently.
Gallery
Tags
Ai · Image Generation · Creative
Like Ramy Saboungui’s work? Request them: post a brief and they’re invited to pitch on it first.
Request Ramy SabounguiMore by Ramy Saboungui
- Web3 DevelopmentTMWLWelcome to the Public Domain Era
- Market ResearchPerfume of the SummerTime to Code
- CopywritingLife on Planet EarthCarnivorous Care
- CreativeHands of the WorldTribute to Humanity
- Team ManagementThe Marshmallow Challenge Team Building through Creativity
- CreativeThere's Always Something NewNo Technology Needed
