https://deepmind.google/technologies/imagen-3/
Imagen 3
Imagen 3 is our highest quality text-to-image model, capable of generating images with even better detail, richer lighting and fewer distracting artifacts than our previous models.
deepmind.google
장소나 배경에 특화된 느낌.
데이터셋이 뭔지 딱 알겠네.
생성 실패가 많음.
퀄리티가 좋고 프롬프트가 정제되어 있어서 좋음.
