Скачать или смотреть Google’s Imagen 4 vs OpenAI's GPT-4o

Google has released Imagen 4, its latest image generation model, with significant upgrades over previous versions. The model delivers more detailed visuals, improved text rendering, and better handling of prompts. Users report more realistic textures, clearer product photography, and enhanced control over character consistency and facial expressions.

Compared to Imagen 3, Imagen 4 shows better color accuracy and generates more believable fantasy scenes. Facial features and emotional expressions now respond more precisely to instructions.

Launched alongside Imagen 4 is Veo 3, a new video model capable of turning still images into short, lifelike clips. Examples include everyday scenarios such as someone eating ice cream or a street vendor at work.

In comparisons with GPT-4o, results varied. GPT-4o produced sharp, high-resolution images with a recognizable yellow tint, while Imagen 4 delivered stronger lighting and prompt fidelity, though occasionally with a cartoonish feel. Both models have their strengths depending on the creative goals.