Testing Gemini 2.0 Flash text-image-to-image capability

This article shows several tests on the new Gemini 2.0 Flash Image model.

The original image

This image (randomly generated by OpenAi Dalle3) is used in our test.

1) Remove objects in the image

remove all the clouds in the sky

2) Adding object to the image

add a small half moon in the sky

3) Change the color theme of the image

change the image to a night scene

4) Follow the same theme to create a new image

follow the same theme and color palette, create an image of a prairie

Reference

https://ai.google.dev/gemini-api/docs/image-generation#gemini-image-editing

AI Summary AI Summary
gpt-4o-2024-08-06 2025-04-17 01:07:56
The blog post tests the Gemini 2.0 Flash Image model's capabilities by performing tasks like removing objects from images, adding new elements, changing color themes, and generating new images based on a similar theme. These tests illustrate the model's text-to-image processing proficiency.
Chrome On-device AI 2025-05-01 01:26:23
Share Share this Post