You Can Now Upload and Edit Images on Gemini

Google is finally bringing the ability to edit images on Gemini via text prompts. You can now upload existing images and edit them on Gemini through text conversations. In addition, you can generate AI images using Gemini and reiterate to modify the image. The image editing feature in Gemini is rolling out gradually to all users, starting today.

Google was the first to showcase native image editing using the Gemini 2.0 Flash model on AI Studio. I tested the model in February, and was blown away by how intuitive it was to modify images conversationally. In response to Google’s native image generation model, OpenAI also introduced native image editing and generation on ChatGPT.

It led to the viral Ghibli trend that skyrocketed ChatGPT’s usage. Now, Google has brought the native image editing capability to its consumer-facing Gemini app. Note that you can edit images on both the Gemini mobile app and the web.

Since I have compared native image editing between ChatGPT and Gemini, I can say that Gemini is much better at keeping the character/scene consistent across generations. ChatGPT alters the overall image after each generation, whereas Gemini is highly consistent. However, for image generation, ChatGPT outperforms Gemini.

Now we need to see whether Google is still using its Diffusion-based Imagen 3 model for image generation or using the native image generation capability of Gemini models. By the way, OpenAI is using the GPT-4o AI model for both image generation and editing.

Apart from that, Google says that an invisible SynthID digital watermark is embedded in all images, generated or edited using Gemini. Moreover, Google will also be adding a visible watermark on all images generated using Gemini.