Last week, Google He gave a good review of LLM’s offer (great language models) in Gemini. Free accounts said goodbye to the models 1.5 And welcome to 2.0 flash, 2.0 Flash Thinking (experimental)the Gems of Gemini and the function Deep Researchwhile the subscribers of Gemini Advanced They won the model 2.0 Pro (experimental). Among all these news, it went unnoticed that Gemini 2.0 Flash happened to have native generation of imagesbut with a very important difference against competitors such as Chatgptthat creates them through Dall-e: not only generates images from a prompt, but The user can also upload them and edit them. As if he were using Photoshopbut through natural language and asking him with written instructions what he wants at all times. It is not a still perfect capacity, but it does work reasonably well and puts the edition of images available to anyone, No need for any expensive and complicated software to use.
How to access Gemini’s conversational photoshop
Gemini 2.0 Flash (Image Generation) Experimental It is not available on the web or Gemini app, but through Google Ai Studio. Probably because Google is scattered with its first attempt to provide Gemini with image generative capabilities, which He went out a year ago with his suspension for being too Woke.
Google AI Studio is an artificial intelligence development platform that allows developers to create and train automatic learning models more easily and efficiently, but than Anyone with a Google account can use for free and take advantage of the access that gives to a long list of Google language models. All you have to do is select on the menu Modelin the left column, the GEMINI 2.0 Flash (Image Generation) option.
New Skill Unlocked: Gemini 2 Flash Model is Really Awesome at Remaining Watermarks in images! pic.twitter.com/6qik0FLFCV
– Deedy (@deedydas) March 15, 2025
This language model jumped to the headlines for their ability to Eliminate water marks in photographswhich evidently involves a problem for companies such as Shuttersock either Getty images. Replaces them with a brand Synthidtechnology developed by Google Deepmind which allows identifying images generated or modified by AI, thus replacing the original water mark with one of Edited with ia. It is a reliable method, since it is also possible to eliminate AI brands with AI tools.
What can you do in the images through written instructions
But for what is groundbreaking, it is to allow users to edit images as if they were using Photoshop, although through an intuitive natural language interface. Just asking it in writing, Gemini 2.0 Flash can Add objects, eliminate them, modify scenarios, change lighting, adjust the angles, approach or remove the image and perform other transformations, respecting the coherence of the represented world.
Photoshop is not outside the AI revolution, and Adobe He has been adding abilities of this type to his tools in recent months. The function Generative filling It allows to manipulate images through written instructions, but its use is not as natural as Gemini 2.0 Flash.
The results do not always offer the same qualitybut Gemini 2.0 Flash is a light model designed to respond quickly and consuming few resources. It is not as powerful as the full versionbut he is taking his first steps in the generation of images, and the expected thing is to see How their results improve in future iterations.
Difference between the generation of images in Gemini 2.0 Flash and other models
The difference between the generation of images of Gemini 2.0 Flash and other models such as chatgpt is that while In the latter an independent -based independent technology model is used (Dall-e in the case of OpenAi, which uses a principle of synthesis different from that of the LLM) to generate images, which are then shown to the user within the chat interface, Gemini 2.0 Flash is both the large language model and the IA image generator in a single system.
Openai advanced last year that GPT-4O was also able to generate images natively, but this capacity has not yet taken to the final product. Probably due to the high computational cost and the risks related to the security presented by the generation of images with AI.
Other aspects in which the generation of images of Gemini 2.0 Flash (image generation) are to maintain The consistency of the characters through successive images and the text representationaspects in which other models still have serious difficulties.