ChatGPT Images 2.0: OpenAI Revolutionizes Image Generation
Le brief IA que les pros lisent chaque soir
Les 7 actus IA du jour, décryptées en 5 min. Gratuit.
Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.
Choisis ton rythme
Gratuit · Pas de spam · Désabonnement en 1 clic
OpenAI Unveils ChatGPT Images 2.0
OpenAI has officially launched ChatGPT Images 2.0, an enhanced version of its image generator, featuring the new artificial intelligence model gpt-image-2. This update addresses a major flaw from the previous version.
Users of ChatGPT and Codex can now access ChatGPT Images 2.0. However, advanced features are reserved for Plus, Pro, and Business subscribers. An Enterprise version is also planned. For mobile users, an app update is necessary to fully benefit from the new features.
The gpt-image-2 model is also available through OpenAI's API, allowing developers to integrate it into their own applications and services.
Notable Improvements
Previously, generating images containing text with ChatGPT was problematic, as the text was often distorted and incoherent. With ChatGPT Images 2.0, this limitation has been overcome. The model can now produce long, readable text that is correctly integrated into images.
Furthermore, the tool is no longer limited to English. It can generate images with text in multiple languages, displayed smoothly and accurately.
The image generation method has also evolved. Instead of requiring a separate command for each image, the tool now offers multiple variations in a single attempt, making it easier to create cohesive content. It supports a variety of formats, ranging from 3:1 to 1:3.
Other Improvements
The AI is now capable of analyzing complex requests before generating images. In some cases, it can even use web searches to refine the output.
ChatGPT Images 2.0 has a better understanding of different visual styles, whether they are realistic photos, cinematic scenes, pixel art, or manga. Each style is reproduced with greater consistency in terms of textures, lighting, and composition.
The tool has also gained in precision. It adheres more closely to instructions, respects requested details, and accurately reproduces elements that were often lacking before.
Persistent Limitations
Despite these improvements, some complex layouts can still pose challenges. Additionally, text in languages other than English is not always perfect.
However, the progress is notable. Compared to last year, the evolution is undeniable.
Brief IA — L'actualité IA en français
L'essentiel de l'actualité de l'intelligence artificielle, décrypté et expliqué chaque jour.