ChatGPT image generation has some competition as Midjourney releases V7 Alpha
Table of Contents
ChatGPT recently introduced its latest 4.0 image generation technology that took the Internet by storm, with users creating everything from cartoon recreations of family photos to lots of memes. The new model was, in fact, so popular that it became too much even for OpenAI's own good, as their GPUs couldn’t handle it, and CEO Sam Altman had to announce a temporary rate limit. However, OpenAI isn't the only one bringing major updates to its image generation technology.
MidJourney, one of the most popular image generation models, just announced the alpha test phase of its first new AI image model in nearly a year, V7, which is much smarter with both text and image prompts. V7 is also MidJourney's first model to have personalization switched on by default, meaning that to use it, you'll first need to rate around 200 images to build a MidJourney “personalization” profile, which will then tune the model to your individual visual preferences.
New features, yet no mention of improved text generation
V7 is available in two different modes, Turbo and Relax, and also comes with a new tool called Draft Mode, which is described as “the best way ever to iterate on ideas.” Draft Mode renders images at 10x the speed and half the cost of the standard mode. While the images are of lower quality, it allows you to brainstorm and edit using even voice commands. The bad news, however, is that text generation may still be an issue for V7 as well, as Midjourney fails to show it off.
Text generation has been one of the biggest concerns with AI image generation, but with ChatGPT's latest image generation technology, we finally saw text being rendered accurately and seamlessly integrated within images, eliminating much of the gibberish we had before. While ChatGPT 4.0 image generation still struggles with certain cases like rendering non-Latin languages, it is still a major breakthrough and something MidJourney still hasn’t properly addressed in its posts – leaving some people worried that the tech trails behind ChatGPT.
Updates are rolling out over 60 days
That said, MidJourney has announced that it will be bringing new features and updates to V7 Alpha every week or two for the next 60 days, with the biggest one featuring a new V7 character and object reference system. This is quite noteworthy, especially since V7 currently lacks a few standard MidJourney features like image upscaling and retexturing, so expect those to arrive in the coming weeks.
“Please Note: This is an entirely new model with unique strengths and probably a few weaknesses, we want to learn from you what it’s good and bad at but definitely keep in mind it may require different styles of prompting. So play around a bit.”
Midjourney CEO David Holz via Midjourney Discord Server
About the Author