Midjourney Remix Mode + Vary Region: ever closer to Photoshop

Midjourney is constantly evolving and its developers are constantly adding new options to facilitate the creation of images. For a few months now, users of generative intelligence have been able to partially rewrite a prompt to transform an image by adding one or more details while retaining a large part of the previously generated image. With Vary (Region), Midjourney now allows modifying an image locally after selecting an area in the image. What fine-tune his “generations” without tearing capillaries during the drafting of prompts to extend, and with sometimes random results. We explain how to proceed with a paid version of Midjourney.

Enable Remix Mode

Once connected to the Midjourney bot in Discord, you need to access the software settings. To do this, simply enter /settings to display the different options. All you have to do is click on the Remix Mode option to access the retouching of quick.

Request a first generation image

Once you have your idea in mind, you must transcribe it in writing and in English in the prompt via the /imagine command. To save time, you can also type the first few characters and complete the command by pressing the tab key. For our test, we asked Midjourney to generate a realistic photo of a woman holding a phone in her right hand. Admittedly, the results are rather random since three out of four images show a woman holding a smartphone with both hands…

Exploit the Vary (Region) option

Once you have made your choice, you must select the image to be generated in large format with the U button (1,2,3 or 4). Again, Midjourney offers the Vary (Region) option. The software then opens a new window in which you can select the area to be modified. In our example, we are going to select the smartphone to replace it with a bouquet of flowers (finally a bouquet of white tulips).

We select a rectangular area around the woman’s hands (it is possible to freely outline the part of the image to be modified with the lasso). Attention, it is not possible to modify the size of the selection once it has been made. You can then edit or add details to your prompt.

We repeated the operation by asking Midjourney to add a cat in his hands, and again the results are quite convincing.

Finally, we also offer the version with the bouquet of flowers.