Save and Share:
I used Midjourney every day for a month, generated thousands of images, and here are my thoughts.
Midjourney is awesome, and I love it.
But it has some weak sides, which must be fixed to get it to another level.
So, here are things i don’t like about Midjourney, and I hope they will be addressed.
- Lack of Web UI where you can do all the things without Discord
- Lack of “Download all”, “Download User defined time period“, and “Download all search results” buttons in WebUI downloader
- Lack of inpainting feature (Dall-E and Stable Diffusion have it)
- Lack of outpainting feature (Dall-E and Stable Diffusion have it)
- Lack of partial repainting (Select The Detail/Area you don’t like and Neural Network will generate and inpaint different variants of selection contents into the same image)
- Lack of API
- Lack of plugins for Krita and Photoshop (with inpainting, outpainting, and repainting features).
- Lack of ability to choose fast/relaxed mode for specific requests (using the parameter, something like
- Lack of ability to repaint/remaster uploaded images in different styles while keeping original composition (a-la Stable Diffusion’s img2img mode)
- Lack of ability to script some repeating actions, i.e. I’d like to have the ability to “Upscale and then Remaster N times” with one click. It would be awesome if we had a feature in WebUI, which allows adding user-defined buttons with quick configurable actions/action chains.
- Lack of compensation for subscription time for downtime
- Small and unreliable queue. If something wrong happens with Midjourney when your image is generated or waits in the queue, it will be lost completely. Interrupted generations and queues are not restored after the restart of the service.
- Some images only show up in Discord but don’t show up in the web interface.