If you want to create images using Midjourney, you’re stuck using Discord.

Sort of.

Midjourney has not released an API, though they are slowly staggering towards one.  There’s undoubtedly a private internal API but for external public users, the technology is completely tied to Discord.  Authentication on midjourney.com is via Discord and the only way to interact with the technology is by submitting /imagine commands on Discord.

This inhibits some logical workflows.  For example, perhaps you want to work out a set of prompts and then batch-submit them so it can run overnight.  Perhaps you want to auto-upsize each image and save them in a directory along with a text file (or a database entry) describing the prompt used.

And just to state the obvious, Midjourney is the kind of technology that works best with volume.  Yes, if you play around with it, you’ll make some nice images just by feeding it fun creative prompts.  But if you’re trying to get a specific kind of image or look, you’ll need to iteratively generate and tune the results.

For example, here’s a typical workflow.  I have a friend who’s suffering from Lou Gehrig’s  disease and is in full-time managed care.  I had the idea to brighten his spirits by sending him funny images of someone who looks like him, and our other friends can comment and have a chat with him.

To make this happen, the steps are:

1. Find a phrase that describes my friend.  For example, “middle-aged Italian man with a friendly face and giant forearms”
2. Add “no mustache” and tweak the age otherwise every image looks like Mario meets Hulk Hogan.
3. Find a fun scene – for example “terrifically handsome grumpy Italian man with giant forearms taking a selfie with his fellow sailors aboard their World War II German U-boat, everyone laughing, sloshing beer”
4. Experiment with lighting, mood, art style, and dozens of other parameters.
5. And then eliminate all the gens where the hands are messed up or someone in the background is missing an eye or other AI artifacts

So you can easily generate 50 times to get what you want.  I mean, that’s what /relax is for.

Alas, to do this, you’re going to be typing all these commands into Discord.  Forget a Python script that generates all the variations you want neatly pipelining to an API.

Well, you don’t have to entirely forget it…

I came across a great article on Medium that takes an interesting approach.  It uses PyAutoGUI to build a bot that talks on your Discord and feeds it prompts from a text file.  I played with it and it works as advertised.  You create a prompts.txt file and then watch the bot type commands.

Just writing a bog-standard Discord bot won’t work because the Midjourney bot expects “/imagine” and then “tab” and then your prompt.  It’s tedious.

Unfortunately, since it’s a fake gui operator waiting for events, I haven’t quite got it modified to create those database entries.  In theory it should be possible to get the attachments and save them – work in progress!

Meanwhile, you might find these resources helpful:

• First, as general advice, create your own Discord server and add the Midjourney bot to it.  Otherwise if you use their official server your results are jammed in with everyone else’s and it’s a visual mess.
• All major styles and commands: This Github is an amazing resource.  Tons of lighting, style, mood, artist, perspective, etc. commands.
• Quick gist of all the major Github commands
• An incredible 50TB Google Sheet showing all artists and styles (OK, it’s not really 50TB but it’s huge)

• Author
• Recent Posts

raindog308

Dread Lord of LowEnd Content at LowEndBox

I'm raindog308, techno polymath and long-time LowEndTalk community Administrator. My technical interests include all things Unix, perl, python, golang, shell scripting, vintage operating systems such as MVS, and relational database systems such as Oracle, PostgreSQL, and MySQL.

I also enjoy using LEBs! I have an empire of little guys for VPNs, larger guys for apps, and big guys for databases and storage. When I'm not in front a screen I'm into German Shepherd dogs, quality knives, target shooting, theology, tabletop roleplaying games, and forest hiking.

I enjoy writing technical articles here on LowEndBox to help people get more out of their systems. You can find me on LowEndTalk @raindog308

Latest posts by raindog308 (see all)

• Crunchbits Discontinuing Popular Annual Plans – The Community Mourns! - November 20, 2024
• RackNerd’s Black Friday 2024: Bigger, Better, and Now in Dublin! - November 19, 2024
• It’s the Season of Giving and CharityHost Has Deals for You! - November 18, 2024