"Naughty" prompts

[waynemareci]

Wonder if you've given any thought to processing rejected prompts

I've seen this using words like "flesh" or "overdose". Right now Imagine() eventually times out and returns a null message. Can anything be done to catch this response?

[zcpua]

https://github.com/erictik/midjourney-api/blob/main/example/imagine-ws.ts#L17

Ws:true

[waynemareci]

I have Ws set to true

[waynemareci]

I failed to run client.init(); this seems to resolve my issue. Can you explain the purpose of client,init()? It's not used in Usage in the readme.md on the midjourney-api home page.

[zcpua]

Wait for Discord websocket to connect successfully

[LukeFerris]

Some interesting further behaviour on this. If you appeal successfully, it looks like that key word (blood in my example below) is then allowed going forwards - but with a slight difference in behaviour. Once the image is created, it seems to delete the original message and displays the result in a new message (see "original message was deleted" at the top of the image).

This thwarts the midjourney api code in ever linking up with the successful image (I suspect) which causes it to eventually return null / timeout.

I wonder whether this bit is solveable..