Midjourney — Check the new command for image → describe

On 3rd April 23, Midjourney released a trial feature named as “/describe”. See below announcement from David in Midjourney Bot:

It means, we can give any image with /describe and MidJourney will give us 4 responses or text prompts on what is inside that image.

I have used it and found it useful in two scenarios:

1. Remember that Midjourney does not allow to give one single image prompt if it not an internet link. So with /describe we can actually put a single local image file and then start working towards our desired output.
2. We do not have to think about prompts as the bot will do the 90% job. I have seen the results and they are good in multiple cases. If you want to refine the output text prompts received from describe, you can further do so by selecting the desired text prompt and then editing it as needed.

Let us give a try :

I attached a drone image and got below 4 text prompts related to the image

Now i can select any of the text prmpt from below 4 options and get another imagine this screen to further edit the prompt as needed:

The results are as below and now we can do regular upscaling or variations as needed.

There might be some errors though as it is still on trial and I have faced one of them as below :

The text prompts that we get from the describe comes with output that has “ - ” as below and and this can cause BOT to interpret it as a parameter and throw Invalid parameter error because of unrecognised argument.

The above error came from the prompt received from the Bot itself as shown below, see number 4 prompt. While generating we can remove the “-” however doing it all the time will be pain. The information is posted in discord for someone to fix it.