Midjourney Prompt Advanced Parameters
Quality
Image quality is another attribute that I often use. First, it's important to note that this parameter does not affect the resolution and does not change it.
It changes more of the details in the image. For example, in the first image below with a Quality setting of 0.25, you will notice that there are many more details on the right side compared to the first image.
However, low Quality also has its benefits - generating images will be faster with less GPU time required.
I usually use this parameter for exploratory generated images where overall direction is good and then let the model enrich details.
In addition to typing in the prompt, you can also adjust the default version in the settings, see Midjourney Basic Settings for a tutorial.
| method | case | note | v5 | v4 | v3 | niji |
|---|---|---|---|---|---|---|
| After the keyword, add a space and include the parameter: --q or -q --quality or -quality | After the keyword, add a space and include the parameter: --q or -q --quality or -quality | Default is 1 Only .25, .5, 1 and 2 are supported, if the input is greater than 2 it will be downgraded to 2 | ✅ | ✅ | ✅ | ✅ |
Seed
When using Midjourney, you may notice that even with the same prompt, each generated image is different. In fact, Midjourney assigns a seed number to each generated image. You can add this seed parameter to your prompt so that the generated images are consistent every time. Here is an official example:
The call method is simple:
| method | case | note | v5 | v4 | v3 | niji |
|---|---|---|---|---|---|---|
| After the keyword, add a space and include the parameter: --seed or -seed --sameseed or -sameseed | woodcut birch forest --seed 123 | Enter the exact same text prompt and add a seed parameter to generate an identical image. Even with this seed, versions V1, V2, V3, test and testp cannot generate exactly the same image; only similar images can be generated. | ✅ | ✅ | ✅But only approximate maps can be generated | ✅ |
How to obtain this seed?
Find the image you want to get the seed for, then there is an emoji button in the upper right corner of its message.
After clicking it, an emoji menu will appear.
Then enter 'envelope' in the emoji menu and finally click on ✉️emoji.
An envelope emoji will appear below the picture, and you will receive a private message push from Midjourney Bot with the Seed number on top of that message.
Image Weight
When you use both img prompt and text prompt at the same time, you can use this parameter to determine whether the AI-generated output is more similar to the image or the text.
The larger this parameter, the more similar the generated image will be to the original image; while a smaller value will result in an output that is more similar to the textual description.
Below is an official example from Midjourney where you can see that as the value increases, it becomes more like the original image:
| method | case | note | v5 | v4 | v3 | niji |
|---|---|---|---|---|---|---|
| Add a space after the keyword with the parameter --iw or -iw | flowers.jpg birthday cake --iw .5 | This parameter has different default values in different versions (even if you don't input this parameter, AI will add it by itself) and ranges. | ✅Default is 1, the range is 0.5-2 | ❌ | ✅Default is 0.25 and the range is -10000-10000 | ❌ |
Stop
This parameter is similar to the stop in ChatGPT, which forces the model to pause.
| method | case | note | v5 | v4 | v3 | niji |
|---|---|---|---|---|---|---|
| Add a space after the keyword with the parameter --stop or -stop | splatter art painting of acorns --stop 90 | The default value is 100 You can enter 10 to 100 | ✅ | ✅ | ✅ | ✅ |
Style
This parameter is mainly used to switch branch models in the V4 model.
In my opinion, this parameter is not very useful and because V5 has already been released and everyone is using it, I think it will be phased out in the future (I suspect that this was actually an earlier version of one of the other branches of V5 that was released to collect data). Therefore, I won't introduce it further.
Uplight
After Midjourney Bot generates an image, there will be a row of U buttons below the image. These buttons do what the Uplight parameter does. Most people use them by clicking on these buttons rather than adding them in prompt.
Moreover, with the current V5 version, U buttons no longer affect resolution but only serve as a function for cropping images. So this parameter is also something that I believe will be phased out and you don't need to worry about it.
Upbeta
Another button that I think will be phased out.
Video
I think this parameter is quite interesting as it's similar to recording gifs. By adding this parameter in prompt, you can record the process of generating an image while converting it into a video format at the same time. Unfortunately, this feature only supports up until V3 and I estimate that official support for Video may also eventually be abandoned since its usefulness seems limited.