MIDJOURNEY V6: MULTIPLE SUBJECTS HANDLING GUIDE
This guide provides:
- Tips and tricks to handle multiple subjects( 2, 3, 4 and 5 subjects) with their positions in the image.
- Multiple examples with proper prompts for different amount of subjects.
Come on, let's go 👇

First, some important TIPS AND TRICKS to get better results:
1) Initiate with a comprehensive sentence describing the entire scene, such as "Three men sitting on a table" or "Four different cats sitting on a couch."
2) Proceed to describe each subject by referencing them like the initial sentence, for instance, "a man in the middle" or "a cat on the left."
3) Prefer using "positional" terms like "on the left," "in the middle," and "behind" over ordinal terms like "first," "second," or "third."
4) Following subject descriptions, provide details about the background and environment if you have specific elements in mind. Elaborating on the setting often yields better results.
5) Conclude by incorporating phrases that articulate the desired image aesthetics. The more precise your description of the aesthetics, the higher the likelihood of achieving the desired outcome.
Thanks to Kate Khoisan for these Tips and tricks.
Here is the link to article:
medium.com

a) 2 SUBJECTS
PROMPT STRUCTURE:
This is the general prompt structure for (2, 3, 4 and 5 subjects).
[General idea of what the photo is about including the number of subjects]. On the left, [s of subjects related to their position]. --style raw
Further dissecting:
[General idea of what the photo is about including the number of subjects]. On the left, [SUBJECT DETAILS]. On the right, [SUBJECT DETAILS]. --style raw
Note:
- Works almost perfectly and understands the positions of both the subjects clearly.
- Works perfectly for two objects as subjects as well(3rd image, 4th image).
- MJ is still not good enough for multiple subjects having different emotions. I have tried multiple times but to no avail. For a single subject, it works decently but for multiple subjects, the accuracy is quite poor.
- It works almost perfectly with different clothes, hair colours, and hair lengths of multiple subjects. As clearly can be observed in 1st and 2nd images.

b) 3 SUBJECTS
Dissection of prompt:
[General idea of what the idea is about including the number of subjects]. On the left, [SUBJECT DETAILS]. In the middle, [SUBJECT DETAILS]. On the right, [SUBJECT DETAILS]. --style raw
NOTE:
- Almost Perfectly understands the position of 3 subjects.
- Works well with the subject's looks and appearance.
- Works well with objects.
- Problem occurs when you try to include the age or ethnicity of a subject. It gives the ethnicity of one subject to a different subject or will make all the subjects of the same ethnicity even though you have mentioned otherwise.
- Still with three subjects you can do that. It gives the desired results 4 to 5 times correctly out of 10.
- I have included two examples in the next section regarding the problem(Do check that out).
- A solution is also provided on how to get over this problem in the next section.

c) 4 SUBJECTS
PROMPT DISSECTING:
[General idea of what the idea is about including the number of subjects]. On the left, [SUBJECT DETAILS]. First [SUBJECT] second from left [SUBJECT DETAILS], Second [SUBJECT] from left [SUBJECT DETAILS]. On the right, [SUBJECT DETAILS]. --style raw
OR
[General idea of what the idea is about including the number of subjects]. On the left, [SUBJECT DETAILS]. First [SUBJECT] second from left [SUBJECT DETAILS], First [SUBJECT] from right [SUBJECT DETAILS]. On the right, [SUBJECT DETAILS]. --style raw
Notes:
- Works decently with objects.
- Problem occurs with Human subjects.
- Especially when you throw so much information about the subjects.
- In the second example, I kept things general and minimal and got the desired result.
- In the third example, I added different ages and ethnicity. That's why I did not get the desired result. The second girl is not a 20-year-old French girl.

d) 5 SUBJECTS
Not recommendable, Mj is not capable yet of handling this much information and that too about five different subjects and their positions.
Still, for you all I have included examples, check them out.
NOTES:
- I have included a Prompt dissection for this one. You can easily understand that from the prompts provided.
- Works fairly well with objects.
- The second example can be created with a single-line prompt. I did it to give you the perspective of different subjects and their positions.
- The results are decent enough only with objects as different subjects.

I hope you will find this helpful
Follow @AI_Vision_Verse for more content
Repost the first post below to share it with others 👇