This morning I got a notification from the Midjourney discord that we could now select version 5 for our image prompts! It's only currently in alpha and is subject to change, but I very quickly got my eager little hands onto it and played around for a bit.
First of all, I changed my settings to select version 5 so I didn't have to type it every time.
Over the past week, I've been checking out a few images that have been generated with version 5. The developers wanted people to choose between two images and rate which one was better to help fine-tune things, and I was very impressed with a lot of the images I saw.
Now, let's see what it can do with a few of the prompts I've already been doing for my world of the Evergreen.
I used this prompt:
wide-angle, cinematic lighting, fantasy, magic:: a man with black wavy hair and orange eyes wearing a silver robe casts a purple spell from both hands, he protects a woman with long red hair, she cowers behind him, scared, afraid:: the background is a lush green forest with lots of leaves, shrubs, trees, that glow green at midnight, full moon:: --no lantern --no fire --aspect 2:3
Which is basically the prompt I was working with in my last post of the man casting magic. I just wanted to see how it would look on this version.
When I first used this prompt on version 4, I got image after image of two silhouettes walking a forest path. Nothing else was happening until I fiddled and tweaked a lot.
This time, on the new version 5, we can clearly see the two characters and they're in the forest. That in itself is an improvement! The man isn't casting magic though, and the woman isn't cowering in fear at his back, but, overall, I'm more pleased with this very first result than I was with the very first result of version 4.
The developers are saying that version 5 works better with more complete sentence structure and demands, so I tried the following:
wide-angle, cinematic lighting, fantasy, magic:: a man with black wavy hair and orange eyes wears a silver robe, a purple magical spell is cast from both of his hands. Behind the man is a woman with long red hair and pale green eyes, she cowers, scared and afraid:: the background is a lush bright green amazonian forest that glows a bright vibrant green --no fire --no lantern --aspect 2:3
The forest is much improved this time! But once more the two characters are standing around and not doing much else. Clearly, to get what I want, I do have to dabble like I did last time.
Out of curiosity, I used the prompt with one of my images from version 4. This was my favourite image from the last post and I think the people in it look wonderful. It's mostly just the background and the lack of purple magic that bugs me.
I was impressed enough with it to reroll it a few times, just to see what I would get. There was one that I liked, but the man looked so depressed that he was casting magic, I couldn't bear it, haha! He's supposed to look brave, heroic, fierce!
This one shared above is my favourite of the lot. In saying that though, I didn't play with the man and woman that much... I reckon if I spent an entire day of it like I did last time, I just might be able to create that ~perfect~ image that I wanted. Will try at some point! You'll notice that all of these are portrait aspect, that's for possible future book cover reasons. 😝
On the note of re-rolling, I really wish there was an easy way to remove images from my gallery without having to go through my entire Midjourney history and emote a red X at what I don't like. Sigh.
Next up, I wanted to test my forest prompts. I've done a few of these in version 4 with varying results, and now I was curious what version 5 could come up with. I was impressed with quite a few of version 4's results, so I had high hopes for version 5!
wide-angle, fantasy, magic, cinematic lighting:: waves splash up around the inside of a bubble centred:: background is a large vibrant lush green forest at midnight, full moon, the leaves glow bright green
It didn't seem to notice my midnight and full moon inclusions, instead the sunshine filters through the forest canopy.
I tried again, this time being slightly more flowery with my words.
wide-angle, magic, fantasy, cinematic lighting:: a bright neon green amazonian forest with leaves that light up bright green. It is midnight, the moon is full and shines through the leaves:: a large single bubble sits on the forest floor:: --no fire --no sun --no lantern
It still seems to be daylight.
a magical forest at night. The leaves of the forest glow a bright neon green. It is midnight and the full moon sits above the forest. On the forest floor is a single bubble, sparkling beneath the moonlight:: wide-angle, cinematic lighting, magic, fantasy
This time I spoke in simple sentences and repeated certain aspects, and put my modifiers on the end of the prompt instead of at the start. That seemed to work a lot better! It also seems to have incorporated the "magic" a bit more, which is nice. Honestly, the forests of the Evergreen in my mind aren't too different from reality, but I like seeing these types of things when I input "magic."
wide-angle, cinematic lighting, fantasy, magic:: a forest surrounds a lake at midnight, the moon is full, the sky is full of stars:: colourful glowing flowers grow beneath the water, a mass of bubbles float above the surface
and then with the modifiers placed at the end:
a forest surrounds a lake at midnight, the moon is full, the sky is full of stars:: colourful glowing flowers grow beneath the water, a mass of bubbles float above the surface:: wide-angle, cinematic lighting, fantasy, magic
I'm not quite sure what to think about modifier placement. It seems to work better at the start for some ideas, and better at the end for other ideas.
For my next prompt, I inputted the very first prompt I ever put into Midjourney, before I knew of more appropriate prompt layouts.
A forest scene at midnight. The trees and flora are neon bright, the moon is white and full, the skies are deep purple. A man and woman in magic robes are silhouetted against the moon. Golden fairy lights border the image. A magical scene.
It doesn't look too much different from what I got in version 4, but I have to say that the silhouettes of the people are much improved! They were quite polygon-esque in my very, very first attempt. 😅
Actually, as a whole people seem to be improved in this new version. Especially their hands.
As I was playing around with the man and woman before, there were a couple of instances of janky-hand, but for the most part they actually looked like proper hands!
wide-angle, cinematic lighting, fantasy:: an adult woman, 30 years old, with long red hair and pale green eyes, clean clear face, high cheekbones, wearing a long silver robe with hood down:: the woman sits at a medieval tavern holding a wineglass:: silhouettes of people in the background
I decided to use one of my previous prompts that had resulted in an image that I liked, of my lady in a tavern.
LOOK AT HER HANDS. They're not bad. Not bad at all.
Yes I am excited over hands.
I can't get them holding their damned glasses and don't have the time to work on it all day, but, HANDS. Speaking of hands, let's get some dude casting some magic from his normally shaped HANDS.
full body, wide-angle, cinematic lighting, fantasy:: a man with black wavy hair and orange eyes casts a purple spell from both hands
I'm really enjoying playing around with Version 5 of Midjourney, even though it's only in an alpha stage currently.
Unfortunately I've been playing around with it for 4 hours now and I really had other things I needed to get done today. xD So I'm forcing myself away and will dedicate a day to it soon.
I never did get my image of Tomias protecting Kelsey with an arcane barrier as an evil creature circles them. Maybe I'll be able to get it this time!
Before finishing up for today, I merged my people drinking at the tavern with the people at the start of this post... you know, it's nice to have consistent-looking people. 😅
Now they're enjoying some socialisation in a very green and flora-filled forest-tavern. xD
Okay, that's enough for now.
Must play with this more later.
Until next time! 😊✨
