Midjourney just dropped its latest
game-changer, Midjourney v6! Today, we're putting it head-to-head with its
biggest rival, Leonardo AI Alchemy. When Leonardo AI Alchemy burst onto the scene,
it made a name for itself with a strong focus on realism and some nifty features like turning still
images into dynamic wonders. On the flip side, Midjourney v6 is all about crafting
detailed, dreamy outputs and rocking a killer art style in its animations. So,
let's find out if the hype holds true. We’re here to help you figure out which
one suits your project best – whether it's text to image, image to image,
keeping those characters consistent, or even cooking up some sweet videos.
Let's get this showdown started! Starting with text to image, let's dive
into Leonardo AI Alchemy. Just a heads up, you need to subscribe to one of their
plans to use it. No pressure, though—I suggest considering it after watching
this video. Stick around till the end, and don't forget to hit that subscribe
button for more AI tech goodies. Now, let's hop over to Leonardo AI and give image
generation a shot. Don't forget to turn on Alchemy V2 on the left side. Leonardo Diffusion XL
excels at artistic styles, Leonardo Vision XL prioritizes photorealism, and AlbedoBase XL
focuses on detailed character creation. Today, we're rolling with the AlbedoBase XL.
We're going to use the same prompt for both platforms and here's what I came up with.
“A pixel art animation of two baristas crafting the perfect latte. One barista is making a
steam swirl, another is wearing a colorful shirt and frothing milk for her latte,
and there are cute pixelated coffee beans dancing in the background. With a backdrop of
‘Coffee Please!’” Hit that generate button. Take a look! These pictures look
really nice. The visuals are awesome, in my opinion. Even though it didn't quite
nail the background thing, the overall result is still impressive. You can also use Live
Canvas Leonardo AI and the Leonardo AI Canvas Editor if you want to tweak your images even more.
Now, let's check if Midjourney can do even better. To use Midjourney v6, all we're going to do is
add a double hyphen and v6 to each prompt. So, let's type slash imagine, then copy-paste the
prompt we used for Leonardo Alchemy earlier and add -- v6 at the end. If you
want to include text messages in the image, make sure to put quotation marks. Now,
let's hit generate and see what we get. The results from Midjourney are just as good
as Leonardo's, and it nailed the backdrop this time. It looks like it followed
our prompt perfectly. I've got to say, it outdid Leonardo's earlier results.
Now, let's dive into experimenting with image to image. We're taking this black and
white image and turning it into something colorful and lively. With Leonardo AI image
to image, we'll start by uploading it. For our prompt, let's type something like “Paint
this picture in a bold style of colorful pop art, each member bursting with personality
and energy.” Switch the fine-tuned model to Leonardo Diffusion XL, lower the
image strength to 0.20, and hit generate. Once again, the results didn't disappoint.
It's vibrant, lively, and full of color. Now, onto Midjourney v6. Let's begin by
uploading our photo. Once it's uploaded, copy the image URL and add our prompt,
including the double hyphen v6 at the end. You can also incorporate your Midjourney
Style Tuner if you want specific styles. While Midjourney did well earlier, it didn't quite
bring its A-game this time. I mean, look at that. So far, Leonardo Alchemy has been decent,
but let's give it another shot to see if it can generate consistent characters. To
start, when crafting consistent characters, the key is to first find the perfect starting
image. We'll use a detailed prompt to describe our character – and I’m going for something
different like “Generate an image of James, a realistic green lizard-human guy, 18 years
old. But most of his features are human-like. His eyes are curious and brown, and he's got a cool
jawline. When he smiles, it's friendly, and his nose is uniquely shaped. Plus, he wears a stylish
jacket, making him a chill humanoid lizard.” Now, let's generate at least ten images and
save them for later use in our model. Moving on, we'll create our model using the
description we provided earlier and name it James. With our model in place, it's
time to generate some consistent characters. So, we'll generate James two times in
different situations and positions to test Leonardo Alchemy's consistency. For example,
“Create an image of James, performing at his own concert. Picture him in a cool t-shirt,
surrounded by fans, with vibrant stage lights.” The results are okay, not bad but not perfect.
The images don't exactly match our idea of James. Let's try placing him in another scenario, like
“Generate an image of James sitting on a bench, hand extended, with a bird perched on
it and a bright smile on his face.” Unfortunately, the results don't look great
– there are some awkwardly placed parts, and the facial features don't
quite capture James' likeness. Let's see if Midjourney can step up its game
in creating consistent characters. First off, we need to make a description for a
character we want to be consistent. Same with Leonardo, let’s use the prompt
“A realistic photo James, a realistic green lizard-human guy, 18 years old. But most of
his features are human-like. His eyes are curious and brown, and he's got a cool jawline.
When he smiles, it's friendly, and his nose is uniquely shaped. Plus, he wears a stylish
jacket, making him a chill humanoid lizard.” The results look promising. If you find an image
you like, you can click V1, V2, and so on to generate more similar photos; let's save them.
Now, upload the saved photos. After uploading, save each image address by right-clicking
and save it in your notepad or chat box. To make this character consistent, type “/prefer
option set” and set the option part as its name, James. Click “plus one more,” and in the ‘value’
part, paste the image address we saved earlier. Now, let’s try recreating the same prompt
we used earlier “Create an image of James, performing at his own concert. Picture
him in a cool t-shirt, surrounded by fans, with vibrant stage lights.” add double hyphens
then James, then double hyphens and v 6. By providing Midjourney with reference images, it
should be able to reproduce James' facial features in various prompts. Let's see how it goes!
Okay, first of all. The face looks really promising, and the images are spot-on with
those extra touches. Let's give it another shot with the same prompt we used before, and
once again, it doesn't let us down. The face, the background, and all those extra details
– everything's just perfect. When it comes to making consistent characters, Midjourney
consistent character v6 is the go-to, no doubt. Now, let's wrap up with a bit of a showdown. We'll
try making video snippets using these tools. For Leonardo AI Alchemy, its “Upscaler” feature can
take a still image, pump up its resolution and details, and then you can turn it into a short,
looping video clip with subtle movements and animations using “image2motion.” Perfect for
adding a bit of life to your artwork or photos. I tested it on the picture of James rocking
out at the concert, and after about a minute, we've got our clip. It looks good, really
great. But there's a bit of an issue – some facial features are a bit off throughout the clip.
Now, onto Midjourney v6. The process is pretty simple. Just add double hyphens and "video"
to every prompt where you want a quick clip. But don't forget to finish with double
hyphens and “v6” at the end. For instance, let's try making “lizard slowly catching a bug
with its tongue timelapse” – add double hyphens, “video,” “v6,” and hit generate. Once your grid is
good to go, give it a thumbs up with the envelope emoji, and the Midjourney Bot will send you a
link to your awesome video in Direct Messages. But here's some cool news! Midjourney v6 is
gearing up to train video models starting in January 2024, and they're planning to tackle
3D training after that. It's something to look forward to! But in this video, it's clear that
Midjourney v6 is elevating its performance, especially in handling complex prompts. While
both platforms deliver high-quality images, Midjourney V6 takes it to the next level.
On the flip side, Leonardo Alchemy stands out for its user-friendly interface. If you're a
beginner, Leonardo AI is definitely worth a try. Once these features are ready, we'll make another
video to show you what they're all about. So, if you enjoyed this video and found it
helpful, be sure to give it a thumbs up, subscribe, and drop a comment below.
Stay tuned and watch our other videos for more straightforward reviews on
the latest happenings in the AiWorld.
