Hey everyone, Rodney here from Kleebz Tech.
Can you get Midjourney-like results with Stable Diffusion? Can you achieve those
results with just four steps? Today, I'm going to give you a quick rundown on LoRA
I ran into the other day: Mid Journey Mimic, which is created to give you Mid Journey 5.2
model-like results using Stable Diffusion. Now, I'm a big user of Mid Journey myself for some
things, so I decided to give this a try and see if it does what it says. Since Fooocus is my primary
Stable Diffusion tool, I did my experimenting with Fooocus. Unless I mention otherwise, all these
images were generated using only the Fooocus V2 prompt expansion, and I went with the first
generated image. So, there was no cherry-picking of good images. I used the same seed, did the
same thing with the checkpoint. I used the Juggernaut XL version 8. This LoRA is designed
to mimic Mid Journey style 5.2, and it works. They say it acts as a detail tweaker, which
supplements the picture with details; a color enhancer adds contrast, brightness,
background depth enhancer, which adds depth to the background. They do recommend its use
only with weight from .1 to .8. You can set more, but the picture, they said, will be colored;
smooth edges can break proportions. In my experimentations, yeah, definitely anything
over .8 weren't the greatest results. Now, I will include a link to this,
obviously, in the description. And if you're not familiar with how to add models
to Fooocus, I do have a video that covers that with detailed instructions. But basically,
with all of these, you're going to download it, put it into whatever folder the default one
for Fooocus is in the Fooocus models folder, and then there's a LoRA folder. And that's all
you do. You put it in there. To download it from this site, if you go from here, just go over to
download options, you click there, and then you click on the file here, and then choose where you
want to save it. Save it to whichever folder you keep your LoRAs in. Now, if you have it running
and you download it, you do need to click on the refresh all files, so it will see the new ones.
And then you just select it in the dropdown. As we can see here, I have it selected, and then
you want to adjust the weight. .1 to .8 is the recommended. That's pretty much what I found. I
actually even found a lot over .5 or .6 really, I most of the time didn't like the
results, but it depended on the image. So, I'm not going to make you wait around for the
lightning results, if you're curious about that, because I'm not going to go into a lot of detail
on those. Because I don't use the lightning model a lot, I'm still experimenting a lot with it. So,
I had it set up to use the lightning setting in Fooocus, and in there, I was using the Fooocus
V2 style, and the model was the Juggernaut, and I did it with the Mid Journey LoRA off and with
.05 settings. And the results of that, the top two images are the default lightning without the
LoRA enabled, and the bottom two are with it set at 0.5. Depending on what you're going for with an
image, I do like the bottom images better. That's basically the lightning result, so it does, to me,
seem to improve it when I played around with it. Now, let's just look at the regular results. The
first image on the top left will be LoRA disabled, and then we had it at 0.25, 0.5, and
0.75. Probably at the lower weights, I find it's more close to the 5.2
in certain aspects. It does change; that is all I can really say there. Like, for
this image, I prefer the one with it disabled, maybe at 0.2 to 0.5. For the Magical
Forest, I would like the bottom two images preferably. But then again, it depends
on what you're really going for with the look. Now, for our wizard, I definitely liked it
around the 0.5 weight. I thought that was the best image overall. And for the Woman Dancing in
the Street, I thought the bottom two were better; it definitely improved. I felt that the
standard one just wasn't that great. As for the Tranquil Forest at Sunrise, I think
the last one with the heavier weight actually came out best. But then again, I like
all of them. I don't know if they hit it perfectly with everything here. I
like the results of this LoRA overall. Now, for our still life, 0.5 is the one I
like the most. As for the astronaut one, I think the original one was probably
fine there. None of them are perfect, I'd have to say; most of these are
pretty bad images. Bottom left got some really bad stuff going on, so
we'll just move on to the next one. Sailing ship, Viking sailing ship, I like
the bottom. I think 0.5, I don't know, it might could be that lightning is making it
much better. Then we have our Hobbit Hole. Now, the original one looks like I went outside
and took a picture of it. In my opinion, the other ones are, yeah, definitely, I can
see the Mid Journey influence through these, and on this one, I think as well, it's holding
the coffee cup properly. For the most part, they're not too out of whack. I usually
don't get those good of results. Just to give a comparison of what I got for
when I went from the lightning settings I showed before back over to the standard, just
the Juggernaut under the quality settings, just so I'd have a comparison. So these were
the lightning ones that you're seeing on the screen right now. But these are a couple of
examples; these are the same prompt run under the standard quality settings, Juggernaut. The one
on the left is the default Juggernaut settings, and the one on the right is with the 0.5
on the Mid Journey. Here's another one, same prompt, Mid Journey is on the right,
with that at 0.5. On the left is just the standard Juggernaut. I can definitely see
the Mid Journey type influence on this. That should give you an idea if you want to
check that LoRA out, especially if you're a big fan of Mid Journey, or you just want to
get similar results, whatever it may be. So, it does have overwhelmingly positive ratings
on the civit ai.com website, and a lot of the reviews are really good. So, go check that
LoRA out, download it, and play around with it. Thanks for watching the video, and if you
found it helpful, please do hit the like button. Any tips for getting better results
with this LoRA, just drop it in the comments for everyone. Any other suggestions for any good
LoRA, definitely put them in the comments. I do plan on doing some more videos like this,
and hope you have a good time creating.
