Midjourney’s Secret Playground: Chaos, Weird, and New Editor

Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:00):
What's up everybody, mid journeyfast hours podcast, early
morning edition early. What is this pre dark, pre dark,
pre dark morning edition, pre sunrise edition.
I don't know. We'll we'll workshop that thing.
We'll figure it out. We were just trying to find a
time to to shoot and we had someupdates come through with mid

(00:21):
journey, which we'll talk about today.
Lots of them, lot of things thathave already dropped into the
latest update, but also some things that are coming up here
as we sort of like start to fallin line with this dripping
system of new things coming. Seems pretty quickly here, man.
Like, I don't know how we're going to keep up all the stuff.
We might have to cancel all the vacation plans.

(00:41):
It's appreciated though, like I this to me is so much better
than still waiting for version 7.
Like I'd rather just have this. I think again, I said it many
times, they overthink this stuff.
Just release it, let us play with it on to the next thing,
you know, keep the hype cycle going 'cause you see it way more
now. Like I feel like 5 weeks ago,

(01:02):
six weeks ago, before this got released, like it was kind of,
it was kind of dying out on things like Twitter.
Like I wasn't seeing as much just mid journey stuff.
I wasn't even out it's. Just like right back in Bang,
there we go, 'cause it's constant new stuff and then
everyone's testing it, everyone's playing with it.
And it's not just like this giant release where everything
gets lost and it's like these. They just did it better.

(01:24):
That is the way they should havedone it.
Yeah, and we got, we got a couple new things back into the
fold. So weird has been dropped in.
We'll talk through that. I'm curious if you're still
using that. Also, what was the other big 10
editor? Editor was the big one.
So finally combine that into oneplace.
Now there's a lot of different panels within that.

(01:45):
They've they've sort of like reorganized that UI of that
particular page too. I'm curious, you know, as we
dive into both of those, one like are you still using weird,
you know, is that still something that you'll go back to
and when, like when are you going back to use it?
Just for fun, I mean like when you look at let me go back to

(02:05):
that prompt from like a few weeks ago where it's like the
work is mysterious. And important.
That's a. Perfect one to use it on it's
you still get I still, you know,contend that you get things with
weird that you never would get otherwise.
You just, it just has this weirdno, no pun intended.
Maybe, pun intended. Weird.
It's weird. Streak to it and you get some

(02:26):
create, you know, you do get some.
It's like chaos when you dial upthe chaos and you never, you
know, you don't use it that often, like super high chaos.
When you do, you just get thingsthat you never get.
So creativity level, just Max out.
Just, you know, if anyone hasn'tplayed with weird yet, by the
way, it's just dash, dash weird and the values are zero to 3000.

(02:47):
So it's, you know, you can go. It's a big range.
That's what that was. My next question is if you've
played with it enough yet since they've dropped it back in to
know if that spectrum has changed at all in terms of like
what you get back, right? Because that was one of those
that was like especially weird in terms of the spectrum, right?
Like you could throw weird 3 in there and all of a sudden you've

(03:09):
got like some really different stuff.
You didn't have to dial it up super high.
Like have you, have you seen a difference there or have you
played around with enough to know or maybe the look that we
dive into today, I don't know. Yeah, we can definitely show.
And I think weird one through 10is like where I live with it for
certain things, but like when I want to get creative and just
like see what it'll pump out, you know, weird.

(03:30):
There's like diminishing returns.
I've always found like up to about 1500 it's good.
And then like after that, like it doesn't get any weirder.
Like 2000 doesn't feel differentthan 1003 Thousand doesn't feel
different than 2000. But between zero and like 1500,
it definitely is. Like, you know, difference
between weird 100 and 400 is pretty weird.
So it's it's good stuff. I mean, it's just fun.

(03:53):
It's fun to just throw in there.If you want to waste some some
hours and some generation time, we're a good parameter for that.
I'm thinking maybe we, we, we dial that up a little bit.
I'm curious to see if that's changed at all, you know, like
in terms of, you know, the, the spectrum itself.
Like it's been a while since I've used weird, but when I did

(04:16):
like I think I used it almost always definitely under 10, but
a lot even under 5. Like it was like 2345.
Like even just like sprinkling it in there would really break
the mold on some things. It it's really useful.
It's like a as like, you know, the, the piece of the whole
analogy that we use all the timewhere it's like, OK, if I dial

(04:36):
up the weird, it's not necessarily because I want that
output image. You might just like a style and
then that's a style reference, you know, like that's how I look
at it. Or I want to go create like a
mood board from something like that.
It's, it's a good way to just sort of like survey the mid
journey landscape and find, you know, things that it can do that
you just wouldn't ever think to prompt for.

(04:57):
And I was, I was running it like, I don't know, last night
or the night before, just to like see how just to like play
with the parameter and see how the, the weighting was and like
the sensitivity. And it was, it felt pretty much
exactly the same as it always has been.
But I always like it for that. It's just like, oh, I can use it
as an S ref now and use this on something else because it's not
going to control is not not the thing you're dealing with.

(05:19):
Weird. I agree, yeah.
I'm just, I'm going to, I'm going to run it for fun.
So I'm going to run the work as mysterious.
Let's let's do this though, likeI, I do think maybe two things
to keep in mind with this one. This is a little bit of a broad
ambiguous term in general. So I don't know if this is the

(05:40):
best prompt to run weird on in the variation.
So I think what we might do is maybe run 2 separate prompts
here. Let's run one that's like this,
and then let's run one that's a little bit more specific.
Yeah. That way we can kind of see what
the difference is here. So let's run that.
Maybe I'll just go back in time here and let's see if I have

(06:07):
anything like specific. The nuns might actually be good
because you might get some really interesting outcomes with
people. I always find that the like the
look gets significantly different.
So let's take this. Sometimes if there's too much
like environmental detail, like it becomes so surreal and weird
that it doesn't even make sense.And sometimes I find it better

(06:29):
with people because you get likea very unique look.
That might be a good one. Yeah, I was thinking something,
maybe something like this actually.
So let's take this off. You back on the style raw game.
Yeah, yeah, I've been using style raw all.
Right. Yeah, let's do, let's do the

(06:50):
same thing here. So we've kind of seen what 0
looks like. This is 0.
Let's do like A5, let's do A10, let's do 100, let's do 1000, and
then let's dial it all the way up and run, OK.
And then let's go back to what we had here, right?
So the work is mysterious. Oh, it's got your profile on it,

(07:16):
so it's it's going to stay fairly.
It does. That's one.
That's one thing to note like when with weird in your profile,
like it still stays pretty true to your profile.
So if you take it off, you get like way more variety of
outcomes. Of that.
That's something that I've, you know, little, little not unlock,
but just something that I was finding while I was doing it.

(07:39):
Yeah, that seems to. Well, let's see, Let's dial this
up. Here, like if you go, if you go
back, go back a few images thereto that guy that was a close up
of the guy like that, that to meis like right there, like I
might take that and then use that as a style reference
because that you weren't going for that, right?
Like, but that's sort of like very cool to me.
And I'm like, all right, let me play with that.

(07:59):
I can I can go and sort of, you know, going back to our micro
decisions episode, where does this where does this take me?
Where does this leave me? Yeah, I mean like that's, that's
a great call out. I mean, you could use this as a
jumping in point to a, to a process, right?
Because like you could find something that's a little bit
out of the norm and then to yourpoint, use that as a image raft

(08:21):
for a style raft for, you know, inspiration for where you're
kind of headed with it. Now, these are all still with
the the profile on, though. I mean, these are getting a
little bit weird. I mean the question I would have
is like do you really want to use weird with profile off
though? Sometimes.
Well, that's that's why I basically.
Use personalization off I mean. Yeah, I basically use it as like

(08:42):
mining S refs in the same way, you know where I'm like I'm just
looking for, I'm looking for something that I can't that I'm
not able to prompt for or I'm just testing mid journey sort of
creative capability with it. Because some of these are great,
like some of the ones of him riding the horse, like you have
a very different look to it, like that one might not be what
you were going for. That one's pretty good.

(09:04):
Right. But it's it's something that's
unique. Well, I like this because it is
specific and let's see how the weird sort of effects it here.
So now we're getting into 100. It's really changing the the
coloring for sure in these images like these are.

(09:26):
Look at that face, bro. Yeah.
You definitely get more of that.Yeah, I mean, but they don't
look, they don't feel great. Like none of these are ones that
I would actually pick. So let's, let's do this.
Let's take this same thing. Now let's take this off.
Yeah, turn your your P code off.There you go.

(09:48):
And then let's run. All right, let's run the same
same thing here, just out of curiosity.

(10:10):
OK, so now here this was what itreally started to kind of get
quote UN quote weird here at this 100 somewhere between this
25 or this 10 and 100 here. Big difference, kept a lot of
the same like I noticed with my global on 7 which I don't love

(10:33):
love but it's got a lot of like cyan and magenta.
Yeah. Which I I actually like the
opposite. I like more of like the yellow
and orange personally, but you see it's breaking, breaking the
mold here. Like none of those have that
same color profile there. Now this is without
personalization on. So your problem, your problem is

(10:58):
like definitely dialed in because it's it's coming out
very similar. This is a very specific one.
Yeah, so that's how you know it's a good prompt, because even
with weird, it's not, it's not getting overly weird.
It's the only thing it's really changing is the face and the
colors are completely different.Yeah.

(11:19):
Still kind of given that same yeah, maybe we'll just try the
work is mysterious one too. Yeah, try that one without
profile. And let's run that.
Because this one, this one should have varying outcomes.

(11:43):
So let's do a 510. Let's see, I'm going to do
another stop in between here. Let's do like a 5000 one
thousand, not really seeing a huge difference still between
1003 thousand. So let's just do that.
Yeah, that's typically. And so it's between 1 and 1000
to me is like the biggest differences.
And then after that like 1500 ish, it's pretty much pretty

(12:05):
much similar. This is also where I think if
you're going to go into this exploration mode of this entire
thing, this is where aspect ratios start to come back and
play too, especially with, well,just in general, but also with

(12:26):
personalization off, right? You're going to get totally
different results for the work is mysterious in general, but
you're going to get even more different results when you, you
know, vary those aspect ratios. So this isn't a four.
What is this 4/5? You know, so maybe you'll get
more people in these versus if you were to do like a 69 or
something like that, it might bedifferent.

(12:46):
So this is with Weird 5. See, it's interesting how it's,
how it's interpreting it 'cause I, I like some of these
compositions too. And I'm like, oh, I could play
with that. So I use it as like a, like I
said before, like a creative sort of outlet to play with sort
of mid journeys brain, see whereit goes.
This could be a good this, this could be a good time to revisit

(13:08):
that, right? Because like you just made me
think of something which is interesting.
What if you just kind of lock insomething in between this broad
spectrum and the very specific where it's like, hey, you have a
specific composition in mind. Let's add that, right?
And then use the variable of weird to explore within it.
Or let's add a certain lighting or shadow.

(13:29):
Or let's add a certain camera angle like lock into one or two
components and then explore within that with weird too.
Like yeah, try it. Try this prompt real quick.
Just try award-winning photo of an ancient warrior.

(13:50):
And then this is what I was getting some real funky ones on.
You want a specific aspect ratioor style.
Yeah, let's go, Let's go 4-5. That should work.
OK. And then what about style?
Are you do you want? To dial it up a little bit,
let's go like 500, no personalization and then let's
go weird go. Let's do a baseline.

(14:13):
Let's do like. Zero do 100.
You don't. Want to do nothing between.
Now go, let's go a little bit higher, like into the 100, let's
do 508 hundred like 1500. I was getting some really.
Want to add 1. I want to add 1 in here.
I want to see what 50 is betweenhere.

(14:34):
Yeah. Let's do this, OK?
Because I was running this last night and getting some really
funky stuff. Ancient warrior, that's your
that's your jam. Yeah.
OK. Your weird feels different than
my weird. I don't know why.

(14:56):
And no personalization on SO I. Don't need on like.
These are way closer to an input.
These are way like. They're all fairly close.
But look at these, even these colors remind me of kind of what
I'm getting with V7 and it's noton, you know what I mean?

(15:16):
Yeah. Like with the personalization I
mean. Sorry.
So where's that? That's 50 all right.
So here's 50. All right, now we're getting
like that one with the dude. Yeah, OK, now we start getting a
little bit weirder. This is 500 or No, this is 100.
So yeah, you start to get more variants in these, like some of

(15:41):
these are really interesting, you know, like you're not.
That's the this is the weird stuff that I'm talking about.
There you go. What is that?
So you'll get some, you'll get some things that work, and then
you'll get some things that are totally just out there.
Let's try one other thing Now I'm I'm curious.

(16:02):
Let's take this, let's take thisdown to two variables.
Let's take this down to like 10 and let's do well, let's do 3
variables, 10100 and 500. Now I'm curious if we had chaos
in here. Let's stack this right and let's

(16:23):
really see how this effects it too.
So let's go with something like A5.
What? Do you think 25 is it 2525 is a
good range and then? I think anything you think, OK,
all right, so let's do that. You get some crazy stuff with
that. Because I this is the other
component, right? Like you, you just brought this

(16:44):
up, which was OK. Well, now you're starting to get
more variants too, which is true.
So then it makes you think, OK, well then when would I use
Chaos? When would I use weird?
What if I stack them? Is there a nice little balance
of stacking those? So here's kind of what is this?
So let's start with, we've got alot of variations here.
This is Chaos 5 Weird 10, again staying with the award-winning

(17:07):
photo of an ancient warrior. This will stay and like we'll
get something like that and thisto me, if you go back to that,
this. Is pretty good like this this
little 510. This close up right here like I
like. Rerun that again.
I like that texture and that little painting on his face.
Like I might play with that guy that's A and I wasn't looking

(17:28):
for that, but it just sort of itsort of like brought me that
way. Like that 25 and 10.
So you got Chaos 25, Weird 10? And there's no.
If you're looking for control, these aren't the parameters.
Right. So let's so let's put this in a
bubble real quick too. What when would you want to use

(17:51):
this? You know, like so you mentioned
sort of like maybe it's a style reference perspective, maybe
it's an image like so maybe it'slike this image style reference
portion. Are you thinking of this being
used outside of that? If So, what comes to mind?
Not much honestly. It's like to me, it's like the

(18:13):
the mid journey playground, you know where it's like, let's get
away from the base model. Let's see what else it can
produce. Because if you're looking for
something unique, you're not going to just get it with just
stylize and base model. Like you're going to need to
personalize and the blend. I used to like doing this with
multi prompts also when multi prompts are still available
because you just you get stuff like like that.

(18:34):
Like I don't even know that that's a great, you know, you
start OK, like I see what's going on here.
Like I can play with this or this next image that you this
one. Like that's a cool composition.
I'm like, I'm really like that. So I I used like my explore page
almost, you know, like how you use the explore page.
Like I'll go in here and run weird and just see what it can
come up with because high weird and high chaos to me is the most

(18:56):
unique stuff that mid journey produces.
Like it's just not you, just no one's prompting that way.
Well, and you know what, we could always, you know, you
could take this one step furtherand use both.
So you've got Chaos Weird 10 andthen you could say, OK, let me
let me click the search within the search, right?
The other thing that's fun to dois run it through describe and

(19:16):
then once you see like what likesomething that's totally out
there and you're like, OK, what is what is mid journey thinking
with this one? You know that one, that one's
probably going to be relatively relatively simple in in terms of
like it's analysis of it. You know what this this is
actually interesting. Like is there a component of
reverse engineering what mid journey is thinking with these?

(19:37):
Right, because we put in just award-winning photo of ancient
warrior. That's it.
And then we dialed up just a little bit chaos and weird.
So then what are some similarities that we're taking
from this, right? Like we've got it's identifying
samurai and all these, for example.
Portrait Canon. So yeah, Yep, portrait like

(19:59):
it's, it's sort of like. Attaching that to a close up
portrait, it's you know, you seeCanon EO SR5 here a couple
times, Canon EOS R6. Like drop down, I was going to
say drop down to that one with the black and white dude.
I just want to see if it's like describes keep going that one
just once. Let's see how it describes the
composition if you describe it. This is an interesting workflow.

(20:24):
Yeah, this is this is how I've come up with a lot of things in
the past, so. Again, close up and and let's
look at these real quick. So a lot of these are close up.
Yeah. And this is without
personalization on, right? So this isn't any influence of
yours or mine here. I do a lot of portraits, so I'm

(20:47):
actually surprised. Like to me, this would be kind
of what I would get. I would get a lot of close-ups
by default. close up, close up. Style of hell, High Newton.
And you've and you've got that. You got him twice.
You got Richard Evidon once and then another guy or Mario

(21:10):
Testino. Testino interesting yeah 'cause
there's the other thing that's fun with weird is to run it on S
ref codes 'cause it'll really like start to if you have like a
a fun S ref code that's that's you know really you just like
playing with it. It's not necessarily like you're
going for anything other than playing with an S ref code, but

(21:31):
then it becomes something that that gets interesting because
it'll take your. Let's pick one.
I just did a couple the other day.
It'll take that style and dial it up a lot.
OK, So what do you want to do now?
Yeah, just prompt it and then throw the weird.
The same prompt that you're using with the S ref and then
throw and then throw the weird on it on the back end.
OK. So yeah, we'll do a, we'll do a

(21:56):
0 just to kind of like show the baseline there.
Let's do A50, let's do 100. Let's do a 500, let's do 1000.
Yeah, that works. Have you have you done any of
the S ref stuff with 7 yet? In In what?

(22:16):
In what? Like.
Just like do an SRS referendum unlock the codes?
Yeah, I've actually been gettingsome good ones.
Lately, I was going to say I gotit changed.
It's been hit or miss for me. Like there like when I first did
it, the like couple days after Iwasn't getting anything good.
Next time I did it, it was like all bangers.
And then the last time I did it was like kind of a mix.
It was weird. Like the second time I did it,
it was almost like every single 1 was really good.

(22:37):
I was like, damn, OK. I need relax to come back with
with it was like permutations because like it's really
annoying having to run everything in turbo just if I
want to do something like random, like I don't want to
keep hitting it a million times even like our time of our 10
works and that doesn't actually I don't think like.

(22:58):
Like that one's pretty dope. Like people take it and make it
something different, like still have some sort of influence of
the S ref, but it'll take it andgo in a different direction with
it like that. I think it.
That's that. One's a perfect example of weird
to me. What?
Are you prompting to get there? That's wild.

(23:19):
Yeah. These are just the skull, you
know? Yeah.
And so you just you can find some things and you can then
blend them in, use them as S refs like that guy with.
That see, that's what that's where it gets to me where it's
like, you know, I, I I want to blend S refs more, for example.
It's just it's not super easy todo it and what I mean by that is
it just it takes a lot of work yeah.

(23:42):
You know I got to go back and find codes and then you know
like this whole sort of like boxjust gets so jumbled because
you're like oh what do I have onnow like am I using 61 AM I
using 7 Do I have standard raw draft?
You know standard Am I using turbo?
Am I in relax? Do I you know, am I running
permutations because now I got to make sure turbos, you know,

(24:03):
like there's just so many thingsto think about and you know,
like that's what I liked about this, right, is like you have
this now up here really quickly.That's awesome, yes, but then
man, and like the S ref thing does need to kind of get figured
out. I wish there was a way to almost
like quickly access the ones that you like the most, kind of

(24:23):
like profiles in that sense. But then you know, like, I don't
know, there's a real UI thing. You and I were just talking
about that. But for me, I think the UI part
of this is starting to get really, really hard, especially
for people that aren't using this all the time.
And I think people that are it'sjust very cumbersome.
It's just a lot of things happening.
Do you have an S ref folder? Like do you generate within the

(24:44):
folder for your S ref randoms? Do I generate within the folder
for S ref randoms? No, no.
I. Don't that's been like my, my
biggest sort of like revelation is generating within the
folders. So like I just made it a folder
like in that top right corner. Yeah.
And you just, you know, you create a folder for S referendum

(25:06):
and then you just go into the folder and then just generate S
referendum. So they're all just.
It's just one folder of. That's it.
Oh, yeah, yeah, OK. But I mean, but I mean like
blending S refs, right? Like getting into the things
that you like to access the codes that you like.
Because like that, that to me isnot a problem.
Like the SRF random thing is nota problem because I, I do it
differently in the sense that I'll just do SRF random and then

(25:28):
I'll go through basically each set and the ones, the S refs
that I like, I'll just like 1 within that grid of four.
And that way, anytime I want to go back to the the different S
refs that I like, I've got them all here.
That's it. See that's.
So it's just a different way. But but I mean like when you get
into like blending, I think you do this more than me and I, I

(25:51):
know other people do more than me in terms of like blending
234S ref codes together. I'm just saying that part of it
it it gets very cumbersome and hard to like piece together the
ones that you like. Unless you're using the same
ones over and over again, it's too hard to access them.
It is. I I do, I do find myself

(26:12):
defaulting to the. One feel like this, feel like
this is like a Rory special right here.
Of course, anything that's just got a lot of chaos in it.
Like this is like graffiti on here.
This is like nothing. This is this is an interesting
code. Might need to play with that.
But like these are ones I haven't even played with.
These are some of the ones from V7 though, yeah, they wanted to
explore further. The juxtaposing stuff is where I

(26:33):
like to live, like, especially where it's like photo realism
and then art, like photo realismand, and doodles photo realism
and, you know, 3D, like blendingthose two, something that's
dark, something that's bright, something that's, you know, that
to me is where the magic happensbecause it balances itself well
instead of trying to like, yeah,like clash.
Yeah. 2 super, you have two super photorealistic styles and

(26:56):
then it's like trying to clash, you know, it doesn't like come
out the way you want to. But if you blend like something
that's like like sketch work andthen something that's super
photorealistic, like they blend well because there's, there's
different pieces to take from them.
But I, I just, I found like 20 of them that I liked over the
weekend. I probably ran like 60 and I was

(27:17):
like. Oh, it's a pretty good hit rate.
I know I was like this is because normally it's like 5.
Yeah, I know. Like you run the R40, it's like,
oh, you maybe get 3 or 4 out of there.
Then you're like, oh, this is cool.
They're definitely been, definitely been, I don't know if
they've been working on it in the background, but some of them
felt like more V7 looking, you know, V7's got like this way

(27:39):
different look to it than V6. Yes, yeah, the SRAF codes look
different. Uh huh.
Yeah, I. Don't know if it's just
applying. V No, I was thinking the same
thing though. I was, I was actually thinking
that exact same thing. I was like the the something
about the application of the SRAF code just looked a little
bit higher tier. Yeah, it wasn't just the image
itself. Exactly.

(27:59):
Have you noticed? Have you ever tried to just like
put in a random number like SRAFlike?
I have. Like 429.
Well, I, So what I was doing like when they first came out is
I was, I remember they mentionedit was like 4 billion or
something. So I was like hitting all the
like the, the milestone ones, I would go down and hit like SRAF
1,000,000, SRAF 2 million or two, 2 billion, two billion,
like all those kind of numbers just to kind of see if there was

(28:23):
anything like an Easter egg in there.
There wasn't from what I remember, but, you know, it was
worth a shot. You know what I'm wondering if
we should talk about this, this editor, like, yeah, you know,
this has been interesting. And I'll, I'm going to defer to
you a little bit on this. I, I have a feeling you've
played with this more than I have to this point.
Yeah, this is. So I'll, I'll kind of lay the

(28:46):
foundation and then I'm going totee this off to you and just
kind of would love to hear your perspective on it so far.
So Long story short, right, You had editor, you had the full
editor, I think, and maybe you and I just didn't know this, but
the full editor I think was limited to a certain tier and
now that's available to everyone.
I kind of forgot about that. You and I are on the large plan.

(29:07):
We're paying shit ton of money, so whatever.
But now this is combined into one place.
I always thought that was a little bit rough around the
edges to have to access those two things in different places,
right? And then there was the component
of OK, edit retexture. You know, there's a lot going on
here to keep track of whether you're an edit or retexture.
And then you couldn't see the results from here unless you

(29:28):
went into this folder or you upscaled it and then it would
appear in your folder, right? So there were just a lot of lot
of things happening here. Now you've got all this in one
place, right? So you've got edit retexture,
you've got suggest prompt in this little icon here.
Move, resize, right? So this is your scale, this is
your aspect ratio, this is your paint, right?
Just like before. Now you've got smart select in

(29:52):
here too, which was that last feature they dropped in Rory.
And then here's like the other part of this, the layers, right?
This is where you can add an additional photo to this.
And this is the workflow I haven't really played with yet,
so I would love to hear if you have.
But what do you think of this? Any takeaways so far?

(30:12):
I think it's a good starting point.
Start like starting point like Ithink they, they could probably
make this a little bit cleaner in certain areas.
Like the layers thing gets a little clunky when you're just
maneuvering it, but it's fun to play with because you the thing
that that you need to, if you'regoing to use the layers, it's
almost like you have to use layers in conjunction with
retexture to like smooth it all out because you can do, you can

(30:36):
blend things together, which is fine.
Like it'll work. And then, you know, if you're
going to use a layer, like if you're just going to add
something here, like add the brain, right?
The one thing here that I don't that I don't like is like, say
you wanted to put that brain on a say you want to like maneuver
that brain on top of the skull, right?
Like to make it fit the, the dimensions of the skull.

(30:58):
Like if you want to like, you know, like, OK, yeah.
So you can't rotate it. Yeah, you can't rotate it,
That's right. So like that, that to me is
something that's simple that would help is just being able to
rotate that or, you know, drop the opacity on it or something.
I know that like this isn't a big thing for them.
Like I don't, you know, I don't really, I'm just thinking in
terms of Photoshop. But like if you want to do like

(31:21):
the smart select feature to me has been really easy and nice.
Like if you just go to the if you go to smart, select right,
smart select and then include and then just like click on the
brain, like the brain itself. Now, if you do a race

(31:41):
background, it'll isolate it. So there.
You go, there you go. It's So you're just going to let
me waste my time brushing aroundthe?
Brain, I tried to catch you before you started doing that
because yeah, because that's like, you know, something that's
super easy if you did the other way around, right?
If you just hit the little back button, hit the, you know, the
undo up top left. Yeah.

(32:05):
Now if you just, if you hit, youknow, erase selection, then
it'll take that out and you can keep the background so you can
take people out, put people in. That was that was a little less,
what's the word I'm looking for?It was a little more confusing
in the last iteration of this, you know, smart select, but they

(32:26):
made that very easy because thenyou can start to stack these
layers and then like, let's justsay if you go or let's just do,
you know, erase background againon this, right?
And then just erase background. Great.
Now if you if you're just going to go hit generate here, like if

(32:46):
you're not going to like not much is going to change.
It's going to paste that thing in.
Yeah, yeah. And then to your point, now
you're going to have to go and put a texture style or something
on it so it blends together, right?
Yeah. So if you did that, like if you
went and then you know, if you go to retexture now and then I

(33:08):
don't know, dial up the dial up the style weight maybe.
Yeah, I was going to say, do I need to go ahead?
Do I need to reapply this? Let's try it once without and
then let's try once with dialingthis up.
So Southwest 100 by default, let's try like a 250, Let's try
on the on the reverse end of that, let's try like a like a 50

(33:31):
just to kind of see the effect here.
OK, so this is the 100. So you don't have to like by
default, you could just hit submit and it's going to apply
that. Yeah, so it didn't even apply
the brain in this, which is why I found it to be a little bit.
They already added it like a flower.
You might have to prompt for this in the style prompt in the
in the. Text, right?
So like maybe what we do is this.

(33:52):
Yeah, suggest a prompt. Cartoon one, you might want to
use the original if you go to the Yeah, that this is where it
gets confusing, right, is that sometimes you're just selecting
like different. You're like, oh, let me edit
this, but it's not the thing youwant to edit.

(34:12):
You'll edit one of the actual variations instead of the actual
image that you input. Also see this is and this is
given the interesting, this is giving the suggest prompt just
on the brain too. So I might have to do this.
Well, I can't click both, can I?I might have to click this and
then let's see if this changes it right?

(34:37):
Let's see. Yes, it does.
OK, so now with that in mind though, is this going to carry
that layer over because it's notselected?
Let's put see. And now we lost our S ref code
too. So this is where it's difficult.
This is where it's like a littlebit clunky.

(34:59):
Let's just go ahead and submit that then.
Without any reference the brain.It might not.
It might. Not well, it does.
It did carry the brain over here.
Let's see. But there's no call out of it,

(35:21):
right? So it didn't like apply it
heavily, but you could kind of see this here.
That's it. One out of four.
Try one of the, try one of the instead of the initial image,
try one of the variations where the brain is in like down, down
one. Yeah, like click that, like use
that one and then use the suggest prompt on this.
So let's see what it does, Skeleton.

(35:46):
OK, so glowing. So like, let's take pink out.
Let's just do a glowing brain. Or if we don't want glow, yeah,
we'll leave glowing in there. Cement re texture.
And then we'll see what we got here.

(36:08):
See this is this is where it's like you get no video, right?
Like no nothing from mid journeyexplaining exactly all the
variables that exist in here. So this is where it's like,
unless you're listening to something like this or you're
going around and spending the time doing this, you have no
guide, right? Like this is where it's just
like, well, OK, I got to figure this thing out.

(36:28):
There it is. So that's you get the brain in
this one at least, right? Yep.
And that's where, you know, that's where having the little
rotate feature could be nice. I mean, I don't, I don't need
opacity or anything else. Just like having like a, a
feature for something because I was playing around with like
colleges, that's what I was doing, you know, just like

(36:50):
stacking, like you know how people do those like cinematic
stacks where it's like 3 shots. And I was just like, oh, let me
put these in there and I'll erase some of the edges and then
I'll like throw in a style code to make it look like it's, you
know, so it's designed instead of just like three boxes, right?
Simple things like that. Because I was playing with the
layers was also doing some like,oh, here's a landscape and let

(37:11):
me put this, you know, this giant creature in here and like
sort of like, let's blend both of them together.
So let's take this creature, let's take this landscape.
I'll cut the creature out, put that in there and then like
retexture it all so it looks thesame.
And it's, you know, I'm not going to use this thing for
editing that much current in itscurrent form, I don't think.

(37:33):
But it's nice to have it like, Idon't think, I don't think it's
like groundbreaking in terms of what it can do, but I I do like
having the optionality to do it,if that makes sense.
Yeah. But this I haven't, you know,
never really like, loved. It's clunky retexture.

(37:54):
Yeah, retexture's good. It's not great, doesn't.
I'm not using it for that though, you know.
But it's not. Yeah, it's, it's, yeah, it's
interesting. Like this is the other part,
right, where it's like you got to make sure you're on edit, not
retexture, even if you're resizing, right?
Because of the first time I ran it, I was just still in the
retexture tab. And then here we go.

(38:15):
OK, no, that's not what I want. I actually wanted to expand it
with the same thing. Yeah, I agree, man.
I mean, the retexture thing, it's got its place.
You know, some, some of the things are, are better than
others. I.
Think this was just like an appeasement, you know, or you

(38:38):
know, it's like you want to add something in.
So this was hold on, let me justlet me do let me share this the
other day this was what I was I was doing this was a little
still a little unlock with this.I don't want to say unlock, but
it's a let me see if I can find it here like this was what I was
enjoying it for. I wasn't really doing anything
crazy with it, but you know thathaving the smart selection

(39:01):
feature in there makes it so much better for things like this
where I was going and doing where are this is just not going
to hold on. Maybe there like something like
this where I was just taking theshoes out of an image, right?
Like, oh, and then instead of itjust being a, you know, product

(39:22):
shot, right, I was like, oh, letme add some feet to it.
So you can then change from likeiteration here like this, just
taking the shoes, putting them on different people like that
was a really simple one that this is where, you know, I got
screwy. But this was interesting and I
found I was using, I tried V61 like a different same image,

(39:42):
right? V61 did so much better with this
in the editor feature. Yeah, V7.
V7's like way too much and I don't have profile on or
anything here, but V6 one's justmore coherent to sort of like
the idea that I was going with. But you can, you know, you can

(40:03):
probably put a product in here. That's probably, that's probably
where it's like one of those things with this mood board
stuff where it's like they're using probably the 6.1 engine or
whatever the hell you want to call it temporarily until they
kind of finalize the specific version for V7.
That's what it kind of feels like, right?

(40:24):
Because you could tell to your point, it's it feels a bit
broken or disconnected compared to 6.1.
You could tell this this featurewas built for 6/1, right?
Yeah, just based on the results that you're getting here.
V7 V 6.1 like there's a big difference there.
Same prompt, nothing different, just V6V7, right?

(40:45):
I like the smart select idea toofor an object replacement.
It's simple, 'cause it does, it does like the thing that it did
here without prompting. I didn't have the I didn't have
the mood board on that I used tocreate these shoes, right Is
that it matched It took the style of the shoes and then
matched the entire image, but like it took that smoky blue

(41:07):
from that my atoll mood. Board and was that was that with
a with any kind of style reference or mood board on when
you did that? No, no, this one, when I created
this initial image, it had the, it was just like a pair of
sneakers in like an environment.So I just erased the environment
and then I put it in here and itmimicked sort of this look and

(41:27):
feel without having a mood boardattached to it, which was nice.
Which is interesting. I'm surprised it did that well.
Like it just took that that shoecolor style applied the entire
aesthetic across the image to a cool image.
And then like it took like I, I liked how it took the like the
paint splatter. Applied it to.

(41:47):
The shoe and sort of applied it here, like that's a creative use
of it. Yeah, like this one.
The good thing about it is because it gets the, IT nails
the shadows, right? Like it nails the shadows around
here. Like you have the shadow, you
know, underneath him. This guy is shadowed
appropriately. The lighting is appropriate.
This one here, same thing you have the shadows that come
through here based on sort of where the light is.
Everything's consistent. It took this sort of gritty

(42:08):
texture like I like how it just added this guy's jacket, you
know, and sort of match the color of the shoes.
Same thing here, like you get the shadow of him and the shadow
of the shoes considering there'snothing there to start with.
So you know that in Photoshop alone would have been insane to
do something like that. Like we look at it and criticize
it for some of this stuff, but like to do just to do this in

(42:31):
general from here to here is wild.
And like, you know, it's some ofthis stuff might be a little bit
inconsistent in the shadowing, but realistically like it works.
So I was impressed with that using 6.1.
So maybe, you know, sometimes just a little tip, maybe sevens
not the best with editing just yet.

(42:51):
I think this one was 6. This was 7.
This one. This one's sort of.
That one you can feel like the quality of the image itself
feels like 7, but like the the structure and composite, like
the whole his legs are just, it's shattered, you know, like
they're just, they kind of wreckit.
But the rest of the image is really.

(43:12):
It's also tough with. How close the shoes are, right?
Like it has to sort of fit the legs.
It does, it does there. So that might not be the best
example this. One, and that's also where you
know what, what's interesting there is I wonder if you change
the aspect ratio if that's goingto help you out.
In this one. Yeah.
Because to your point, where those legs are, you know, mid

(43:33):
journey kind of does this like it should kind of act as if the
aspect ratio isn't a barrier in that sense and it should just
proportionately put the legs up even no matter where the crop
is. But mid journey kind of has a
tendency to almost like want to fit the crop slightly.
And I'm just wondering if you were to take that, yeah, put it
into like a 3-4, even like a 916, how, how better those legs

(43:58):
get? B 6.1 Let's run this.
And also, you know what I see? Like you'll never see shoes that
size usually in a 3-4, yeah, unless you were prompting like a

(44:20):
like a close up, like extreme close up angle of shoes and
you've specified it in the prompt that you like the crop
comes up to the waist or the knees or you know what I mean?
Like it's product photography orsomething.
And so that's that's get gets into one of those areas where
it's a little bit. This is a tough ask I think for
mid journey but let's kind of see what happens.

(44:43):
Nailed it there. I mean like that's a little
weird physics there. I think that looks pretty good.
This is. Like a that feels like a wide
angle lens maybe? So that could work like that.
Curious about the standing up? Let's see.
Like that fits better, yeah, butthe legs are.
Like perfect. Yeah, this one, it's a weird

(45:06):
composition. This one weird composition.
I also probably have standing. I don't, I have standing on a
wet street. So like that one works
definitely. What if you put in instead of
lifestyle? What if you put in product
photography? See here 9/16 This looks better.
Already. Yeah, yeah.
So aspect ratio could be if thatlooks stiff, that looks a little

(45:29):
more natural. Yeah, this one definitely looks
a little bit more natural. Let's see. 77 just wants them to
sit. But the quality on this is
really good. Like that's, there's a lot of
stuff going on there. Also lifestyle tends to put in a
person they they want to see a person in the frame.
That's why I'm thinking if you switch out lifestyle for

(45:50):
product, I'm wondering what how different the results are.
Product Photography Submit Edit STB 6.1 Let's see so we have our
9 by 6. So this is, you know, you can do
this with a lot of other products.

(46:11):
This doesn't have to be sneakers.
Right, right. Let's.
See is this one? Oh, we did this one.
And again, I think this is this is probably one of the harder
asks because of the shoe size and where it's at.
Yeah, this is difficult there. See, it's like just when that's
probably why I was doing lifestyle because when I was
saying product, it was adding a shoe that one.

(46:32):
So 61, again, a little bit more consistent in terms of what
we're looking. Forward to the yeah.
And then, yeah, I mean, I'd say,I'd say 6.1 to me is more
consistent. You might be able to then sort
of like retexture it if you wantto, Like let's just see if I
want to let's see like one of these, like maybe let's

(46:56):
retexture this with V7 and P, see how that works.
So that gives it a better interesting aesthetic.
Or if it just sort of mutes everything.
Sometimes retexture sort of likemutes everything for me.
I don't know if it's the same for you, but you lose a lot of
detail. Well I you know what, I always

(47:16):
if I'm going to use retexture, Ialmost always use an S RAF for
us like either it's a code or anS RAF.
Like this, like, you know, like these actually look pretty
decent. Yeah, like I like it.
It didn't. It's not the same color scheme,
but I'd probably just have to add the mood board back in there
so. So then you could take that like
take that one with that like that.

(47:37):
I like that furthest from from the left.
Do the all left one all the way to the left.
Yeah, like I like there's like alittle graffiti in the in the
check. Then if you were to apply any S
ref, whether it's an image or a code, let's see, I bet it would
crush. Let's try first, Let's try.
All right, maybe this is a fun one to play with.

(47:59):
We'll add this and then I'll add.
Let's see. Let's see how we let's see how
we do here. Now this is now this is mood
boards though, so I wonder how that's going to affect the

(48:20):
quality too, because you know, the mood board stuff still a
little bit. So I gave it a little like I
gave it a tough task there because if I go and instead of
this one, let's just use let's go with the cosmic start with
this one. Do that.

(48:45):
Let's try this one. Let's do this one.
That's OK. Go back and take one of your S
ref codes that you just unlockedto that you like and apply that

(49:07):
all. Right, let's see.
Because I think that's going to,that'll probably give the best
result. Actually, let's just go in here.
Or take any of those images and just apply it as a SRAF image.
Let's see where the SRAF codes. All right, let's see what we got
here. Just try.

(49:30):
Let's try this one. See if this one's good SRAF.
We'll copy this. It's back to edit, and then
we'll just add this one in here instead.
Turn P off, Submit added, no changes.

(49:52):
Oh, it went to instead of Recheck.
There you go there. You go there it is.
All right, so let's see what we got here This.
One and then this is where you could apply the S weight, right?
So you play style weight to thisand really dial it in.
So here was our little gritty neon.
I have to play with that. Yeah, I screw with that one.

(50:12):
That's not bad. So it's applying the mood board
pretty, pretty well. Here.
This one, my darker profile, it's definitely really dark.
Not bad. This one has like all the
elements of so re texture. Not bad with not bad with the
mood boards on V7 like it's applying the style
appropriately. This one.

(50:32):
I wonder if we do one more. Let me just do you find my Pope
images here? I think the Pope actually just
passed away today, or at least overnight or something.
Like, oh really? Think so.
I think that happened where let's see, because.
There was a yes, you, you're right.

(50:55):
RIP to to a wonderful piece of art.
He waited. He waited till the day after
Easter, yes. Did he, did he make it?
It's kind of like very, it's like very karmic, not karmic.
What's the word I'm looking for?It's like serendipitous on
Easter. But here's the let's see, I got
to go back to edit. See, this is where even we get a

(51:17):
little screwy. Let's try this.
Let's do this one on the course.I can do the retexture, submit
the texture. This one's got a fun little
lifestyle slash doodle componentto it, and we'll see how this
one comes out. Yeah.

(51:40):
Interesting. SO I mean, yeah, you look at the
you look at the difference in adding mood boards.
It definitely it definitely works in retexture, but I think
that this was a this was interesting to sort of go
through because you can you can apply this at many different
ways for a lot of different. It's not going to probably work
with a bunch of things that havea bunch of text on it.

(52:01):
But you know, if you had like a credit card, like I'm just using
that as an example because it's pretty simple, right?
If I get credit card image, you can just like throw that in
there. It's like generate backgrounds
into Infinity. You have a, you know, a, a mug,
something like that. You can do, you know, these
simple things that can be super useful in terms of assets, most

(52:22):
likely. That's just my feeling on it.
Yeah, but let me stop sharing here.
Yeah, so I mean like so we went into weird, we went into editor,
like deep dive. I think weird is worth re
exploring again, editor, the full editor.
It's nice that they're all together.
It's nice that everybody has access In terms of the layers
themselves, that part of it feels like there could be more

(52:47):
tight, you know, like consistentcustomizable process that exists
there still, right? Like I'm I'm sure you can do
great things in there. It's just like, do you want to
spend the time to go about it that way or would you rather do
it a different way, right, whichis all what kind of what we're
all talking about in mid journeyanyway.
There's always this line of thismakes sense to do here and it's

(53:10):
fun to do here versus like, can you do it here?
Yeah, but is it actually worth doing it?
Yeah. So.
I mean, that's where I'm just like, I like having it.
I don't necessarily think I needthem to focus on it personally.
I think, you know, it's like, doI need them to have a rotate
button? No, I'll just do this a
different way. I'll just, you know, do that in

(53:31):
Photoshop or Figma and then bring it back in there and just
retexture instead of trying to like screw around, right?
So. 100%. You know, these are these are
the little things that are goingto keep dripping out though that
are just going to keep like, youknow, it's not going to be like
life shattering updates, but something to see what we can,
you know, play with and start tolayer in and feel out, you know,
feel out how it works. So it's good that I don't have

(53:53):
to worry about this and with thewhole V7 at the same time.
Yeah, agree, agree. You know, and there's so there's
a lot of other stuff here like I'm wondering if we just kind of
go through these notes to kind of wrap this up like in terms of
office hours, because there was a lot buried in there.
I'll I'll just kind of like readthrough some things.
You tell me what you want to kind of focus on whatever pops

(54:13):
out. I do think this is probably
worth it. Quality Q4 coming into the fold.
We haven't had AQ four in a longtime.
It was like 5 or 4 maybe. Yeah, yeah, yeah.
So this is under active experimentation, but it's the
idea that you're going to get higher quality for slower
renders. Yep, This has got this has got

(54:36):
our names on it because I don't care about I I would rather have
the top tier quality and wait a day for it, you know,
hypothetically, then get something that's half as good
that I get instantly. So this is kind of like one of
those things where when Q4 drops, that feels like something
you're going to want to play with, probably going to get all

(54:57):
the micro contrast, tonal contrast, detail structure,
sharpness that you could ask for.
So that's that's on the horizon now.
They're playing around with somethings with obviously character
reference Omni, you know, like that whole Omni reference thing
still feels like that's a coupleweeks away.

(55:18):
There was a point here about S ref Super S ref plans to develop
a significantly improved Super Sref system.
What do you think that means? Don't even know but I like it
because S ref's to me like the best thing that that mid journey
has. Like it's so it's so good, so
much better than most of the other tools.

(55:39):
Like I think if there was like adifferentiating factor outside
of creativity and just like visual explosion, it's like SRAF
to me is just better than every other SRAF.
I I I feel that. Yeah, like I feel that gap in
terms of like where the next tool is and what you're able to
do with it 'cause most tools youcan't blend.
Like Gemini just released blending styles, right?

(56:00):
Like what was it a year ago? Almost probably if you could do
that. So it was like take these two
styles and then sort of mash them together.
Like, yeah, you can image promptthem, but not like style plus,
yeah, yeah, plus control and then apply both.
So I think still to my to me, myfavorite sort of big
differentiator versus all the other things.

(56:21):
But it's just me, man. I don't know.
How do you feel about that? Yeah, I I do too.
Like I, I, I don't know what it means.
Yeah, I still would like a library of some sorts like so
kind of what we were talking about, it'd be nice to have some
sort of organization or library of S ref codes that you like and
use. Now there there are a couple
ways that you can go about doingit right now, right?

(56:43):
Like going back to that folder system that you talked about,
you could essentially set up folders specific to S ref codes,
you know, like, so if you've got5 to 10 that you like to use,
you could take the time to set up specific folders for those,
right? And then with that folder
integration that they launched to prompt within those folders.

(57:04):
So everything kind of stays tight.
But I mean, like in terms of like grabbing them and applying
them and infusing those into theprompt box, that part of it
could be easier. It would also be cool to have
some sort of like maybe searchable S ref library, right?
Like based on style, like so you, you know, those third party
sites that exist, right? Like they've got a bunch of

(57:25):
things that people have unlocked.
It'd be cool to like go through and say, OK, I want to go into
this kind of style and look for S ref codes relative to this.
That would be cool. Maybe just an S ref explore
page, like just add it instead of like, you know, just top hot
whatever. And then some like S ref
category and then all people that are generating with codes
like. That like that.
It's it's easy like it's a good.I'm sure it's a filter.

(57:48):
You know, if you're putting S ref random in, right, just like
whatever, whenever someone hits at random, like upload it to the
thing and then we can just sort of scroll like our own dedicated
S ref code feed. To me, it doesn't seem fun to,
to develop that. But it's to me that's like a, a
good first step in terms of it 'cause then then it becomes just
like, all right, let's go, let'sgo in there and find, you know,

(58:09):
let's go find stuff. Cause a lot of stuff you think
on the explorer page is going tohave S ref codes.
It doesn't. That's Yeah, that's true too.
Yeah. The other thing they they
mentioned was a lot of just, again, like going back to this
improvement in the model that just takes place over time.
Improved prompt accuracy, anatomy, artifact reduction, a

(58:32):
more aggressive aesthetic updatepending user feedback.
I don't know what aggressive means, but I do I, I hope it has
something to do with moderation because I've been getting
flagged like a mug lately. And I'm just worried that like
all of a sudden I'm going to be locked out of my account because
I used the word skin or something, you know?

(58:52):
And it's just like, yo, I'm justtrying to like I was trying to
do something with like a girl biting a match and like just the
way that I was trying to describe it and then using my
tool to describe it. It was like it wanted to throw
things in like provocative or skin.
And all of a sudden I'm just like getting flagged.
Like I can't even generate the image.
Definitely can't use it as an image reference.

(59:13):
And some, you know, some of these cases too.
So that part of it that could beimproved.
They mentioned something about acreative mode.
Rory considering introduction ofa new quote UN quote creative
mode if necessary for increased user control without excessive
sliders. So maybe they're they're taking
your feedback there. They don't.

(59:34):
They don't. Maybe it's just like more
prompt, more focus on the text prompts, maybe less, which
sounds the opposite of creative mode, but I'm.
Yeah, yeah. Not trying.
To think about that. I like it for increased user
control. So increased control, but
creative. So yeah, it's kind of a, an

(59:56):
interesting dynamic there. I I don't know how that works.
We don't, we don't have many details.
It's just off sour note guys. So we we're taking guesses,
yeah. We missed the context for
everyone that needs context. We we missed the call last week.
I at least I did. I only caught a few minutes.
I only caught a few minutes, yeah.
So we're missing contacts. We're just totally riffing on

(01:00:16):
this right now from a from a discord a post.
We, we are a new ultra fast draft mode.
So I mean, it was already fast, don't really care too much about
that. Let's see exploring innovative
real time user interfaces enabled by instant speed models,
good aesthetics maintained despite ultra fast generation.

(01:00:37):
OK, video. They dropped a video update, so
video model training completion expected by the end of the
month. So was that like 2 weeks away?
Yeah, tuning and and release in one to two months.
If you don't have any pickups later, yeah, let's let's expect
that in like August or Septemberprobably, maybe.

(01:00:58):
I would rather wait it and and let it be extraordinary, you
know? Don't need to ruin a good thing
right now like V7 is hype. They don't need to.
They got the the unnecessary blowback for V7 because people
were just like again, it's just quick hit hitter reactions that
don't know anything about the tool.
It's just like it's a brand new architecture.

(01:01:21):
You have to figure it out, not just.
It was a perfect storm. You had the the long wait for V7
then you had ChatGPT drop in their image Gen. update.
Like, what was that couple days a week before?
And then you had that right? Like just the the expected
instant reaction from people that probably put in a few

(01:01:41):
prompts didn't get what they want, right.
OK, well, did you bring enough images?
Did you explore enough with all the different variables?
Right? Because now again, does weird
mean the same thing? Does stylized mean the same
thing? Do aspect ratios render the same
kind of images? You know, like all those
different things, you have to sort of like reassess and there
wasn't that happening. And it's the other thing too

(01:02:03):
with that right, is just like I've seen, you know, to people's
point, I don't want to rank 1000images.
It's like, I get that. That's, that's fair.
That is fair. Do you want the best experience
with the tool that can do the probably the most or most
variety of things? Like, you know, that's, it's
just like a sometimes you've gotto make a little sacrifice

(01:02:24):
depending on and I can totally understand why people be like,
I'm not doing that. You know, it's just like that's
that's a non starter for me because I don't feel like
wasting time and only good fine,like they they can, you know,
you can use other tools on, but it's to me that's like you put
your time in there and you get what you need.
So it's just like again, I get Iget so turned off by everyone

(01:02:46):
that just like dumps or or it's just like the hot take of
hopscotch like this. This is dead.
I was just about to say I'm so tired of that.
God. So mid journey is dead.
RIP mid journey. RIP Photoshop.
You know, like I saw that one the other day.
I'm just like, are you kidding me?
Like you're literally just engagement farming.

(01:03:06):
There's no you tell me you don'tbelieve this and please tell me
not because otherwise I'm just going to like absolutely clown
you because that is just ridiculous to make a state use.
Photoshop. When people say that I'm like
that, I don't know if you ever used it.
You probably not because you would know there are 1,000,000
different things that you could do in there.
Some of these things are now making them, making themselves

(01:03:29):
over to something like mid dreamwhere you can't accomplish these
things. Again, though, it kind of comes
down to well, are are you already?
Did you build that muscle using Photoshop?
It's going to there's a stickiness there that would have
to take place right for some folks to then move away and go
to a different product and use use that same thing for a
different, with a different tool.
Like that's still a thing the adoption anyway.

(01:03:52):
We, we are, we kind of know what.
That is but been trying to get off Photoshop for 10 years.
I still can't you know? Like I want every reason to not
pay for it's. Still in the dock?
It's still in your dock at the bottom.
Right, I just like. I can't get rid of it because it
still can do things that no other tool can do.
So like, it doesn't matter to me.
Like they don't need to be #1 inAI, they still have very good

(01:04:13):
functionality based on what theycan do.
So Adobe, shout out, still a still a customer.
Don't hate you guys, do hate youguys sometimes, but.
But also, like, they, they have their, their pockets are deep,
you know, like, they're, they'reworking on stuff.
They're going to be fine. Photoshop's not going anywhere
for the foreseeable future. Too many products.
I think that's their problem. I I agree, agree.

(01:04:34):
That's that's always this the case when you get to these like
big companies is instead of it being just like very tight and
succinct and you know what it isand like just focusing on
consistency and quality, they just expanded, expand.
And that's kind of like the problem with this VC or public
model anyway. It's just like they want more
revenue, more products to get the revenue, you know, expansion

(01:04:56):
on the products to get the revenue.
It's like you look at it from a just like a purely logical
standpoint. It's like how can Adobe keep up
in AI when they have like 50 products and 50 products that
have 10,000 uses each and like 10,000 functions.
It's like, how are they going to, how are they going to do
that? It's just so much.
It's different than like Figma where it's like Figma has one

(01:05:19):
product and they have like 3 different things you can do in
there. It's like Fig Jam.
It's the design files and it's aslide deck.
Like that's it. It's not Indesign, Illustrator,
Photoshop, Premiere, After Effects, you know, Lightroom.
I know those have a million functions, so.
And then there's like 10 that you're not even thinking about.
There, I know there's a million,yeah, those are just the ones

(01:05:41):
that I use, right. And then it's like, yeah, they
have, you have all these products, but not even to
mention the features that you then have to in it.
It's like it's like mid journey now, like trying to get to a new
model because they have all the parameters, all the different
functionalities. And now you got to bring all
that stuff over, yeah. And like Reeve's just like
everyone's like, it's awesome. Yeah, it's good, but it only has
to do text image right now. But are you also hearing about

(01:06:03):
Reeve like you were a couple weeks ago?
Nope, Chachi BT sort of stole a.Lot.
And so then it's like, you know,there's nothing wrong with
Reeve, you know, like, in fact, it was like that was one that
if, if I had more time, I would love to spend more time there.
But you're not hearing about it.Like there's the staying power
of being early and being the muscle that people have, have

(01:06:23):
sort of like built over time. And that's with like ChatGPT.
It's like, is ChatGPT still the best model?
I, I have no clue. I don't even keep up with the
benchmarking or what's going on anymore.
I just use it. It's there, right?
It's what I, it's the daily driver.
Same with mid journey, like I think ChatGPT is sort of that
whole image. Jen has changed some things that

(01:06:44):
I would maybe thought about using mid journey for in some
cases, right? And I think it's stolen some
market share away from some other places too, like a
Photoshop. Like, I mean, the ability to go
into a thumbnail, for example, or image and take a night, take
something that's in the background, isolate that as like
its own image. What?
Like, that's crazy. That's crazy stuff.

(01:07:06):
That sketch to image stuff to meis still nuts, like sketch to
image stuff so well, yeah, we were using like control net
before where you your sketch hadto be good too.
Like it couldn't just be like general idea.
I mean, it could be, but like this one, this one just works,
takes your direction and sort oflike has a creative level to it
as well. So that's that's what you know,

(01:07:27):
again, Reeve really good image quality.
I like it. I'm not going to use it every
day. I might use it for certain
things. But like, you know, I think you
you what you gain and sort of that good look, you can also get
it in mid journey now and then play around with it a million
different ways. There's only so many things you
can do. We're I think we're all looking

(01:07:48):
for, we've talked about it a lot, but we're all looking for
those instead of the 20 tools, we're looking for the three
tools, you know, And so the onesthat start to bring these things
into one place are going to accompanied with right, like
again, like the adoption are going to win.

(01:08:09):
Like the fact that ChatGPT was already a place that you were
going to show up and how many people are using it?
Like an insane number of people are using 500 million or
something. It's like, OK to then bring in
some sort of like image Gen. component that's reasonably
good. That's a huge lift.
Reeve on the hand, the other hand could be really, really
good. But what else does it have?

(01:08:30):
You know, like is it just another tool that I have to go
to? And then how am I going to
separate that in my mind from using that versus mid journey,
that versus runway frames or that use versus image Gen. or
whatever the tool is? So you know, like at at the same
time, it's like who's going to win the video part of that too,

(01:08:51):
because video like if you could bring video into, let's just say
hypothetically, like ChatGPT or open AI had Sora, but it feels
like very disconnected as a partof the workflow compared to the
other things. Maybe that's just because that's
maybe that's just to me, I don'tknow.
But like it feels different thanthe image Gen. part of it does.
That's not so good. It's right there.

(01:09:12):
It's mid journey. If mid journey launch is a good
video model, I'm probably not using any other video model.
Maybe I'll have the circumstances where, yeah, like
runway's good for this or cling's good for that still like
that could exist and maybe that's temporary, but like the
ability to bring that into a tool, I'm already using the out
of going to use that. So that's where this is

(01:09:32):
interesting, this kind of like fight for market share.
Here's a here's a tool that I just, I'm going to show it
because I think it's awesome. I've been playing with it.
Which tool? Recently it's called Flora, but
it's for it's for basically if you are building workflows or
trying to do things, you know, alittle bit a little bit more

(01:09:52):
systematized like it's great. Like this is what I was playing
around with yesterday, which I found to be awesome, which was
it has Gemini 2.0, which can nowtake images and blend them
together. So like this multi carat.
So I can basically upload an image and then I have them run
in like a node based fashion into the same into the same like

(01:10:14):
So I have a multi image reference and then gem and I
will produce with both of them in the picture, right.
So you can have multi character and then with gem and I you can
also then iterate right? So I can take it and then get
different scenes for this shot type or for this, you know, for
these two characters, right and get the over the shoulder
looking at her, you know, over her shoulder looking at him and

(01:10:35):
get the the burger that matches the table like the table here.
You know it's going to match thetable here.
So to me this is great because then you can also you can
basically build a storyboard andthen add video.
So it's pretty it's a pretty awesome sort of experience in
here. Look, I did this with a bunch
of, you know, with a bunch of the car stuff where it's like,

(01:10:55):
all right, let me feed this mid journey image in here and then
have it create different angles,right that we did.
I did this a while back, thoughtthis was awesome.
This is like the new stuff that I was playing with, which was
great, just like taking the shoes from before, right?
And then saying, I want to put him on this guy and then we
start there. And then you can, you know,

(01:11:16):
begin to iterate your product shots.
So this is what I'm hoping Omni reference starts to look like,
right? Even something like this, you
can add an LLM into the mix where it's like, let me take
this shot, describe the composition that I want and then
like run that in to an image generator in Gemini 2 point O
with this as a reference image and sort of like take this

(01:11:38):
composition and put those shoes in it, right.
So that's like there's a lot of stuff you can do in here,
especially if you have some someworkflows sort of dialed in
already, like this one was just a basic, you know, describe this
and then attach this and then basically generate this.
So it's cool, man. Like I've been playing around
with it a lot as of reason to see where I can go with it.

(01:12:01):
But I think like this whole nodebased sort of experience,
bringing everything into one workspace is nice and I think
it's like fun to play with, especially if you're you're
doing this stuff all day every day like we are.
Yeah, there there's some interesting things there.
I I mean the OK, so a couple different things.

(01:12:22):
One curious to see. Yeah, to your point, Omni
reference drops character reference, right.
Like then again, when that if that happens in a good enough
way, you probably won't use that.
Yeah, right. But here's the interesting thing
though. I really like that cohesive

(01:12:43):
workflow of the image to the video right there.
Yeah. So then that gets me thinking,
what's that experience like for mid journey?
Could you then take, what was that whole, what was that
patchwork? Could you take patchwork and
apply that like like some version of that into this,
right, where maybe you give people two different options to

(01:13:04):
play with? There's the existing UI that you
have, right, where everything kind of goes into create and
then you can switch from maybe photo to video and blah, blah,
blah, blah, blah. Maybe there's a way to tag and
you know, kind of like consolidate these things into a
project based sort of like organization or file.
Maybe there's also a patchwork style canvas like you just

(01:13:25):
showed, right, that could apply here.
And they've kind of built patchwork already.
Again, I don't know how great itis because I haven't seen it
since they soft launched it and I never saw it again.
But like if you've already kind of built out the infrastructure
for that tool, you couldn't you maybe you apply that in that
sense to, you know, to get that kind of workflow where you can

(01:13:45):
visualize stills to animation really quickly without having to
change tabs. Because I think like when you
start to change tabs, right, just ADHD brain like you're just
you're kind of like you lose focus of what's in front of you.
Yeah. And to have all the tools and
the kind of visualization of what you're trying to accomplish
in one spot without having to get distracted is a huge unlock.

(01:14:07):
Maybe they apply it in that way,maybe they're thinking about
that, maybe they're not. I don't know.
But that's where my mind went when you showed me that.
Well, it's nice to like string thoughts together, like, you
know, instead of like looking at2 separate places like you just
said, like even just scrolling down to find the image that
you're looking for just then like putting them side by side,
like no, OK, let's let's redo this, let's play with this,
let's generate a video on this one.

(01:14:28):
Let's resize this, you know, 'cause then you can take it a
million different ways and I like I like that because there's
a there's another tool that I'vebeen playing with that's way
more comfy UI ish, but it's cloud based called weevie.
It's great. It's got like everything like
that's just got Flora's got likethe basics that are all you
need. It's got like ChatGPT, clawed

(01:14:49):
Gemini. And then on the image side, it's
like flux luma ray photon and you know, some of the other flux
models and Gemini and the video,it's got cling runway mini Max
VO2 XYZ. So it's not a million, but
Weevie has like everything because it literally looks like

(01:15:09):
a comfy UI workspace. But it's not as complicated and
it's very easy to sort of play with the workflow builder,
which, you know, this stuff might not be for your daily
creative needs, but if you're doing this stuff at scale, you
know, sometimes this is buildingthese work flows and systems
out. They could be like agents in a

(01:15:30):
way. And it's just an easy way to
bring things in. And I want to take this image,
combine it with this, turn it into like product shots and then
resize it, you know, 5 differentways, right?
Like and then you could build that workflow and then just drop
the stuff in, hit run and it allcomes out.
So I I think the hard part stilllies in creating image by image,
the variation that takes place within creating those images.

(01:15:51):
Meaning if you've got a theme baked out in your head, it's
kind of a lot of work to get an image that feels perfectly
cohesive of let's say it's a different character, but you
want that same lighting, that same style applied feels like
it's again, a continued part of an ongoing series, not detached.

(01:16:14):
It really needs to feel tight and cohesive and you know, like
consolidated into one place to tell that story.
Like it, I get so trapped sometimes and just like creating
cool one off things that I get away from the themes.
And I think that's part of the reason why is because like it's
so hard to create image by imageone at a time to connect those

(01:16:38):
dots. Meaning like, like if I'm
thinking about this nun series and I want a nun doing all this
crazy shit, it might take me a certain amount of time to get
one image, then to turn around and get them doing something
totally different based on the mechanics or the physics of the
image or whatever it is. That could take me a while.
How do you consolidate that process?
Because then that starts to get really interesting in terms of

(01:16:59):
like stringing stories together in a very like streamlined.
I hate straight hit me if I say streamline from now on.
But like, you know what I mean? Like that part of it is, is
probably still one of the last things that feels like this
super strenuous, time consuming process and maybe character
consistency and you know, thingslike that Sora piece of it, like

(01:17:22):
they they help, but like it's not there yet.
It's just crazy to me that like this stuff, like even this one
that I'm going to show you here like this is this is possible
now, like taking these shoes, right?
This is different. This is the weevie now and then
like changing it just subtly, but then also getting a front
view and then like maneuvering the shoes and it's picking up

(01:17:42):
all the fine grained details, especially the logo here.
And then like the the ridges, it's taking everything and doing
it really well. You know, having this all in one
space to generate at once and then be like, OK, I need this to
make a make a square, right? I can make this 9 by 16.
But like this is the this is what I'm talking about.
You can use everything in this one.
It has the out painting, it has,you know, all these different

(01:18:06):
models that you can play with. Then it's got these, you know,
generative. It has these like different sort
of sketch to image image. Image.
Control Nets you have video stuff you have all of this even
on the workspace you have like the you have like run any LLM
you can do a prompt concentratorimage describer like there's a
lot of things. This is what I feel like sort of

(01:18:29):
exactly what comfy UI is is working with, but you don't need
to be crazy to do it. Like, you know, like comfy UI,
not, not crazy. It's just like it's a lot.
You have to have the right hardware too.
You can't just pick it up. You can use the, you know, the
certain tools on cloud based. It's not as fast, but something

(01:18:50):
like this is a nice way to sort of bring everything together,
which I sort of like this workspace idea.
Now patchwork. The idea is there, right?
Like you see the idea and it makes sense.
It's just like, how do they? The execution needs to be
tighter. Yeah, 'cause it was so confusing
when they hold that out. I agree, just I agree.

(01:19:13):
But they did, they did mention V8, right?
They, they mentioned V8, they mentioned that they're, you
know, starting testing training for 7172.
So there's going to be iterations before, you know, V8.
I remember, I don't know if thisis like 2 weeks ago now, but
David really being optimistic about the fact that image

(01:19:33):
quality, consistency, things like that are going to be like
there's plenty of room for thoseto even get better than they're
at right now, which which was great to hear because we're not
even talking about V8, right? We're talking about 771, maybe
7-2, you know, like so those things are still on the horizon.
But yeah. Hopefully, maybe we'll see that

(01:19:57):
this year, I don't know. We'll see.
We'll see what happens. I mean, let's, let's see what
happens here in the next month or two.
I'm curious to see if they get more aggressive with talking
about video. We'll see how like that stays on
track too. Curious, but.
Sounds like there was a change. Remember when we said that a
couple weeks ago where it was like, it just sounds like
there's a little bit more, All right, we need to get stuff

(01:20:20):
done. Yeah, yeah.
It was sort of like, I don't know if it was, I don't want to
say it was LAX there for a while, but it felt very laissez
faire on certain things. And then it was just like, bang,
we're here. Next update, next update, next
update. Expect this.
Expect this. Yeah, that's.
What I'm hoping for, man, I really, you know.
Same with the community. Same with the community aspect.

(01:20:41):
We talked, we we ragged on that for a while.
But I do think with all these other things in the fold, that's
the other way to play this, right?
It's it's not only pushing things faster and shipping it
and you know, like putting their, their foot on the gas,
but like they, they've got so many users.
How do you keep them? How do you?
Keep improvement. That's all it is.

(01:21:02):
It's just like perceived improvement.
Like even if it's a micro feature, like we added weird
back great, we added remix back great.
Gives a ton of people something to talk about, ton of people's,
you know, stuff to try and then you go on to the next thing.
Yeah, instead of just giant model release.
I also, I also think like even with something like that editor,

(01:21:22):
when there, when there's so many, like with that layer
aspect specifically, they do need to continue to educate
people on, you know, at least the foundation of how to use
each little thing. At this point.
They have to, right, Because otherwise what you, you've spent
how much time building and then now no one's using it.
You can't really afford to do that either, right?

(01:21:44):
So they do have to sort of like do a better job of like drinking
their own champagne and kind of like really force feeding and
showing us exactly how they're how they're doing this stuff and
how to use it. Yeah.
I mean, it could be. I mean, it could be laying the
foundation for three DI mean. I think Korea just dropped
another. And they did.

(01:22:04):
They did. I'd be.
Damn good, yeah. Yeah.
So, you know, who knows when 3DSgoing to get here, if they're
even going to focus on it. There's there's like not I'm
like just excited again for justimages.
I feel like there was a there's been a good shift back like it
was just all video. It's only anyone anyone cared
about and I'm taking like the social litmus test.

(01:22:25):
From what I see the. Most of.
Right. That's all I'm making my
judgement on. And now it's like I see a lot
more images again, and I'm happyabout that because.
Well, and remember, no one therewas that huge gap in time where
unless you were using mid journey, you would see like the
average person put out somethingon social.
And what I like my litmus test is like other just B to B

(01:22:48):
marketers using Dolly for something.
And I'm just like, bro, like just don't even put it out.
Just don't even just don't even do it.
You know what I mean? And now right with, with that
Chachi BT component making that huge leap forward, you're seeing
that part of it start to take place and it's just a better

(01:23:08):
quality. It's it fits into the feed
better. There are plenty more use cases
now, right? Like so it's been interesting to
see that that take place, but. We're we got more common?
More common, more common. We'll we'll wrap this up.
Rory, you're doing a little vacation, so we'll have to
figure out the next the next recording, but we'll figure it
out, you know, and a lot of goodstuff coming, guys, a lot of

(01:23:31):
good stuff to be excited about. I mean, we're in a, we're in a
really good time, I think in 2025 with just the.
Fact that we didn't even talk about cling 2 point O today is
we didn't talk about telling youhow much is going.
On we didn't talk about the the Korea 3D thing.
We didn't talk about a lot of stuff Higgs field.
I don't even think we've talked about, which is, you know, just
the whole another video tool, right.

(01:23:52):
So like, what are we? These are the times, guys.
These are the times. It's Dan.
It's not slowing down South, geton board.
Hold on. Because hold on.
If we've hit, we've hit like sort of event horizon where it's
it's no longer drew and I can't drew and I can't keep up.
I can't keep up. I have to choose, you know, so

(01:24:12):
there's only so much you can deep dive into and like figure
it out. It's sort of just.
Like very true. So we're just like, there was
like a tsunami and we're like just riding on top of it, just
like keep going. You know, there's no, there's no
end to it. So let's you know.
With that being said too though,if you're still here, just hit
the Like button. Like and subscribe, baby.
Like and subscribe, maybe comment, share with a friend,

(01:24:35):
show your mom something just to give us a little help here cuz.
It sounds. You guys are crushing as of
lately and we like it and that'syeah, it's.
Helping us, loving all the shoutouts, loving, loving seeing your
guys work, love seeing the different creative, you know,
outcomes of having the same features.
Some some shit I would have never thought of.
So it's it's really it's a it's always fun to see.

(01:24:55):
But yeah, man, let's let's put abow on this thing.
Thanks for joining us, guys. We'll keep you posted on kind of
these changes as it come throughwith mid journey.
We still got plenty of things coming.
So we'll, we'll keep showing up if you guys do and we'll see you
guys in the next one. See ya.

All Episodes

Episode Transcript

Popular Podcasts

On Purpose with Jay Shetty

Ruthie's Table 4

The Joe Rogan Experience

.css-15opob5{left:0;position:absolute;top:0.8rem;} All Episodes

.css-14f5ked{margin:0;word-break:break-word;display:-webkit-box;-webkit-box-orient:vertical;box-orient:vertical;-webkit-line-clamp:2;overflow:hidden;}Midjourney’s Secret Playground: Chaos, Weird, and New Editor

Episode Transcript

Popular Podcasts

.css-r6mb8g{margin:0;word-break:break-word;display:-webkit-box;-webkit-box-orient:vertical;box-orient:vertical;-webkit-line-clamp:1;overflow:hidden;}On Purpose with Jay Shetty

Ruthie's Table 4

The Joe Rogan Experience

All Episodes

Midjourney’s Secret Playground: Chaos, Weird, and New Editor

On Purpose with Jay Shetty