AI Learning Lab

Apr 13, 2024 - How to Generate Consistent Characters in MidJourney

38CmDm5bCbY
Video2024-05-261:41:091 views

Description

In the latest episode of the AI Learning Lab, Kyle Shannon dives deep into the fascinating world of artificial intelligence, exploring its creative applications in music, comedy, and visual storytelling. He discusses the potential of tools like Udio and ChatGPT for generating engaging content, from crafting stand-up routines to creating immersive audio experiences. The session highlights the importance of character consistency in AI-generated imagery, showcasing how platforms like MidJourney can bring characters to life across various contexts. With a blend of humor and insight, Kyle encourages viewers to embrace these technologies as amplifiers of creativity, pushing the boundaries of what is possible in digital art and storytelling. For more engaging discussions and demonstrations, check out the AI Learning Lab on TikTok: [AI Learning Lab](https://tiktok.com/@aiLearningLab). #ArtificialIntelligence #CreativeAI #MusicGeneration #Comedy #VisualStorytelling #MidJourney #ChatGPT #AIArt #digitalcreativity Chapters: 00:00:00 Intro-Music Background 00:11:31 Creative Uses for GPTs 00:12:52 Udio 00:16:57 Prairie Dog Politics 00:21:25 ADHD Coffee Routine 00:25:03 Midjourney on Discord 00:31:45 Style Reference Images 00:36:42 Character Reference 00:41:41 Midjourney Settings 00:46:00 Ideogram AI 00:50:51 OpenAI Technology 00:56:01 Vintage Book Illustrations 01:01:01 Inspiring Image Results 01:08:09 Character Reference Example 01:14:17 Rabbit R1 AI Device 01:21:57 LinkedIn Post 01:28:51 Apple Should Have Invented 01:34:25 Pythagorea App Development 01:37:06 Fake AI Demos 01:40:00 Final Thoughts 01:40:31 Outro

Chapters

Transcript

0:02 [Music]
0:19 every time I see you
0:21 now that in
0:24 my every time I see your
0:27 mouth I hear that smile
0:32 [Music]
0:38 morning
0:40 [Music]
0:45 out you believeing
0:48 me again
0:51 today you convince
0:54 me again
0:57 today you're leaving this hotel looking
1:00 for someone else's Golden
1:06 Ring should say so
1:14 [Music]
1:15 long
1:20 cry so
1:26 long don't you cry for me
1:30 [Music]
1:36 Sharon drink the cigarettes and keeping
1:38 warm out on the
1:41 road chasing down the lifestyle out on
1:44 Highway
1:48 24 New York state was a rolling Breeze
1:51 in a sunshine with a blue sky falling to
1:55 the chill of old September creeping in
2:03 you will be
2:05 made again
2:08 today you will convince
2:11 me again
2:14 today you're leaving this hotel looking
2:18 for someone else's Golden
2:23 Ring should say so long suan
2:30 [Music]
2:38 so long
2:43 suzan don't you cry for
2:52 me hello Robert Rossy thank you so much
2:54 for the lightning bolts very generous of
2:57 you happy Saturday night Saturday Night
2:59 Live
3:05 people my name is Kyle Shannon it's the
3:07 AI learning lab and dog singing
3:11 extravaganza
3:17 [Music]
3:36 I think he's done thank you Robert Rossy
3:40 very nice thank you so much
3:49 [Music]
4:16 [Music]
4:18 [Applause]
4:20 [Music]
4:23 he's not
4:25 done hey Source Camp what's happening
4:27 better Angels good to see you Zack
4:30 [Music]
4:35 laughing thunder hit level
4:37 16 thank you sir or
4:42 ma'am I think sir
4:46 [Music]
5:05 all right we are warmed
5:09 up let's do
5:11 this
5:13 um played around with some more yudo
5:15 today I yud for like two hours this
5:18 morning was offline so that means they
5:21 were ramping up their servers and I got
5:25 better results today so I have a feeling
5:27 they have they've turned their quality
5:28 back up
5:31 if indeed it was turned down before but
5:33 I think it it it's the only thing that
5:35 could have explained how shitty it got
5:37 for a day um I also heard a really cool
5:42 thing that
5:45 um I think it was Greg Brockman said
5:48 something about it being really nice
5:49 when you
5:50 can help a customer do what they do and
5:55 then the CEO of Udo thanked open AI for
6:00 their support in
6:02 scaling so that's now two companies that
6:06 have remarkable
6:08 technology that I did not know open AI
6:11 was behind one of them is hey Jen so all
6:15 those really amazing translations where
6:17 people's lips get synced and the it
6:19 sounds like someone really talking in
6:22 that different language I think that's
6:24 open ai's technology and then yudo the
6:28 puno Killer
6:30 comes out and it turns out they've got
6:33 open AI behind
6:35 them so I have a sneaky
6:39 suspicion that when the real next
6:43 version of of chat GPT comes out it's
6:45 going to
6:47 have all this kind of stuff in
6:50 it you know whether it's voice synthesis
6:54 or music generation or translations
6:58 video or video to video or sunno or like
7:02 I think it's just going to all be in
7:03 there um so I find that fascinating I
7:07 find it just
7:09 fascinating Saturday Night
7:12 Live welcome welcome welcome where are
7:15 the viewers dropping that always happens
7:18 so I think Tik Tock you know does a
7:20 blast out to people and most people look
7:22 at it and go like yeah I'm good an old
7:26 guy
7:28 no I'm looking for
7:30 twerking I'm looking for twerking
7:33 girlies they see my fat old ass they're
7:35 like
7:38 nah harod D what's happening saw a car
7:42 tag today that said AI rules I wanted to
7:45 run up to the window and talk to them
7:47 that's really funny yeah you know we're
7:49 going to be coming into a time where if
7:51 you're sporting an AI rules bumper
7:53 sticker someone's going to come up and
7:55 kick your ass cuz uh as these things get
7:59 better people people are getting a bit
8:00 angrier I'm noticing I love tuning in
8:03 Kyle hey major laser what's happening
8:05 always makes me ponder about the AI
8:07 powered nitrogen capsules yeah they're
8:10 coming man I uh I'll I'll share a fun
8:15 thing with you um so we we talked last
8:17 night a little bit about uh udio can do
8:20 spoken word um and then I said it could
8:24 it could do comedy so today I I use chat
8:28 GPT to help me write a comedy routine
8:31 well not a comedy routine a
8:33 joke um for my friend's company and uh
8:37 it did it and I sent it to him he was he
8:39 he used to be a stand-up
8:42 comic so it's crazy all right here we
8:46 go so here I am at
8:49 yudo the joke was written well it's
8:52 three jokes the jokes were written in uh
8:55 chat GPT and then two of the punch lines
8:58 sucked but all of the setups were good
9:00 one of the punch lines was fine and two
9:04 of the punch lines sucked so I fixed the
9:05 punch lines and then I probably did
9:09 this probably made 12 12 or 14 versions
9:14 until I got this one which was decent
9:16 enough yeah the first thing I felt was a
9:19 rock have you heard about these
9:21 minimalist running shoes called zero
9:23 shoes their tagline is feel the
9:26 world yeah the first thing I felt was a
9:28 rock
9:31 the second thing
9:33 regret I mean if I wanted to feel every
9:35 Pebble and piece of glass on the road
9:37 I'd just tape a couple of orange peels
9:39 to the bottom of my feet and call it a
9:42 day with zero shoes you're not you're
9:44 not just feeling the world you're
9:46 feeling every poor life choice you've
9:49 ever
9:49 [Applause]
9:51 made yes so have you heard about these
9:55 you know it's not quite there but it's
9:58 pretty flipping closed like it it
10:01 understands punch lines and inflection
10:03 and audience
10:05 response and it for the most part put
10:08 the stuff in there I tried to put
10:10 in laughs you know I put them in
10:13 brackets I put laugh or pause and it
10:16 kind of ignored that [ __ ] but anyway
10:18 it's it's
10:20 interesting fascinating fascinating
10:26 [Music]
10:27 fting trying to think idea of something
10:30 we could do tonight I can't remember
10:32 what it was
10:33 though I don't know anyway crazy right
10:36 using chat GPT for role playing to build
10:39 characters and
10:40 stories can we play with that um sure I
10:44 don't know a lot about that world but
10:46 that shouldn't stop us because we have
10:47 chat
10:49 GPT
10:53 um I mean I think there's enough stuff
10:55 in the Corpus that you can probably just
10:57 get it to
11:02 probably just get it to do its thing
11:04 just act like a dungeon master
11:08 right build characters and
11:12 stories
11:15 um I don't know what good would look
11:17 like unfortunately I like I know so
11:19 little about that world that I don't I'm
11:22 happy to do something I just don't
11:27 know I don't know what to what's even it
11:30 to do can we talk about some creative
11:32 uses for
11:34 gpts that's an interesting idea Joker
11:37 has an RPG
11:39 GPT all right what's the name of it
11:47 Joker launch your digital business AI
11:50 have consistent looking characters yet
11:52 yes they
11:53 do um Mid Journey now has consistent
11:56 characters they have this thing called
11:58 character
12:00 uhu let's see RPG
12:02 Joker I don't know if Joker's no results
12:13 [Music]
12:21 fan name that sitcom
12:28 [Music]
12:34 realm Weaver man chat jpt came up with
12:36 that didn't it realm
12:44 Weaver Joker's
12:47 gpts start chat a dungeon master start a
12:50 campaign list the commands let's list
12:52 the
12:55 commands what was that site with the
12:57 comedy sorry for the usual I'm late f
12:59 you that's okay frumple um it was udio
13:03 ud.com UD io.com it's a music generation
13:09 uh tool I'll I'll show you here you've
13:10 got to you've got to actually do
13:13 something um to get it to do voice so
13:17 here we'll we'll flip back over to that
13:19 here's here's how you do
13:21 it so you say create please enter a
13:25 prompt
13:26 right okay so
13:29 first thing we're going to do is I'm
13:31 going to put in square brackets spoken
13:38 word oh this will be interesting
13:42 um
13:44 political
13:48 stump
13:53 speech about
14:00 the scourge of prairie
14:16 dogs all
14:17 right that'll be interesting to see now
14:20 couple of things you have to do there's
14:22 this thing called manual mode and if you
14:24 look at the information for that if you
14:26 roll over it it basically says
14:29 if that's off then they're going to
14:32 rewrite your prompt when they rewrite
14:34 your prompt they tend to add musical
14:38 um things into it so it makes it musical
14:41 so you have to turn that on manual and
14:42 then just your prompt is there and then
14:45 you also have to
14:47 do
14:49 um you have to write your own lyrics so
14:51 we'll go over to chat
14:53 GPT let's
14:58 do right write a 30- second political
15:01 dump speech about the scourge of prairie
15:10 dogs did anyone get that by the
15:13 way that sitcom thre's company someone
15:16 got it who got
15:18 it Price is Right not even close Joy pie
15:21 got it first oh no sorry ver ver
15:25 verifier wait
15:27 no uh wonders wond wait no do G need a
15:32 bottle wait Silver Fox wait
15:35 frumple hang on hang on bunch of you got
15:40 it I didn't even sing it good all right
15:43 so who is the first one
15:46 uh your name is spelled in
15:50 runes Misty
15:53 Grimes nice Misty Grimes wins the uh
15:57 wins it so many of us got it
16:00 watched it every week I
16:02 know I didn't even hear what's going on
16:04 just saw saw him trying to find
16:08 it uh I heard it can do comedy too yeah
16:11 it can do comedy too okay so ladies and
16:14 gentlemen so we're going to grab our
16:16 little
16:17 speech we're going go back over to
16:20 here we're going to paste it in so
16:23 that's it put spoken word and there
16:26 might be other things you could put
16:27 there I just I know that works so when I
16:29 find something that works I tend to
16:30 stick with it let's see we'll just do
16:33 political stump
16:36 speech I'll say
16:38 two
16:44 roaring
16:47 crowd and by the way the the prairie dog
16:50 thing as funny as it sounds when when I
16:52 first moved to Boulder apparently
16:55 prairie dog preservation in Boulder
16:57 pisses off the rest of Colorado so where
17:00 prairie dogs are are nuisances so it's
17:04 pretty funny okay so we'll create that
17:07 we'll see if it can do a political
17:08 speech I'm I'm curious if it it doesn't
17:10 do it because it has the word politics
17:13 in it could be
17:15 interesting speech I'm I'm curious if it
17:18 it doesn't do it because it has the word
17:20 politics and it could be
17:26 interesting fresh prince of B air
17:32 close I was singing the song as you did
17:35 the tune that's crazy maybe it's a we
17:38 have a psychic psychic
17:41 connection all right Prairie's
17:47 Edge ladies and gentlemen today we s
17:49 Fross roads in our community we Face a
17:51 critical challenge that threatens our
17:52 beautiful landscapes yes I'm talking
17:53 about the spirit of
17:56 yes ladies and gentlemen today we stand
17:58 at Crossroads in our community we Face a
17:59 crit challenge the threat all right so
18:01 it's going too fast I have a feeling so
18:04 um udio can only do 30 seconds
18:08 so I think there might be too much here
18:10 it's time to address these issues head
18:17 on let's get rid of that
18:22 sentence let's get rid of that
18:26 sentence and then I'm going to say
18:31 slow
18:32 D
18:34 wait
18:37 DB
18:40 liberat political spunch stump speech to
18:44 a roaring crowd boom all
18:48 right I'm up to 96 of my 600 monthly um
18:55 Generations I can do
19:00 I got flagged for
19:02 using what word that began with
19:07 P just ask J Jack
19:12 [Music]
19:17 Ritter oh really
19:21 unbelievable Twitter Tik Tok is so weird
19:26 man so weird
19:29 [Music]
19:35 ladies and gentlemen today we stand at a
19:37 Crossroads in our community these
19:38 creatures while part of our ecosystem
19:41 have multiplied beyond control damaging
19:43 our fields and making it difficult for
19:44 local farmers to protect their Li hoods
19:47 it's time we address this issue headon
19:49 with sustainable
19:52 human ladies and gentlemen today we
19:55 stand a Crossroads in our community
19:57 these creatures while part of our system
20:00 have multiplied beyond control damaging
20:02 our fields and making it difficult for
20:04 local farmers to protect on their
20:06 livehoods it's time we address this
20:08 issue headon with sustainable Humane
20:11 solutions that respect our environment
20:12 yet protect our cultural heritage I
20:14 promise to implement all right that's
20:17 awful all right so what we're going to
20:18 do we're going to do spoken word um
20:23 edgy
20:25 standup comic
20:29 in large
20:31 theater
20:34 large
20:37 theater all right so we got to go to
20:39 chat jpt and say
20:45 uh write a 30 second standup comedy
20:50 routine about someone who drinks iced
20:53 coffee at night period actually make it
20:57 the standup comment
21:01 themselves that is making fun of
21:03 themselves for drinking iced coffee at
21:06 night
21:07 period you could also throw in the fact
21:10 that they have
21:12 ADHD and they use that as an excuse to
21:15 normalize the barbaric
21:25 practice yeah livelihoods yeah that's
21:27 one of the tricks with these things
21:30 is wait 30 I ever see someone sipping
21:34 coffee at midnight and think who hurt
21:36 you spoiler it's me I blame ADHD
21:40 um make it
21:43 three back to back
21:57 jokes okay so here our
22:00 comedy we don't even read it this is not
22:03 this is how you do this now people you
22:06 want best practices here here's your
22:07 best practices just copy and paste [ __ ]
22:10 directly out of chat GPT don't proof fre
22:15 it especially if you're doing something
22:17 that's factual like a legal brief just
22:20 copy and paste it straight out of chat
22:21 GPT and hand it to a judge you'll be
22:25 fine trust me it's they're going to they
22:27 don't make mistakes anymore
22:29 [Laughter]
22:33 can you show how to create consistent
22:35 characters in AI I absolutely can we'll
22:37 do that
22:40 next and in fact I will show you in both
22:46 um I will show you in both Discord and
22:50 the web version of mid Journey since a
22:52 lot of people don't have access to the
22:54 web
22:57 version ever see someone sipping iced
22:59 coffee at midnight and think who hurt
23:05 you that's me I'm self-destructive but
23:08 hey
23:11 caffeinated I tell people it's my ADHD
23:14 like my mind my coffee is always on the
23:18 Rocks honestly I'm just prepping for
23:20 tomorrow's tired at this rate I'll sleep
23:23 when I'm
23:26 dead not horrible
23:29 ever see someone sipping iced coffee at
23:31 midnight and think who hurt
23:34 you that was sort of Comedy timing
23:37 that's me I'm self-destructive but hey
23:41 caffeinated that's the appropriate laugh
23:44 I tell people it's my ADHD like my mind
23:46 my coffe is always on the
23:50 Rocks honestly I'm just prepping for
23:52 tomorrow's
23:54 tired at this rate I'll sleep when I'm
23:56 dead
24:01 that was funny
24:03 that's that that's actually this is
24:05 actually really interesting cuz this is
24:06 a comedian bombing and these jokes
24:09 deserve to bomb ever see someone sipping
24:12 iced coffee at midnight and think who
24:15 hurt
24:16 you it's like polite polite polite
24:20 laughter like three
24:24 people that's me I'm self-destructive
24:27 but hey caffeinated
24:30 [Laughter]
24:31 that's
24:32 uncomfortable that's the virtual
24:35 comedian's bombing and I feel for for
24:37 him inflection I know it's it's pretty
24:43 bad um all right
24:48 so let let's see let me jump over to
24:52 Discord
25:03 [Music]
25:06 okay so here we are on Discord can I
25:09 make this bigger for you I think
25:15 so oh let me close this
25:20 boom plus yeah there we go okay so this
25:25 is mid Journey so this is just a channel
25:27 in the AI on
25:29 Discord and
25:32 um if you've never seen mid journey in
25:35 Discord this is it if you don't know
25:37 Discord by the way this is Discord down
25:39 the Le hand side you have these
25:40 different what they call servers
25:42 different communities and I'm in the AI
25:45 Salon community and then within that
25:47 there's channels so it's just like slack
25:49 if you've ever used slack for work it's
25:51 slack is basically a ripoff of Discord
25:53 but done for
25:55 companies um and so within our Channel
25:59 we've got a mid Journey play Channel
26:00 where you can go play with mid Journey
26:02 so and which is nice cuz it's it's a
26:05 it's a like a low um a low usage channel
26:09 so there's not a bunch of people you
26:11 know dumping things in here um so the
26:13 way you make an image in mid Journey you
26:15 go slash imag you type in slash imagine
26:18 and then it'll ask you for a
26:20 prompt and then you type in your prompt
26:23 and it'll make an image and then there's
26:24 all sorts of ways you can modify that um
26:27 but before we do
26:31 that um I'm going to upload an image and
26:34 the image I'm going to upload well wait
26:38 there's a we've got this woman
26:40 here let's see if we've got a face
26:43 there's a
26:52 face e that's a
26:55 little a little cliche
27:02 the medieval
27:04 Wai all right oh wait there was a face
27:07 oh that's interesting all
27:10 right so let's let's grab this upper
27:12 leftand face I'm going to I'm going to
27:14 hit U1 which means upscale one so I'm
27:16 going to upscale that that little
27:19 face all right so that's going to be our
27:21 our character
27:24 reference all right so there it
27:27 is and and first thing I'm going to do
27:30 here is I'm going to rightclick and I'm
27:31 going to copy the link to this image not
27:34 the image the link to it the so this is
27:37 being stored on Discord servers
27:40 somewhere so I just copied the link to
27:41 it now I'm going to type in slash
27:47 imagine and then I'm going to go um
27:51 let's
27:53 see um child let's see
28:02 closeup
28:04 of
28:07 child's face
28:10 as
28:11 they sled down a
28:17 hill in
28:20 a
28:26 snowstorm okay then I'm going to
28:30 do
28:31 space-- AR which stands for aspect ratio
28:35 oops I didn't do D Dash I did equals
28:37 equals so go Das Dash AR then I'm going
28:41 to do 16 colon 9 which is the aspect
28:45 ratio and then I'm go space dash dash CF
28:50 character reference CF space and then
28:54 I'm going to paste in that URL so that
28:57 big that big ass URL is is the is where
29:01 that image is so I'm basically saying
29:03 use this as your
29:06 reference uh for for the face
29:10 here
29:12 um and then I hit return and it should
29:15 if I did everything
29:16 right yeah now it's now it's off
29:19 generating things waiting to start so
29:21 it's going to go off and it's going to
29:22 do
29:25 that you can get them in paid Deco here
29:28 yeah that's a good
29:31 idea checking can't get consistent
29:34 characters in stable diffusion well you
29:36 can in stable diffusion laughing Thunder
29:38 if you use a thing called dream Booth or
29:41 if you use Leonardo you can train up
29:44 your own model in
29:46 Leonardo so there's her crazy orange
29:49 eyes right so there's our consistent
29:51 character and now we could say let's do
29:55 slash
30:03 slash
30:06 imagine I'm going to do the same prompt
30:08 but I'm going to do
30:10 um let's see
30:16 um um
30:22 children's
30:24 book
30:27 illustration of a closeup of a child's
30:32 face and then
30:34 CF that same I got to go get
30:37 the hang on got to go get the reference
30:41 image copy
30:46 image oh wait I did that wrong copy
30:48 image
30:50 link copy link and then go back over
30:53 here and paste
30:55 it and go bang
31:00 so here's those are pretty
31:03 good I mean they look like her
31:07 right creepy but
31:10 cool so same face different situation
31:14 and now this one these should look like
31:15 illustrations right these are
31:17 photographic these should look like
31:19 illustrations so that's how you do
31:24 it now you can also do a thing called a
31:26 style reference which is pretty
31:32 cool and the style reference you can
31:35 only do in
31:45 here did it do those as illustrations
31:49 not
31:50 really oh I'm going to try something
31:52 here all right let's
31:55 see um
31:59 I'm going to copy all this children's
32:01 book
32:02 illustration we're going to add
32:04 something so we're going to
32:07 add dash
32:10 dash uh space CW space z now so CF is
32:18 the character reference and then
32:21 CW is it's not it's D- CW no space um d
32:28 - CW is character weight and if it's a
32:31 100 it will it will keep her clothing
32:35 right and if it's zero it'll just keep
32:37 her face so so if I say uh closeup of a
32:40 child's
32:41 face
32:43 um in
32:46 a uh
32:50 face closeup
32:53 of a girl in a red hat and
33:01 green
33:04 jacket as they sled down a hill in a
33:08 snowstorm and we'll do let's see
33:11 children's of of a medium shot medium
33:15 shot of a girl in a and so so I said
33:18 close up before so now it should be not
33:20 as close and she shouldn't be wearing
33:22 that same outfit she should be in a red
33:25 hat and a green jacket all right let's
33:27 see how we
33:29 do
33:31 oh I always do this copy you have to
33:35 type slash imagine I'm sure you're all
33:37 telling me you didn't type SL imagine
33:40 you dumb dumb yeah yeah all right there
33:42 we
33:43 go can consistent characters be used in
33:47 video too like Runway ml so what you do
33:49 launch your digital business is generate
33:52 your still images in mid journey and
33:54 then take them into something like
33:56 Runway to to animate them
33:59 typo on hat where oh red head it still
34:03 got it look there's our red
34:06 [Laughter]
34:13 head oh this is looking pretty cool see
34:15 medium shot let's see if it's got her
34:17 creepy little eyes yeah pretty good this
34:22 this one on the upper right's not very
34:23 good but the other three are pretty good
34:30 that one especially number one if we
34:32 upscale
34:36 that
34:38 right so there she is from that image so
34:44 her
34:45 sledding it's pretty
34:48 close pretty close you know and again
34:51 like like with all of these things here
34:53 here's the good news the good news is
34:55 this exists um thead news is sometimes
34:59 it's really
35:02 inconsistent is mid Journey the best for
35:04 consistent characters so cool yeah Isn't
35:06 that cool um now let me show I'll show
35:08 you another thing called style reference
35:10 I'm going to show you that here because
35:12 um it doesn't work in the web site
35:14 although I'll test it tonight but it it
35:16 wasn't working before okay here's a
35:18 really good one so let's let's uh let's
35:21 upscale this this one with the green the
35:24 green uh clouds with the silhouette so
35:27 that's number two so I'm going to
35:29 upscale number
35:30 two so what we're going to do
35:34 now is we're going to take our same
35:36 little prompt from
35:48 her I'm going to type slash
35:54 imagine I'm going to paste that all in
35:56 there and then I'm going to do Dash Dash
36:00 sref which stands for style
36:03 reference and then I'm going to right
36:05 click this and copy the link to this so
36:08 now we're
36:11 using we're
36:13 using the girl with the orange eyes as
36:17 the character reference and we're using
36:18 this image as the style reference and so
36:21 this should give us a pretty
36:23 dramatically
36:24 different set of illustrations
36:30 [Music]
36:42 this is impressive yeah this is pretty
36:44 slick isn't
36:48 it Dolly cannot do something similar not
36:52 with uploading what you can do in DOI is
36:54 you can upload an image let's say you
36:57 upload an image of your yourself and
36:59 have it describe that image and then
37:02 write your prompt and say you know use
37:04 the face of the thing you described but
37:07 you can't really use it as
37:09 a as a thing but look at this like
37:14 totally you know consistent with the
37:16 style of that other
37:18 thing um that one sort of looks like
37:21 her I guess more than the others but so
37:25 let's upscale number four
37:30 and then we'll do a creative upscale
37:32 from
37:34 there and this should give us a pretty
37:36 slick little image do you not have Alpha
37:40 mid Journey yet so much easier yeah I do
37:42 AI nurse I'm going to I'm going to go
37:44 show how to do this in the alpha of the
37:46 web the web version but since since you
37:49 have to have a
37:50 thousand um generations to be able to
37:54 use it you know not a lot of people do
37:56 and some people want to know how to use
37:59 uh mid journey in Discord so I figured
38:01 I'd just do
38:18 both Tams this is so good first time I'm
38:20 seeing it in real time on Mid Journey
38:22 yeah it's slick isn't it what are your
38:24 settings uh I'm just doing let me see FL
38:28 settings I don't think I have anything
38:35 fancy style medium styliz
38:38 medium oh I've got it in remix mode I
38:41 don't know why I have it in remix
38:49 mode High variation mode and
38:54 fast so that's it but I mean look how
38:58 good that
39:02 is oh you know what we could
39:04 do do tell Kyle all right so here's what
39:08 we're going to do we're going to
39:09 download this sorry we're doing a chat
39:11 add
39:13 moment sorry save
39:17 image let's just call her oh okay that's
39:20 good book illustration save okay so now
39:23 let's
39:25 go back over here
39:28 spoken word we're going to
39:30 go
39:33 um
39:37 spooky voice
39:39 [Music]
39:41 over
39:46 remembering the
39:49 past all
39:53 right then we're going to go here we're
39:55 going to go
40:00 write a 30 second
40:09 script
40:14 for a young woman
40:20 reminiscing that faithful day she went
40:25 sledding not faithful faith
40:32 hateful she went sleding
40:40 um make it creepy and
40:51 spooky whispery distant that's good it
40:53 was a cold gray afternoon oh that's
40:56 perfect okay
40:58 okay copy I don't know what it said
41:01 whispery distant I like
41:04 that
41:12 spooky
41:15 whispery
41:17 distant young
41:22 woman voice
41:24 over remembering the past
41:30 create
41:34 okay this should be interesting would
41:37 you be willing to post some of the
41:38 settings you used in mid journey in the
41:41 AI Launchpad tools
41:46 area
41:50 yes do you me do you mean the srf and CF
41:53 stuff or or like a screenshot of my
41:56 settings
41:59 cuz I don't think my settings are are
42:01 anything remarkable although I like I
42:03 don't think I should have remixing
42:05 turned on so that's that one's a little
42:07 weird that I have that turned on like
42:09 that all right let's see if this is any
42:14 [Music]
42:16 good it was a cold gray afternoon the
42:18 kind that's a young
42:22 woman it was a cold gray
42:25 afternoon okay let's see uh let's see um
42:31 young
42:33 woman
42:35 speaking in a spooky
42:38 [Music]
42:39 whispery
42:44 distant
42:47 tone remembering the past all right
42:50 let's see if we can
42:52 get oh I know another way we can do
42:55 this just for the new to MJ to learn
42:59 what settings to use or where anyone can
43:02 help Okay um yeah let
43:07 me let me
43:10 see I can take a screenshot of this
43:33 [Music]
43:35 where am I going to put it I think I'll
43:37 put it
43:40 in I think I'll put it in water cooler
43:42 in a in a threaded
43:46 post oops
43:59 um
44:00 Mid Journey
44:04 settings
44:06 for AI
44:09 learning
44:12 lab
44:16 example
44:19 and dh- C
44:22 ref
44:24 and dash dash s
44:29 ref
44:33 um that's good
44:36 enough all right there's that okay now
44:39 let's go see if we got a young woman
44:42 [Music]
44:46 seats it was a cold okay that's bad so I
44:51 think what we're going to do is we're
44:53 going to put
44:55 bracket Young
45:06 female bracket return let's try
45:15 this plus all the Tik
45:21 Tock oh yeah all these Tik toks are on
45:24 YouTube if if you want if you want back
45:26 recordings of these follow the YouTube
45:30 channel at AI learninglab dtt as in- Tik
45:35 Tock AI learning lab is someone else but
45:38 AI learninglab dtt is the archives of
45:42 these lives on
45:51 YouTube mid Journey okay but in my
45:54 opinion requires more prompting it does
45:56 but one of the things you can do is you
45:58 can go to like ideogram or chat
46:02 GPT have because both of them augment
46:05 your prompts and let them write the good
46:08 prompts and then take those over to Mid
46:10 Journey here I'll show I'll show you how
46:11 to do that too all right let's see if we
46:14 got
46:18 her it was a cold gray what the
46:25 hell it was a cold gray let's see uh
46:31 spoken let me try
46:35 um
46:38 female spoken word is is does does
46:42 spoken word imply that only men can do
46:45 spoken word performances D is there
46:49 something that I don't know about in the
46:51 English language did I did I miss a
46:55 memo yudo is hit and miss un female
46:58 spoken it's really hit and miss like I'm
47:02 not getting anything like it's it's all
47:04 like crusty old men I mean we're
47:07 swelling all but
47:11 geez how many times do I need to say
47:14 it's a lady listen hello yio listen it's
47:18 a
47:22 lady it was a cold gra okay I give up
47:27 it was a cold gray afternoon finally the
47:30 kind where the snow muffles everything
47:31 into silence we were just kids dragging
47:34 our sleds up the old hill by the woods
47:36 the one they told us to avoid I should
47:37 have listened as I pushed off the world
47:40 slipped away it was just me the rush and
47:42 something else Whispers from the trees
47:44 Shadows stretched like fingers across my
47:46 path when I reached the bottom the
47:47 laughter of my friends seemed miles away
47:50 I looked back once something in the
47:52 woods watched me its eyes like Hollow
47:54 promises and have letting go I've never
47:55 been sing since
48:01 H that was weird let
48:04 me is there too much here I think
48:07 there's too much here
48:08 again never been sleding since let's see
48:29 [Music]
48:31 is she going to take a breath well she's
48:33 super excited Pate she's sleding she's
48:36 remembering sledding so no she can't
48:38 take a
48:39 breath it talks fast if you use a lot of
48:41 text yeah I killed some text
48:44 there it's really funny someone please
48:47 what is the name of this AI this AI is
48:48 called udio UD IO it is remarkable and
48:53 and this is not really what it's known
48:55 for what it's known for is
48:57 music um we're we're trying to get it to
49:00 do talkie stuff
49:03 decent please be female it was a cold
49:09 gra it was a cold gra Jesus
49:13 jeez I okay I give up whatever wait is
49:17 there
49:29 lyrics tip adverse for additional
49:34 control female female
49:38 voice how many times let me do spoken
49:42 word oh wait instead of female let me
49:45 do woman speaking young woman
49:50 speaking I know I said I was not going
49:52 to do this and now here I am doing it
49:54 again young women speak
50:18 spoken udio could eventually be used to
50:21 do voiceovers for entire audio books
50:24 absolutely could well and that's again
50:26 what
50:27 what I'm a
50:29 Little shooketh by Danielle
50:33 is the two of the tools that I think are
50:36 kind of the most impressive tools um he
50:39 Jen's video
50:41 translation and yudo with this music
50:45 stuff both apparently have open AI
50:48 behind them so I'm thinking that this is
50:51 open AI technology powering both of
50:53 these is my guess so crazy cza it's cza
51:01 it'sa soza Danielle listen Danelle it's
51:05 so crazy isn't it I got yio to make 1940
51:09 style mint radio shows music commercials
51:12 news briefs and station IDs yeah that's
51:15 amazing recel that's
51:18 great see this is the thing
51:21 where like once you once you
51:24 get how to get a little bit of control
51:27 with these tools
51:29 um you can just take a concept like I
51:32 want to do a 40s radio show and have it
51:35 do all the sound effects and just
51:39 everything it was a cold gray afternoon
51:42 that's too dramatic but it's a woman it
51:45 was a cold gray afternoon the kind where
51:47 the snow muffles everything into silence
51:50 as I pushed off the world slipped away
51:53 it was just me the rush and something
51:55 else Whispers from the trees
51:57 Shadows stretched like fingers across my
51:59 path when I reached the bottom the
52:01 laughter of my friend seemed miles away
52:03 I looked back once something in the
52:06 woods watched me its eyes like Hollow
52:08 Promises of Never Letting Go I've never
52:11 been spling
52:12 [Music]
52:17 since got one too many sentences in it I
52:22 think so let's see when I reach the
52:24 bottom the left two miles away
52:52 woman speaking slowly
53:00 in a spooky whispery distance okay let's
53:02 try it one more
53:05 time line break in each sentence Oh
53:08 that's not a bad
53:09 idea this literally just came out and
53:12 it's that
53:14 good she's totally holding a cigarette
53:19 holder music producers now have free
53:23 tireless session musicians well here's
53:25 the thing Jeff Jeff
53:28 it's not um what this is not doing is is
53:32 it's not splitting any of this music
53:33 into tracks um it is if you know
53:37 anything about music it's really janky
53:40 it it's it's audio Fidelity is not great
53:44 but um like using it as a starting point
53:49 as like a as like an ideation tool or
53:52 for doing stuff like we're doing here if
53:54 you're doing some sort of cool audio
53:57 book voiceover kind of
53:59 [Music]
54:01 thing it was a cold gray eye oh come
54:04 [Music]
54:08 on it was a cold gray all right I'm sick
54:12 of men now all right we're done we're
54:14 done
54:16 here
54:18 um okay so that was
54:23 um that was mid journey in Discord let
54:25 me let me show you a quick little hack
54:27 um if you're if you are frustrated
54:30 writing
54:31 prompts jump over to chat GPT and just
54:35 say
54:42 um
54:44 create an illustration in the style of
54:47 Courier and
54:51 Ives of a young girl sledding down a
54:55 hill
54:57 with a
54:58 big
55:00 laugh on her
55:05 face period uh should be 16 by9
55:10 wide okay so this going to go make us an
55:13 image
55:20 now more control in 11 labs to pick the
55:24 voices yeah but what what 11 Labs is not
55:27 doing two block Tom is anything like the
55:29 background music or like when the
55:31 standup comic does his thing the laughs
55:34 all that sort of stuff this is doing
55:36 like a complete
55:37 production so again different tools are
55:40 good for different things if you want
55:42 some predictability then use something
55:44 like 11 11 Labs voices where you'll get
55:47 exactly what you
55:48 want um okay so here's this if I click
55:54 on the image
55:56 and then I click on this little I for
55:58 information there's the actual prompt it
56:01 wrote illustration in the style of a
56:03 vintage children's book featuring a
56:05 young girl sliding down a snow covered
56:06 Hill the scene captures the girl in a
56:08 big joyful laugh on her face right so I
56:11 can now copy this prompt and take it
56:14 over to Mid
56:18 Journey I'm going to take everything but
56:20 the aspect ratio stuff so if I flip over
56:22 to Mid Journey now type in slash imagine
56:27 type in that and then go D-
56:31 AR 16 colon 9 for aspect
56:35 ratio 16 colon 9 and then I'm going to
56:39 go Das Dash
56:43 CF and we're going to go
56:46 grab the
56:50 original image that
56:53 one so I'll do copy link and then we'll
56:57 put that here and then I'm going to go
57:00 D- srf for style
57:02 reference
57:05 space and then we'll go get that
57:10 creepy creepy green thing
57:15 that copy image or oh wait copy image
57:20 link and then put that there and so now
57:25 we should get something that's
57:27 somewhere between that Courier and IES
57:29 looking thing and what we had before but
57:32 with our little girl's
57:42 face Kyle you mentioned doing the same
57:44 trick with ideogram yeah exactly so with
57:46 ideogram same same basic idea so if you
57:50 go to do I have ideogram up here
57:52 anywhere I don't know if I go to
57:54 ideogram
58:00 and let's just say let's just go find
58:09 something we'll do that one so here's
58:13 the prompt for
58:16 that and then here's the magic
58:19 prompt right so we'll grab the ma magic
58:22 prompt so that's the magic prompt is the
58:24 thing that ideogram wrote so I can copy
58:27 that actually this this will be an
58:29 interesting one let's copy that which is
58:32 a really distinct style so here's here's
58:34 our freaky ass Courier and hives looking
58:38 looking thing I also noticed I didn't do
58:40 a character weight and so it went back
58:43 or I didn't do yeah character weight of
58:45 zero where it just uses her face so it
58:47 went back to her old colored clothing
58:50 the or from the original photograph
58:52 which that's kind of interesting um all
58:55 right so let's put um slash
59:00 imagine and then I'm going to dump in
59:02 that whole big prompt from um ideogram
59:05 then I'm going to
59:07 go Das D AR Let's
59:10 do let's do a portrait we'll do three
59:13 colon
59:14 4 and then we'll do dash dash
59:22 CF dash dash oops Dash Dash see
59:28 ref and let's go get her
59:36 face and yeah this is a hell of a lot
59:38 easier to do in the web version of this
59:40 CU you don't have to go copy links and
59:42 do all this bizarre [ __ ] weird ass
59:45 coding
59:46 [ __ ] um I know coders don't mind
59:49 doing this [ __ ] but most people
59:52 do okay and then um let's not do a style
59:57 ref on this one let's just let this
59:59 style be what it is we'll do dash dash
1:00:02 CW and I'll do a weight of 10 so so
1:00:07 we'll keep a little bit of the maybe her
1:00:10 hair um if you do a 100 it'll keep all
1:00:13 of her clothing if you do zero it'll
1:00:14 just be basically her face let's see
1:00:17 what this turns into so this should in
1:00:20 theory be some version of
1:00:23 that but you know mid Journeys in
1:00:26 interpretation of
1:00:43 that oh you said Deco here earlier
1:00:46 meaning
1:00:51 ideogram oh this one's cool so okay not
1:00:56 for nothing
1:00:58 people looky here this is this is where
1:01:01 we're getting into some things this is
1:01:03 the reason you play around with AI
1:01:06 because had you guys not said oh go do
1:01:09 this go do that I wouldn't have been
1:01:10 mixing and matching these things but
1:01:13 notice I didn't come up with that prompt
1:01:16 right someone else came up with that
1:01:17 prompt I'm like oh that's a striking
1:01:19 image and so I just went there grabbed
1:01:22 that prompt but because we're using a
1:01:24 character reference
1:01:27 I mean look at this one that is crazy
1:01:30 good upscale
1:01:33 three and then let's upscale this
1:01:36 creative that's [ __ ]
1:01:42 gorgeous so like you can get to just
1:01:45 truly remarkable
1:01:47 inspiring
1:01:49 work without having to have ideas on
1:01:52 your own and then what happens is what
1:01:55 my experience is
1:01:57 is I'm [ __ ] around with stuff and I'm
1:02:00 just like I don't know what to do and
1:02:01 I'll go over here and I'll try something
1:02:02 I'll go over here and I'll try something
1:02:03 I'll come on these lives and you guys
1:02:05 will give me ideas and then we'll kick
1:02:07 something out and that'll that'll spark
1:02:09 a whole new set of
1:02:11 ideas and just keep running down those
1:02:14 rabbit holes whether it's with words or
1:02:16 images or music or spoken word or video
1:02:20 it it literally doesn't matter or code
1:02:28 let these tools amplify your
1:02:36 intent amplify your
1:02:39 Humanity this is how we end up with
1:02:42 things that are way way
1:02:45 way beyond what
1:02:48 our capability would
1:02:52 be look at that holy [ __ ]
1:02:56 if you wonder why everyone Raves about
1:02:59 mid Journey this is
1:03:06 why this is
1:03:13 crazy so
1:03:18 good stunning isn't
1:03:21 it stunning
1:03:30 I can make chat GPT hallucinate pretty
1:03:32 much any time you got to stop giving it
1:03:35 mushrooms
1:03:40 dude looks like she's in a video game
1:03:42 doesn't
1:03:44 it they just like like that's totally
1:03:47 you know she's like from another
1:03:52 planet she's totally like from another
1:03:54 planet
1:04:00 crazy like look at the depth
1:04:06 right dang even pores I
1:04:11 know I
1:04:14 know it's just nuts okay so now this is
1:04:19 mid Journey Alpha so this is the website
1:04:21 version of mid
1:04:23 Journey so let's see if I go yeah yeah
1:04:27 so here's kind of my is that my history
1:04:29 I don't think I did these that's not me
1:04:31 is
1:04:32 it are these mine oh yeah they
1:04:38 are oh yeah this was I was doing Sydney
1:04:41 stuff I was doing stuff for the musical
1:04:43 that's what that
1:04:45 was
1:04:47 so right now if you've got a mid Journey
1:04:50 account if you made if you've made any
1:04:52 images in mid journey and you go to Mid
1:04:55 Journey
1:04:56 and sign in with your Discord account
1:04:58 you can see all the images you've ever
1:05:00 made which is really cool and I don't
1:05:02 think many people know
1:05:03 that if you've made more than a thousand
1:05:08 images you can get access to Mid Journey
1:05:11 alha which lets you create images see
1:05:15 where it says imagine right here well I
1:05:17 don't know if you can see it but right
1:05:18 there it says imagine and this is just
1:05:21 like the The Prompt box in Discord
1:05:23 except you don't have to type imagine
1:05:25 you can just type your prompt so let's
1:05:26 let's type let's put
1:05:31 in oh actually that's really cool I just
1:05:34 pasted the URL of that image and it put
1:05:37 it in there that little girl so so
1:05:39 there's three little icons here there's
1:05:42 the picture is we're just going to use
1:05:44 the prompt the the image as the prompt
1:05:47 the paperclip is we're going to use this
1:05:50 image as a style reference that function
1:05:53 as of 3 days ago was broken but we'll
1:05:56 test that tonight and then the little
1:05:58 character to the left there this is
1:05:59 using her as a character reference so
1:06:02 let's hop back over to
1:06:06 ideogram and grab this magic prompt from
1:06:10 the poppy the red poppy
1:06:13 girl hello my name's
1:06:16 poppy I'm a poppy
1:06:19 girl so we're going to put in that big
1:06:22 ass
1:06:23 prompt are we aren't we how did I not
1:06:26 copy
1:06:29 that copy maybe I hit
1:06:33 paste paste there we go okay and
1:06:38 then you should ask Claude to generate
1:06:41 her
1:06:42 biography using Vision oh that's that's
1:06:45 a pretty cool
1:06:46 idea yeah Claud just added Vision right
1:06:50 okay
1:06:51 cool
1:06:53 let's shrink this up okay
1:07:00 let's do it in that so so we're going to
1:07:02 do
1:07:03 portrait let's up the weirdness I'm
1:07:06 going to do the weirdness to
1:07:09 1200 and I'll bring stylization down to
1:07:13 200 all right that's good and then we've
1:07:16 got character reference here and I'm
1:07:17 going to put at the end of this you can
1:07:19 still use your little Dash dashes Dash D
1:07:22 CW I'm going to do a dash dash Oops I
1:07:25 did it in the wrong Place hang
1:07:32 on let me do it at the end here D-
1:07:42 CW zero we'll do we'll do zero so it
1:07:46 just gets her
1:07:47 face um and I think that's it
1:07:51 bang Okay so
1:07:57 and what's weird is it
1:07:59 didn't I don't know why it doesn't have
1:08:01 the ones I just did in Discord oh there
1:08:03 they are okay they're all in here now
1:08:09 yeah so here are the new four one the
1:08:12 the the new ones that are being
1:08:13 generated these last four
1:08:36 o she got creepy little skin
1:08:39 [Laughter]
1:08:44 bumps that's creepy
1:08:49 good okay so now now I'm going to do a
1:08:51 thing so you can either do variations of
1:08:54 this you can upscale it you can remix it
1:08:58 you can pan it or you can um Zoom so I'm
1:09:01 going to zoom out two
1:09:03 times and we should see more of
1:09:13 her Kyle where did the original image of
1:09:16 the woman in the red head come from that
1:09:19 came from IDR I was just scrolling
1:09:21 through
1:09:24 idag I was just it was just like
1:09:26 whatever was I was just kind of
1:09:27 scrolling until I saw something that
1:09:29 caught my eye and and then I just
1:09:30 clicked on it so it's
1:09:33 from Kus to
1:09:40 noou I don't know if you can see that k
1:09:43 r e u s number two n o
1:09:47 y u e
1:09:58 so here she is zoomed
1:10:04 out all right that's the one we
1:10:08 want yeah that one's the
1:10:12 one and then we're going to
1:10:15 do let's do strong variations of this
1:10:22 one and then we'll pick one of those and
1:10:24 we'll upscale it
1:10:35 oh curious to annoy you is that what
1:10:37 that name
1:10:39 was cure us to annoy you
1:10:43 yeah that's really funny k r e s curious
1:10:48 to Noy you that's pretty
1:10:52 cute a little too deep for my particular
1:10:59 that's
1:11:01 cute that looks like a skin
1:11:04 disease she looks terrified
1:11:16 there I think we're going to do upscale
1:11:19 not
1:11:21 creative we'll do
1:11:23 upscale we'll do upscale creative
1:11:26 on this
1:11:38 [Music]
1:11:39 one what's the topic now just logged in
1:11:42 uh we are playing with
1:11:46 um consistent characters in mid Journey
1:11:49 so we found an image of this girl with
1:11:52 orange eyes and then
1:11:54 we're playing around with you know can
1:11:57 we get her to show up in different
1:12:06 scenes pretty
1:12:11 amazing actually like the you know
1:12:13 what's amazing is like we we just went
1:12:15 from like you know we got her sort of
1:12:17 sledding her face and then you have that
1:12:20 looking feel right which is like that's
1:12:23 that and then maybe she has a nightmare
1:12:25 at night and this is what sledding is
1:12:27 like in her nightmare right like this
1:12:29 totally is inspiring a whole story about
1:12:32 this little
1:12:34 girl
1:12:40 [Music]
1:12:49 fascinating crazy look at that
1:13:11 [Music]
1:13:13 like even just ripping through those
1:13:14 images like that tells a kind of story
1:13:17 doesn't
1:13:19 it it's like the story of the
1:13:23 exploration I was watching Cory Cory
1:13:25 Sandler I don't know if you're on
1:13:27 tonight Corey but the way Corey
1:13:29 Sandler um Works she just she'll create
1:13:33 an image she does a lot of hers in Del
1:13:36 in chat GPT and she'll she just runs
1:13:39 down these rabbit holes trying to get a
1:13:41 look that she likes and if she can't get
1:13:43 it then she'll run down another rabbit
1:13:44 hole and she just keeps going and going
1:13:46 and going and going and it's like you
1:13:50 know this ability to just very very
1:13:51 quickly rip through these
1:13:54 things you know is a weird kind of
1:13:59 Storytelling you
1:14:00 know and what did it take us half an
1:14:03 hour and I'm you know I'm blabbing most
1:14:05 of the
1:14:07 time
1:14:13 um
1:14:17 anyway how is the face the same for each
1:14:19 image so I'm using a thing in mid
1:14:21 Journey called character reference
1:14:26 so you you add an
1:14:31 image so if I
1:14:35 want this crazy ass
1:14:38 dude I can say let me go let me go back
1:14:41 over to
1:14:42 ideogram let me find a picture of a
1:14:52 dude actually that would be kind of cool
1:14:54 let's grab this statue of
1:15:04 Liberty it's a really long
1:15:08 prompt be curious to know if it can even
1:15:11 do it all right so go back over here to
1:15:15 Mid Journey type in a long ass prompt
1:15:20 then I'm going to make this image the
1:15:23 character reference and I'm I'm going to
1:15:26 leave I'm going to leave the default
1:15:28 waiting cuz I want it to pull in the
1:15:32 hair and
1:15:35 then I'll turn weirdness down a little
1:15:37 bit because this is already going to be
1:15:40 weird and then I'll just hit
1:15:44 return and this should be interesting
1:15:46 we'll see what this does Kyle if you
1:15:48 have so much against rabbit why did you
1:15:52 buy one
1:15:56 um my my rabbit should be here in 10
1:16:07 days will that work in Del or just mid
1:16:10 Journey right now it's just mid
1:16:15 Journey but you can do it in
1:16:18 um Leonardo you can train a model in
1:16:20 Leonardo you can also do it in Deco here
1:16:24 has has a char character reference
1:16:27 thing
1:16:29 um I don't think you can do an nagram
1:16:32 right
1:16:33 now oh this is
1:16:44 cool pretty
1:16:52 amazing that one's actually quite
1:16:54 beautiful
1:16:56 I'm going to vary this
1:17:07 strong you bought an AI rabbit the
1:17:09 orange one yep what exactly does the
1:17:11 rabbit do so what it does Jesse is it
1:17:14 um does a couple of things here let me
1:17:17 jump over we'll chat Ed over there for a
1:17:20 minute
1:17:21 rabbit R1
1:17:26 so there's two things I find compelling
1:17:28 about it one is its industrial design is
1:17:30 just gorgeous like I love the scroll
1:17:32 wheel and then that little gray button
1:17:34 is like your it's like a walkie-talkie
1:17:37 but instead of talking to a friend
1:17:40 you're invoking the AI so so if you want
1:17:43 to do a prompt you just hit your
1:17:44 walkie-talkie button talk into it and
1:17:47 then it submits The
1:17:48 Prompt um if you double tap it it
1:17:51 activates the camera which is on a
1:17:53 cylinder as well so it'll flip towards
1:17:55 toward you or it'll flip away or if it's
1:17:57 not active it points straight up it's
1:17:59 got a little screen it's got a little
1:18:02 speaker and a
1:18:03 microphone
1:18:07 um and
1:18:10 so so it's a it's a little device to to
1:18:13 talk to AI but it can it can show you
1:18:15 things it can it can find things off the
1:18:17 internet but they've developed this
1:18:19 thing called the large action model and
1:18:22 what that does is you can train it up or
1:18:26 they've trained it up on some things
1:18:28 like you can have it send an email or
1:18:31 one of the examples they show is they
1:18:32 trained it up on how to make a
1:18:34 mid-journey image in Discord so you can
1:18:36 hit your little walkie-talkie button and
1:18:39 go hey make me a photograph of a little
1:18:41 girl sledding you know down a hill in a
1:18:44 in an illustration that looks like the
1:18:47 1800s and it will then write the prompt
1:18:50 and log into mid Journey for you and um
1:18:54 put that up there it's made I think the
1:18:55 company's called
1:18:57 rabbit the CEO of the company uh the the
1:19:01 company that did the Industrial design
1:19:02 is called teen teenage engineering they
1:19:05 make musical instruments like electronic
1:19:07 really cool electronic musical
1:19:09 instruments and the CEO of rabbit um had
1:19:13 a previous company that he made a cool U
1:19:16 musical instrument with them a
1:19:19 controller so let's see what we got here
1:19:27 that's
1:19:28 bad the the chin
1:19:32 dribble all right well the first one's
1:19:34 really good that one's really good
1:19:36 that's very disturbing let's do upscale
1:19:39 creative oops not that one oh
1:19:42 crap that one that
1:19:45 one oh let me see what it's
1:19:51 doing there we go
1:20:00 surprised it didn't have a musical
1:20:01 component the
1:20:03 rabbit yeah Oh you mean like an explicit
1:20:05 one maybe it's got an Easter egg in it I
1:20:08 don't know I mean I think quite honestly
1:20:10 I think what the rabbit
1:20:12 R1 is is probably just a custom circuit
1:20:16 board that's it's it's probably just a
1:20:18 Raspberry Pi with with a cool case
1:20:20 around it and then they're just you know
1:20:22 connecting it via Wi-Fi or Bluetooth to
1:20:26 the
1:20:27 internet and then it's it's using your
1:20:31 computer to um to do all the all the
1:20:35 stuff and it's also they did a deal with
1:20:37 perplexity so it's got perplexity
1:20:38 sitting underneath it all right so
1:20:41 here's
1:20:46 this
1:20:51 um did I do that
1:21:01 why did that why is that
1:21:03 not big maybe it wasn't
1:21:06 finished ah there we go
1:21:25 [Laughter]
1:21:29 uh save image
1:21:31 as
1:21:35 um
1:21:43 crying
1:21:44 k
1:21:47 s butter Chell
1:21:57 see we go to
1:21:59 LinkedIn and what we do is we
1:22:11 go in
1:22:14 1784 my great great
1:22:44 and there now we'll do
1:22:48 uncle
1:22:52 uncle k s
1:23:00 belli is that how you spell
1:23:02 belli
1:23:19 oops squandered his talents
1:23:25 in
1:23:28 obscurity and the
1:23:31 bottle
1:23:32 period here's one of the
1:23:38 self-portraits he
1:23:40 created before his untimely
1:23:44 death in an eel fishing accident
1:23:53 [Laughter]
1:24:06 oh squandered in
1:24:09 1784
1:24:10 comma my great uncle Kos belli
1:24:13 squandered his talents in obscurity and
1:24:16 the bottle here's one of the self
1:24:19 portraits he created before his his
1:24:21 untimely death in an eel fix fishing
1:24:24 accident his untimely death
1:24:28 at at
1:24:32 32 in an eel fishing
1:24:38 accident oh is it it's not is it going
1:24:40 to be able to see that yeah
1:24:43 good all right there we
1:24:50 go so that's now up on LinkedIn
1:25:00 [Laughter]
1:25:07 oh it's so
1:25:09 good oh you have a very talented
1:25:12 [Laughter]
1:25:16 family it's
1:25:18 comedy Kyle laughing at Kyle I just I
1:25:22 just imagine people going is is he
1:25:25 mentally stable we know the
1:25:30 answer an eel fishing accident laughing
1:25:34 my ass
1:25:42 off
1:25:46 oh yes belli okay good spelled it right
1:25:50 look at
1:25:51 me that's Beethoven right no that's me
1:25:56 that's one of my that's one of my Kyle
1:25:58 Shannon dreams
1:26:02 images all right um what other questions
1:26:06 you got oh look Ron Morris just liked it
1:26:08 I guess Ron's in here Hey
1:26:15 Ron all right any other
1:26:19 questions did you see that meta can
1:26:22 generate images in WhatsApp Now using SL
1:26:25 imagine I did not didn't are are they
1:26:29 doing that with with mid
1:26:31 Journey Brandon I kind of remember meta
1:26:34 said they cut a deal with mid Journey am
1:26:35 I remembering that
1:26:39 right I don't have
1:26:43 WhatsApp where did you order that rabbit
1:26:45 from I ordered it from that site from
1:26:47 the rabbit
1:26:50 do rabbit.
1:26:53 te 199
1:26:55 bucks the other reason I ordered it
1:26:58 quite frankly is for 199 bucks I mean
1:27:01 it's not cheap but like the the Humane
1:27:04 pin is $700 for that stupid little thing
1:27:08 that shoots the laser in your hand
1:27:10 that's overheating thanks I'm getting
1:27:11 one now like here here's my here's my
1:27:14 justification for
1:27:16 this this is absolutely this is a
1:27:19 product in my opinion that Apple should
1:27:23 have produced like apple is always the
1:27:26 one who reinvents the category right
1:27:31 whether this thing succeeds or not is
1:27:33 irrelevant but this reinvents the phone
1:27:36 category by saying hey the whole idea of
1:27:39 apps and having to navigate through
1:27:41 dozens or hundreds of apps in my case
1:27:44 cuz I'm a I'm a digital hoarder so I
1:27:46 just I never uninstall
1:27:49 anything um and this basically says
1:27:52 instead of all those apps you just have
1:27:54 a button and you just tell the phone you
1:27:56 just tell this device what you want
1:27:58 it'll go figure [ __ ]
1:28:00 out that's something that Apple should
1:28:03 have invented not these guys right so so
1:28:06 one is it's the first of its kind as a
1:28:10 category with a screen and then the
1:28:13 design is so slick that it's like even
1:28:16 if it's a piece of [ __ ] it's going to
1:28:18 look really good sitting right there
1:28:22 there yeah yeah right in front of see
1:28:24 the side quest drive right there see
1:28:26 that little orange box it's going to go
1:28:28 right there if I don't use it but if I
1:28:31 do use it cool like if it's good great
1:28:34 but if
1:28:36 not it goes in the garbage
1:28:38 [Laughter]
1:28:48 pile it'll be a collection piece of art
1:28:51 yeah exactly absolutely
1:28:56 yeah um yeah batch one just shipped and
1:29:00 listen they said batch one was going to
1:29:02 ship in
1:29:04 March if you've ever done a Kickstarter
1:29:08 if you've done ever done anything in
1:29:11 Hardware when when the company says
1:29:15 we're going to ship the first one in
1:29:18 March it might be March but it's often
1:29:21 March of two years down the line they
1:29:24 actually sh in March they actually
1:29:25 shipped from the factory in China or
1:29:27 wherever they're being made on March
1:29:29 31st and they're going through customs
1:29:31 and all that stuff right now so third
1:29:34 week in April we're supposed to have
1:29:35 them April 24th I think is the US the US
1:29:38 date where they're shipping or where
1:29:40 they're supposed to arrive I don't
1:29:41 [ __ ] know
1:29:43 um so that's that's promising because
1:29:47 that means they have their [ __ ] together
1:29:49 to some
1:29:53 degree you have a tool for 2D to 3D
1:29:57 conversion um I don't yet all the all
1:30:00 those tools are eh they're okay they're
1:30:02 not great um I figure another 6 months
1:30:05 to a year for the 3D tools to not
1:30:08 suck um but they're going to they're
1:30:10 going to get there like again I I think
1:30:12 that
1:30:14 most I think by the end of 2024 most
1:30:20 tools will
1:30:22 be really good good the music stuff's
1:30:26 going to be really good the coding
1:30:28 stuff's going to be better we're going
1:30:29 to have agents large language models are
1:30:32 going to have
1:30:33 reasoning we're going to have seven or
1:30:37 eight gp4
1:30:39 quality large language models and then
1:30:42 we'll probably have one or two GPT 5
1:30:46 quality models that'll be multimodal and
1:30:49 crazy context window lengths like by the
1:30:51 end of the year
1:30:54 these things are going to be in the in
1:30:56 the neighborhood of good enough to do
1:30:58 almost any work in in my humble opinion
1:31:03 well there's nothing much humbl about
1:31:05 him all he does is spout on about what
1:31:08 he thinks about
1:31:09 everything I don't know if you knew that
1:31:13 Luma is decent I did a I made a 2d to 3D
1:31:16 tool but it's not working perfectly
1:31:18 there you go Luma on Discord has one
1:31:21 can't remember the exact name of it
1:31:25 so that's two two votes for Luma AI
1:31:27 daily said llama 3 and GPT 5 are near in
1:31:31 completion
1:31:33 interesting yeah I think um gp5 went
1:31:36 into red teaming last week or the week
1:31:39 before red teaming is where they try to
1:31:42 break it and get it to do all sorts of
1:31:43 evil
1:31:44 [Laughter]
1:31:49 [ __ ] I've been using pi tonight to coach
1:31:52 me through how to share Google app
1:31:55 script #
1:32:00 clueless perfect but that's that's the
1:32:03 deal I mean yeah use these tools to help
1:32:06 you do
1:32:08 everything can't wait for the actual
1:32:10 agent I know the agents are going to be
1:32:11 nice I just I want to even be able to
1:32:13 play with Devon because Devon looks
1:32:15 pretty [ __ ] slick just to watch it do
1:32:18 what it
1:32:21 does although it is going to run up your
1:32:24 chat gptt for Bill Lex fredman
1:32:27 interviewed Sam Alman this week really
1:32:30 was it this he he interviewed Alman this
1:32:32 week no
1:32:38 really did I miss that I well
1:32:41 obviously let's
1:32:44 see Lex
1:32:48 Freed
1:32:50 Man alt man
1:33:01 March
1:33:03 18th that was with Ilia
1:33:07 sver internet's about to
1:33:11 break that was March
1:33:14 18th wait that was
1:33:17 sover March
1:33:22 20th I think I saw that one yeah I I
1:33:24 don't think that was this week was it
1:33:26 was there a new one this week I think
1:33:29 that was March
1:33:32 18th the problem is I don't have $300 a
1:33:35 month for for all of these $20 a month
1:33:38 subscriptions it's so bad I'm paying
1:33:40 what am I paying I'm paying 30 bucks a
1:33:42 month for [ __ ]
1:33:45 um the [ __ ] am I paying 30 bucks a month
1:33:50 for one of the image tools
1:33:56 I don't even remember yeah it's
1:33:58 ridiculous I've got a little bit of an
1:34:00 R&D budget at storyvine some that's
1:34:02 where I'm paying for some of these
1:34:04 things from but no Runway I'm only on a
1:34:08 mid Journey St no I'm on a $10 mid
1:34:11 Journey subscription and a $10 um Runway
1:34:15 subscription
1:34:25 playing with pythagorea right
1:34:27 now what's pythagorea Joker that sounds
1:34:31 cool the [ __ ] is
1:34:34 that
1:34:38 piaga
1:34:47 AI p
1:34:58 pythagora completely changes The View on
1:35:00 software development
1:35:03 timelines New Era of software
1:35:05 development begins backed by y
1:35:08 combinator watch
1:35:13 demo hi and welcome to pythagora Dev
1:35:16 tool that builds production ready apps
1:35:18 from scratch by talking to you with
1:35:21 pythagora we don't just generate code we
1:35:23 build apps so let me show you how we do
1:35:26 that when you start building your first
1:35:28 app first you have to give project
1:35:30 description at that point you will meet
1:35:32 our first agent spec writer who will
1:35:35 help you write better description in
1:35:38 case it's needed he sounds like the he
1:35:40 sounds like the uh John malovic Russian
1:35:44 dude in uh in that Matt Damon movie you
1:35:49 know you know which what I'm talking
1:35:50 about it's like I will Splash the pot
1:35:54 when I want to Splash the
1:35:57 pot it was funny cuz Matt Damon was
1:36:01 talking about malovich and malovich was
1:36:04 like I'm a really bad actor and I'm just
1:36:06 going to push this after the spec writer
1:36:08 is done as far as I can until they tell
1:36:10 me to
1:36:12 stop writing the project description
1:36:15 then architect agent will take that
1:36:17 project description and create
1:36:19 architecture for this specific project
1:36:22 yeah that looks pretty cool next one is
1:36:23 a tech so this looks like a this looks
1:36:25 like a Devon something in the
1:36:26 neighborhood of Devon all right so
1:36:28 that's kind of full p
1:36:31 pyora p y t h a g o r a.
1:36:36 a pythag pythagora
1:36:41 pythagora super
1:36:46 [Music]
1:36:49 cool Russian accent I need this right
1:36:53 now
1:36:56 oh source Camp there you go P there you
1:36:59 go this is why we come here this is why
1:37:01 we hang out on a Saturday night with an
1:37:03 old dude trying to learn how to use Tik
1:37:06 Tok Devin was a fake
1:37:10 demo at developing software code it
1:37:13 wasn't a fake demo it was just a demo
1:37:15 that hasn't been released yet but people
1:37:18 who have used it so
1:37:20 so it's it's very easy to say that it
1:37:23 was
1:37:25 you know that it was a fake demo it
1:37:27 wasn't a fake demo it was a it was a
1:37:28 demo it was a marketing demo so it may
1:37:31 or may not work like the marketing demo
1:37:33 but it's real software that people have
1:37:34 used it's just not out yet same with
1:37:37 Sora like Sora You could argue that Sora
1:37:40 was a fake demo um but people have used
1:37:43 that too and you
1:37:45 know it's just we're we're early so so a
1:37:48 lot of these tools they're announcing
1:37:50 them to get a little bit of Buzz and
1:37:52 then but with the one you won't need
1:37:55 apps well you still need apps but it's
1:37:58 it's sort of triggering them it's
1:37:59 managing all the apps in the
1:38:05 background so I guess it does not rhyme
1:38:08 with
1:38:15 diarrhea
1:38:17 oh a little like Boris badenov and
1:38:19 Bullwinkle he was a little like that
1:38:24 [Music]
1:38:26 if you can say it correctly we'll give
1:38:28 you one for free that's
1:38:32 great yeah some people say Gemini was a
1:38:35 fake demo like I I don't think well yeah
1:38:40 listen a marketing video is not a demo
1:38:44 of the software let's put it that way
1:38:47 and so to the extent that that you
1:38:51 know the the uh Gemini Ultra thing they
1:38:55 didn't launch
1:38:56 it they probably showed some elements of
1:38:59 that that were actual interfaces for
1:39:01 something they have in a lab somewhere
1:39:03 they just didn't launch them hang on I
1:39:05 just drop
1:39:14 something come
1:39:17 on what's wrong champy you got to go out
1:39:41 ah there was a boy in the kitchen
1:39:43 Champion was looking for food he was
1:39:44 looking for handouts he's just a taker
1:39:47 he's just a taker of that
1:39:50 champy all right
1:39:53 and
1:39:54 [Laughter]
1:40:00 squirrel so what did you do Saturday
1:40:02 night exactly all right listen I'm gonna
1:40:05 get out of here um do me a favor if you
1:40:09 be so kind follow my channel I still
1:40:12 have not hit 35,000 that last 100 is a
1:40:15 [ __ ] uh follow my channel if you
1:40:17 haven't and um subscribe to the lives
1:40:21 pick up one of the uh the uh
1:40:24 video series in the corner there if you
1:40:26 want to support the channel cheese
1:40:28 please yeah exactly um and keep coming
1:40:31 back so I'll see you tomorrow night
1:40:33 happy Saturday night to everybody um
1:40:36 tomorrow is Sunday usually comes after
1:40:39 Saturday anything going on tiger hosted
1:40:43 at the Masters today
1:40:45 so he won't be in the in the final round
1:40:49 there but uh it looks like it could be a
1:40:51 good one anyway uh all right cool good
1:40:55 seeing y'all hope this was
1:40:57 useful it's been it's been lately it's
1:41:00 been like let's build some [ __ ] you
1:41:03 know so so hopefully it's been useful
1:41:06 all right peace out have a good one