Apr 25, 2024 - How AI is Revolutionizing Video Storytelling

Description

Chapters

Transcript

0:01 [Music]
0:21 oh oh
0:25 [Music]
0:31 oh
0:35 [Music]
0:43 [Music]
1:04 there's been something baby I've been
1:06 trying to
1:07 [Music]
1:09 say for an age and it seems I don't know
1:15 how pass in the future now surrounding
1:20 me surrender to whatever tree can be
1:25 found there's been a little
1:28 trouble since you came to my
1:32 [Music]
1:35 rescue and if you like all of the rest I
1:38 would have quit you long ago but I
1:41 couldn't do
1:45 that oh tell me now why I never went to
1:52 well make my crazy making cold as
1:55 [Music]
1:57 hell I don't want to let you wish me
2:02 well but try still going to have to find
2:08 own way
2:09 [Music]
2:15 through good evening good
2:18 people what is happening I know I took a
2:21 night
2:23 off I'll tell you what you spend seven
2:26 hours in the
2:29 capital takes it out of
2:32 [Music]
2:38 you you know what I realized is
2:41 government is full of
2:43 people those people in high
2:45 school that did their
2:48 homework and learned Like
2:52 rules and followed
2:54 [Music]
2:58 them so I felt a little out of
3:00 [Music]
3:17 place you really didn't get a night off
3:20 I didn't I was I was [ __ ]
3:23 toasted I was just I didn't get home
3:25 till like 8:30
3:31 yeah we were supposed to the our session
3:33 was supposed to start at 1:30 the
3:36 session didn't even start until 4:30 and
3:38 it's not like they're keeping you up to
3:40 date you're just like
3:42 there and you know you're just supposed
3:46 to know where to go and what to do you
3:48 know like read manuals and follow
3:51 [Laughter]
3:53 [Music]
3:57 rules getting how that an F major
4:04 [Applause]
4:07 chippy yes he can
4:12 [Music]
4:23 [Music]
4:42 do me a favor let's I I want to try
4:44 something tonight um let's try to get a
4:47 as many interactions as we can go and so
4:49 like tap the screen do I don't know
4:51 whatever the whatever the other ways to
4:53 interact are
4:55 um I feel like Tik Tock
4:58 has has forgot in this channel a bit and
5:01 they're not getting us out there
5:04 enough and you know listen to be
5:09 clear that that could be very much on
5:13 the host of this
5:16 channel you can make
5:20 money it could be that this isn't
5:22 getting pumped in the algorithm cuz no
5:24 one wants to watch it that is possible
5:30 [Music]
5:33 but I have no idea how the algorithm
5:34 works and so and I and you know being a
5:37 non-rule follower I'm not going to do
5:40 their stupid [ __ ]
5:42 [Music]
5:51 battles well it is over 18 after all
5:54 that dog dog loves his Kyle that dog
5:56 does love me you know that dog the
5:58 minute I leave the house he lays by the
6:02 door for eight hours just waiting for me
6:05 to come
6:06 back pisses everyone else off they're
6:09 like we want to love the dog too but
6:11 he's just waiting for
6:12 [Laughter]
6:16 you I didn't even want the stupid
6:21 [ __ ] we go to the we go to the shelter
6:25 there was a a black lab there and he
6:27 just just wasn't right didn't have the
6:29 right personality just wasn't right and
6:32 cuz we like Labs we've always had Labs
6:34 so we went there to get a lab went there
6:36 to get a
6:38 lab and
6:44 then and then that thing shows
6:47 up that's not a
6:49 Lab look at that thing look at that
6:53 thing that's a bad dog that is a large
6:56 tubby boy that is a tubby Tubby Dog are
7:00 you a big fat boy who's a big fat boy
7:03 are you a big fat boy oh Jam's a big fat
7:06 boy he's so big so anyway so I didn't
7:11 even want him I'm like you know what's
7:13 the return policy on this thing oh and
7:18 then
7:21 then they they proceed to tell
7:24 us so he take him out I'm like this
7:27 thing's Ramy this this is a Ramy dog
7:31 so so I'm like What's the return policy
7:34 on this thing you know joking just being
7:35 an [ __ ] dad and uh so we go to get
7:39 him and they're like just so you
7:41 know
7:43 champ has been returned
7:46 twice and if you return
7:49 him it's it's off to the it's off to the
7:54 chamber
7:56 so so the pressure was on
8:02 I tried to I tried to hate him I tried
8:04 to not like him but then I played my
8:06 guitar and he sang and I was like ah
8:09 [Music]
8:15 [ __ ] but I think that's why they're
8:16 pissed off they're like you didn't even
8:18 want him and he loves you more I'm like
8:21 well who else Ruff House is with
8:24 him you know how you got to play with a
8:26 dog like that that little Thug of a dog
8:30 you gota you gota you gotta Whack Him in
8:33 the nose and wrestle with him beat his
8:35 belly oh he loves it yeah dogs are the
8:39 best Joker dogs are the [ __ ]
8:43 best I feel like chat GPT is a bit like
8:45 a golden retriever which just like hey
8:48 boss I'm ready to write some more [ __ ]
8:50 for you what do you want you want me to
8:52 do it I'll do it good this
8:57 time llm is unconditional love for your
9:00 your creative
9:02 [Music]
9:27 braak so
9:31 I thought my my rabbit R1 was supposed
9:33 to come today but I think it's it might
9:35 come tomorrow I don't know I don't have
9:37 an official tracking number from
9:40 [Music]
9:42 them but
9:44 um some people are starting to receive
9:52 theirs I'm Out Of Tune here
9:58 [Music]
10:01 um I have two 80 plus Peta dogs hugging
10:05 the blankets that's the only way to do
10:07 it Sharon here I will teach you I I'll
10:09 teach you what I learned in in dog
10:11 training
10:12 school so we had two inbred Labs our
10:18 first two Labs were two inbred Labs from
10:20 a a puppy pound or puppy
10:23 mill they were inbred and they were
10:26 [ __ ] crazy as [ __ ] loons
10:31 the the first experience it was uh Max
10:34 and Cleo and she she was uh like a
10:37 little blonde lab and he was red and so
10:42 we we get these P we get a pair my my
10:44 mom gave them to us cuz you know if
10:47 you're if you're going to if you're
10:48 going to give people a dog why not give
10:50 them
10:52 two and make them Labs something that
10:55 has boundless amounts of energy so so he
10:59 get
11:00 the things and we're walking around uh
11:03 you know with the puppies in the lawn
11:04 and the puppies are doing their thing
11:06 and Max goes to you know do what dogs do
11:09 take a poop right Cleo just eats the
11:14 poop fresh out of the fresh out of the
11:17 [Music]
11:21 funnel she spent the rest of her life
11:23 eating poop she loved it she loved poop
11:28 and we're Googling how do you what do
11:30 you do if your dog likes poop you know
11:33 like vitamin deficiencies give it a
11:35 nugget I you know no she just liked it
11:38 some dogs like poop so so that was how
11:42 the relationship with them began and
11:44 then were like we're like two years into
11:46 owning them and they're just [ __ ]
11:47 nuts they're just nuts and you can't
11:50 walk them they're not social and so we
11:53 hire a dog trainer and then the the dog
11:56 the dog trainer on the first day
12:00 tells us tells us that you know when you
12:04 know when it comes time to train dogs
12:06 we're not training the dogs we're
12:08 training the
12:09 owners and then he told us what all the
12:11 rules were like you can't let them up on
12:14 the couch or up on the bed and you got
12:16 to have rules and you got to be
12:18 consistent and I was like ah I
12:21 understand now I and then in that moment
12:24 on that day with our very first dog
12:26 trainer I accepted the fact that I would
12:28 have poorly behaved dogs for the rest of
12:30 my life and indeed I
12:38 have oh
12:41 champy you should see champy attack the
12:46 maale [ __ ] rips it through the mail
12:48 [Music]
12:52 slot we we got we got a we got a
12:55 postcard from the from the post office
12:59 your dog has been flagged as an [ __ ]
13:01 you better get your [ __ ]
13:06 together I'm like yep that's right in
13:09 keeping oh exactly right exactly why
13:12 have a dog if you can't snuggle with it
13:14 on the couch exactly cuno yeah
13:17 boundaries my ass get up on bed and come
13:19 here and
13:22 snuggle
13:25 oh I love custom GPT what is my dog
13:28 thinking
13:31 oh man my dog's a little crazy too I can
13:34 totally
13:39 relate did my package ever arrive at
13:41 your
13:43 office maybe when did you send
13:47 it I don't think so
13:50 karuna but it might we've got my office
13:52 has got like a special mailbox area and
13:55 Monique my co-founder's been out did you
13:57 was it recent or was it a while ago
14:00 she's she's been out in Europe for so
14:03 like we probably haven't checked mail
14:04 two weeks ago yeah it's it it's probably
14:07 still in the mailboxes I'll I'll
14:09 I'll do I
14:15 have I'll have to see if I should have
14:18 an email that we got something
14:21 so but I don't have the mail key so she
14:24 gets back Sunday I think so I'll
14:28 check um but very cool thank you I don't
14:32 I thank you in
14:35 advance oh man all right so what's going
14:38 on
14:40 um I got access to LTX Studio which is
14:44 interesting so we'll play with that
14:47 tonight um if you don't know what LTX
14:50 studio is it's basically put in a prompt
14:53 and it makes a movie
14:57 now there's J and then there's janky and
15:01 this is this is even this is jankier
15:04 than than most Jank it's it's a very
15:06 janky piece of software but it's
15:10 promising like it looks like the kind of
15:12 thing that in a year it's going to be
15:15 pretty interesting if they can survive
15:18 right
15:20 um I know that the two Founders I think
15:22 it's two Israeli guys um they're not
15:25 Hollywood guys they're just large
15:27 language model guys and it's what's
15:30 what's cool you know I'll show this to
15:31 you
15:32 tonight we do enough work in
15:36 here that what they've
15:39 done is just a fairly sophisticated set
15:42 of
15:43 prompts that's using some relatively
15:46 straightforward large language
15:48 models um but they did it in a in a
15:50 really clever way they did it in a way
15:53 that is just kind of this modular
15:56 storytelling engine that I'm a big fan
15:59 of cuz that's kind of what Story vine is
16:01 my you know my company is a modular
16:03 storytelling platform so it's it's it's
16:06 it's got it you know it's it's got
16:08 elements that are very familiar to to
16:11 the way I think think about stories and
16:13 the way we do at Story vine um and it's
16:16 pretty slick it's pretty slick but it's
16:20 like like the first thing that I made it
16:21 you know it made like a 27 second video
16:24 but it's a 27 second video with five
16:26 scenes and three scenes each or three
16:30 shots
16:32 each and there's like a little voice
16:34 over so so all the stories are kind of
16:36 the same they have voice over and
16:39 they're just shots kind of stuck
16:40 together but it's kind of cool so we'll
16:41 play with that and then there was a post
16:43 from Runway today that um I didn't dig
16:47 too deep into it I had a really busy day
16:49 today
16:54 um
16:56 um and Runway Runway post we'll go look
17:00 at that too cuz runway's post was
17:02 basically like the future of
17:05 AI is in models like Sora and Runway
17:09 that actually understand the physics of
17:11 the world wait wait how was Capitol Hill
17:14 Capitol Hill was good it was good good
17:16 good it was okay so let me let me tell
17:19 you so here's how this all went
17:22 down so we we got approached we so this
17:25 bill like dropped out of nowhere the the
17:28 the context on the bill here's this is
17:31 my
17:34 very non-politically Savvy
17:38 uninformed
17:40 um guess not guess
17:44 speculation
17:47 um one of the things I learned yesterday
17:49 is Colorado was very late on data
17:54 privacy
17:55 legislation and apparently like they
17:58 still don't have their [ __ ] together
18:00 and you know States like California got
18:02 it together
18:03 first
18:05 um the Senate Majority Leader Rodriguez
18:08 who's put this bill forward I I just get
18:11 the sense he wants something through so
18:14 that he can say we've got legislation on
18:16 the books for AI harm or whatever like
18:20 you know in this case he's talking about
18:22 uh you know systematic bias within the
18:25 within the systems is what the what the
18:26 bill is about on its surface um
18:29 [Music]
18:30 so I think they just want to get
18:31 something through so it was dropped
18:33 really quickly without a lot of
18:35 fanfare and so Dan Murray from the Rocky
18:40 Mountain AI interest group and
18:43 myself basically paired up on behalf and
18:47 me for the AI Salon him for the Rocky
18:49 Mountain interest group and then we
18:52 worked with this team that knows about
18:54 this
18:55 [ __ ] and you know learned about it and
18:59 and so the the hearing was
19:03 yesterday
19:05 and Monday we draft you know we had a we
19:08 had a letter of a draft we had a a
19:10 letter drafted we got as many people as
19:12 we could to sign it this team vetted all
19:15 the people from Colorado and made sure
19:16 they were all real and you know their
19:19 businesses were businesses and all that
19:20 sort of [ __ ] um we did a we did a press
19:25 conference on Tuesday where there there
19:29 were three of us three small business
19:32 people um on a like a zoom call with
19:37 reporters and then right after that we
19:39 dropped the letter and the press release
19:42 which said more than a 100 Colorado
19:44 businesses opposed this bill and then a
19:48 bunch of us signed up to testify at the
19:51 hearing yesterday and so that's that's
19:53 what that's what I spent seven hours at
19:55 the capital doing was just mostly
19:58 waiting um
20:00 and then the hearing was kind of
20:01 fascinating like I'd never been in one
20:03 before and they would bring up four
20:05 people at a time they they um Rod
20:08 Rodriguez the the House Majority Leader
20:11 sort of presented to the it was the H
20:13 the Senate Judiciary Committee so there
20:15 was five five people on the committee he
20:18 presents the bill to them and then they
20:20 invite up you know testimony and they
20:24 started with opposition testimony and
20:27 then um so so when you register to
20:31 testify you you choose support oppose or
20:35 amend right and so they did oppose first
20:39 amend second and support
20:42 third and uh they would bring up four
20:45 people at a time and it's just like you
20:47 see on TV they had the little table with
20:48 the little long microphones that you
20:50 know you have to turn on the microphone
20:51 each time you
20:53 talk and then you give your statement
20:56 and then all four people would give
20:58 their statement and and then the
20:59 Senators would ask that batch of people
21:01 questions and then they'd excuse them
21:03 and the next four would come up and the
21:05 next four and so and so that's what it
21:07 was and it it was it was a really
21:09 fascinating process and like you know
21:11 some of the Senators could not give less
21:15 of a [ __ ] they didn't ask questions they
21:17 were checked out they were like just
21:20 this side of nodding off um other
21:23 senators were super engaged uh and uh
21:28 but but you know it was long and like
21:30 you know
21:31 about 7:00 like we're just getting
21:35 rolling and it's like 7:00 at night and
21:38 apparently like like they just keep
21:39 going like this is this is how they do
21:41 it and uh so at some point like the
21:44 clerks in the senate in the in the
21:48 hearing room the clerks start bustling
21:50 about and then 10 minutes later they're
21:53 bringing in like fast food so all this
21:56 fast food comes in for all the Senators
21:58 and the Senators just eating fast food
22:00 and like we're passing out there's no
22:02 water there's no food there's no nothing
22:04 you're just sitting there
22:08 like so I think I I think I lasted until
22:12 like eight eight or something like that
22:15 and then uh and then took off but but it
22:17 was good I did did my testimony got some
22:20 questions um I got quoted one of the
22:24 things I said was um something like this
22:27 technology is literally changing on a
22:29 weekly basis and that became the
22:31 headline of one of the news stories it
22:33 said you know in quotes this technology
22:35 is literally changing on a weekly basis
22:37 and I got quoted in there so that was
22:39 pretty cool and then here I'll show
22:41 you I'll show you I got picked up the
22:44 our press conference got picked up by
22:46 some some news news channels as well
22:51 here's the local
23:06 Clues a state committee will take up a
23:09 bill today that would protect consumers
23:11 when it comes to artificial intelligence
23:13 Denver sens W Reese joins us live from
23:15 the state capital in W part of this bill
23:17 is getting some major push
23:20 back good Wednesday morning to you Brian
23:22 and Veronica this bill it's very broad
23:25 but one part of the bill very
23:27 interesting it says that businesses are
23:29 going to have to let consumers know when
23:32 they're using AI so let me just walk you
23:34 through this this morning the bill also
23:36 says that any synthetic digital content
23:39 produced by an AI system must be marked
23:42 and detectable get this though any
23:45 content produced by an AI system must be
23:48 marked and
23:51 detectable right so if you do anything
23:53 with AI you have to mark it right that's
23:55 what they're saying here more than 100
23:57 businesses develop ERS and entrepreneurs
24:00 who focus on AI they're just not on
24:02 board with this bill business owners
24:04 feel the bill is being rushed and claim
24:06 that there was no input from the
24:07 community or from small business owners
24:10 during a press conference earlier this
24:12 week a business owner made a plea to
24:15 lawmakers I urge lawmakers to consider
24:18 the implications of how the current
24:19 overly broad Bill might smother
24:21 innovation in fact I strongly encourage
24:25 uh lawmakers to engage with AI experts
24:28 with entrepreneurs with leading business
24:29 figures who are familiar with the
24:31 Dynamics and potential of AI the
24:33 opportunity before us is truly
24:37 transformative and we're talking about
24:39 some big impacts we also learning press
24:43 release that one business owner says
24:44 that if this were to become a law he
24:47 would consider maybe moving his business
24:49 out of state of course we're going to
24:51 stay on top of this El there is going to
24:55 be a Committee hearing today but Eli
24:57 that test that the press conference
25:00 actually he and Kelly both there there
25:02 was a a man Eli wood and Dr Kelly KBW I
25:06 think is their last name from Boulder um
25:09 they both said they would leave the
25:11 state if this thing passed and and a
25:13 bunch of people said that a bunch of the
25:14 people that opposed the bill said that
25:17 so I don't know I've seen I've seen
25:19 about 10 different articles on this and
25:22 and a lot of what was said in in our uh
25:25 press conference is quoted in all of
25:27 those articles and then some of what we
25:29 said in our testimony was quoted in
25:31 other stories
25:32 so so I don't know I think it's good the
25:35 it it passed 3-2 it passed along party
25:39 lines uh Democrats
25:41 3-2 um and it passed to either amend it
25:46 or kill it right it it didn't pass like
25:48 it passed but it like it made it to the
25:50 next it didn't get killed in in
25:52 committee um you should have told him
25:55 you could make money with chat gbt
25:59 should have taken my [ __ ]
26:04 button but anyway it was absolutely
26:07 fascinating it was it was really kind of
26:08 cool you know
26:11 so
26:20 yeah all right you left the button at
26:27 home yeah Colorado this was Colorado
26:30 Bill and and by and also this
26:34 bill is and Rodriguez talked about this
26:38 in the in the hearing this bill is
26:41 copied from a from a Connecticut bill so
26:45 Connecticut has very little Tech
26:47 Innovation they don't they have very
26:48 little Tech stuff um and so basically I
26:53 guess there's some working group of of
26:56 state senators that
26:59 have decided that you know they're tired
27:01 of having their asses handed to them
27:03 with like big legislation like this and
27:07 so they sort of got together and
27:09 Connecticut went first and then Colorado
27:11 basically just copied it so it's a
27:13 copycat bill from Connecticut and then
27:16 as they're amending the Connecticut one
27:18 like while we were testifying like it it
27:21 came out that there's there's an
27:22 amendment that dropped like within the
27:24 past 24 hours um that we were commenting
27:29 on a previous version of the bill but
27:31 then some of the other people in the
27:32 room had gotten the new version of the
27:34 bill and they were commenting on it and
27:35 said it was still a piece of [ __ ] so I
27:37 don't
27:38 know we'll see fascinating fascinating
27:41 like like I said this is a world I don't
27:43 know I'm not comfortable in it I you
27:45 know it it was a very
27:50 um just I you know I was way outside my
27:54 comfort zone cuz I know nothing about
27:56 this [ __ ] and like you know and like as
28:00 a guy who doesn't read a
28:03 manual I don't know all the Nuance
28:06 details of a 35-page [ __ ]
28:09 bill right I know the the broad stroke
28:12 so like it's a little stressful like you
28:14 know have a reporter ask you well what
28:16 do you think about this specific part of
28:18 the
28:18 bill so anyway I think I survived it got
28:24 got quoted a couple of times and you
28:26 know we got some coverage so I think
28:27 that was that was the
28:29 goal um and apparently you know be
28:32 because the these are some of the early
28:36 like the feds aren't doing [ __ ] with AI
28:38 right they aren't doing any legis
28:39 they're not passing [ __ ] because you
28:41 know they they're completely [ __ ]
28:43 dysfunctional so this is falling on the
28:45 states so connetic Connecticut and
28:47 Colorado are kind of first out of the
28:49 gate and so a lot of other states are
28:52 looking at this legislation and so um
28:56 one of the Articles got picked up in in
28:59 Canada like so so I don't know how far
29:02 this is going to go
29:04 but there you have it
29:10 crazy I think Facebook AI generated
29:12 images or Watermark they are yep listen
29:16 I don't I don't mind I listen I don't
29:18 mind I don't mind marking stuff as AI
29:21 but like every single thing every single
29:25 thing so if I create you know um some
29:30 sort of software service that uses a
29:31 little bit of AI in it I've got to mark
29:34 that when do I need to mark it do I have
29:37 to lab and and do I have to label every
29:39 piece of content that's created as
29:43 synthetic well what if I use sentence
29:46 complete in Microsoft Word because
29:47 that's AI do I have to label that
29:50 sentence and then if I if I write the
29:52 second paragraph with AI but the third
29:55 paragraph with me but the fourth
29:57 paragraph I start in AI but I write half
29:59 of it do I have to label it on a perw
30:04 basis it like it's just it's just overly
30:07 Broad and it's got weird circular
30:10 definitions in it like it was like I it
30:13 just I it there were these self self
30:15 referential def definitions where it was
30:18 like it's AI if it generates something
30:21 and then if if it's AI wait it's AI if
30:25 it's generates something then if it
30:26 generates something it's AI it just
30:28 weird ass nonsensical sort of [ __ ]
30:32 and like really unclear they said
30:35 high-risk this this only applied to
30:38 high-risk AI systems but they decided
30:41 they they said
30:43 that a high-risk AI system is any AI
30:48 that could possibly generate a result
30:51 that was
30:54 biased so like you mean [ __ ] chat GPT
30:58 you mean any AI you could prompt any AI
31:02 to give you a biased answer right so so
31:05 just by their definition any AI is you
31:09 know falls under this any AI is
31:11 high-risk a AI so anyway it was just it
31:14 was just silly um but it was a cool
31:17 thing was as we were leaving we we just
31:19 sort of we we kind of tapped out at you
31:22 know I think there was still probably 45
31:25 minutes to go and we just tapped out and
31:26 as we were walking out um Senator
31:29 Rodriguez was just standing in the
31:31 hallway he was talking to these high
31:32 school kids that testified and so we
31:34 spent like 15 minutes talking with him
31:36 and he was really chill he was very
31:38 politician and and you know charming and
31:41 all that sort of stuff but you know he
31:44 was he was like listen it doesn't really
31:46 have any teeth in it you know it would
31:49 be the attorney general that have to
31:50 bring things we're not going to punish
31:52 anyone if they just make a
31:54 mistake it doesn't go into effect until
31:56 2025
32:00 but you know he was basically saying
32:02 like yeah I gotta we got to get we got
32:04 to get something we got to get we can't
32:05 do what we did with the Privacy stuff so
32:08 so so it's interesting it fascinating
32:10 fascinating stuff
32:12 so all right the media can tell lies
32:16 without
32:18 Source I used often twice really
32:23 often the entire internet is biased well
32:26 well so are people people biased that
32:29 was good Dr Kelly uh kibu I think is her
32:32 last name I I forget her last name um
32:34 out of Boulder she's a clinical
32:37 psychologist and she's she's creating a
32:39 system like she she's like spent her
32:43 entire career you know trying to
32:45 diminish bias right so so she's like
32:48 like yeah I care about bias I care about
32:51 it a lot but this billing you know ain't
32:53 doing it right what you're trying to do
32:56 here is is is is not doing what the
32:59 intent is and that's kind of part of the
33:02 the problem of the bill is
33:04 that it's a well it's a Well titled bill
33:08 right it's like the
33:11 AI consumer
33:13 protection against some you know syis
33:16 systemic bias or something like that
33:19 like like you're like well do you want
33:21 bias well no you don't let no
33:24 but this isn't this isn't BAS Bas on how
33:28 AI works and how developers develop with
33:32 it so but anyway you know it's it's uh
33:35 we'll see we'll see it was you know it's
33:37 it's very much early stages but it was
33:39 fascinating it was titled by AI it
33:42 probably was
33:49 actually so if they're concerned about
33:51 bias how are they planning to enforce
33:53 Google Etc well that's kind of what we
33:54 were saying is like one of the Eli wood
33:57 the one of the other guys that was in
33:59 our press conference one the metaphor he
34:01 said in his testimony he said he said
34:04 listen this is kind of like regulating
34:07 cars right you got Ford and you got
34:09 Chevy and you got Dodge and you regulate
34:12 cars at that level this legislation is
34:15 the equivalent of you know holding the
34:19 car dealers liable for for Ford making
34:24 shitty projects you know shitty products
34:27 because most of us are using um you know
34:30 commercial large language models that we
34:33 can't look under the hood we can't you
34:35 know we don't we don't know what's in
34:36 the training set we don't we don't have
34:38 any control over
34:39 that and my point was you know these
34:42 things are changing so fast and it's
34:44 like if you just look at the the Llama
34:46 models that came out in the past
34:48 week and and then you know Apple dropped
34:52 some really small models to be installed
34:54 on phones and it's there's just it's
34:59 B basically you know part of what the
35:01 legislation said is well you have to
35:03 hire lawyers to analyze your code
35:07 determine how much bias is in it and and
35:10 create an assessment that you then
35:12 report to the attorney or to the yeah to
35:15 the attorney
35:18 general
35:19 really if I'm making like a little
35:21 content generator for my little video
35:24 company got to go to the attorney
35:26 general with from a fancy ass ethics
35:30 lawyer that understands Ai and that can
35:33 analyze AI code and large language
35:35 models the [ __ ] scientists don't know
35:37 how these things [ __ ]
35:41 work so anyway I would like to
35:46 understand how that testimony group was
35:47 formed who spearheaded it an association
35:51 yeah it was a group
35:53 called it was rxm group
36:04 adding any controls would be inod
36:06 reducing bias yeah doing any fine tuning
36:08 ads
36:10 bias um wish these legislators would
36:12 actually use AI before writing a law
36:14 about it that was our basic all of us
36:16 what what all of us said is we want AI
36:20 we want or we want regulation we want
36:22 smart regulation like we want like like
36:25 what your intent here is good
36:28 but you have a whole community of people
36:31 that know about this [ __ ] so why not
36:34 involve them so that we're writing
36:36 legislation that acknowledges how it
36:38 works and and legislate at the right
36:40 level right and don't over legislate too
36:44 soon part of the part of the um part of
36:47 my testimony was um if legislation like
36:51 this had passed in the mid 90s when I
36:53 was doing the early early worldwide web
36:55 work it would have totally [ __ ] us up
36:58 and and so I said you know this is way
37:00 too early and it's way too fragile right
37:03 now like we're all moving fast trying to
37:05 figure out how to do this like you know
37:08 I didn't quite say we don't know what
37:09 we're doing but I I I pretty much said
37:12 you know everyone's trying to figure
37:13 this stuff out and I made a distinction
37:15 I said I said the AI that's been around
37:17 for 30 years is different than what
37:21 happened on November 30th 2022 the
37:23 launch of chat GPT made all of these
37:26 tools available to the rest of the world
37:29 and the rest of the business community
37:31 and we're all just trying to figure this
37:32 out and how it works and right and if
37:35 you put you know these sort of overly
37:38 strict burdens on small companies
37:41 they'll just go [ __ ] it they just won't
37:42 build anything or they'll leave the
37:44 state
37:46 so it was
37:48 cool I mean cool it was
37:58 [Music]
38:01 yeah exactly Common Sense reasonable
38:03 regulation yeah and it's like it's like
38:07 you know legislate the use of the
38:09 technology not the technology itself
38:12 right and this is this is one of the
38:14 things Sam Alman said he you know he
38:16 basically
38:19 said anyone that's
38:21 producing a to like a a Frontier Model
38:25 that's the equivalent of GPT 4
38:29 should be regulated and then he was
38:31 saying anything more powerful than that
38:33 like GPT 5 and higher what Alman was
38:36 saying was that [ __ ] should be licensed
38:38 so if you're producing models that are
38:41 strong enough that it can do like you
38:43 know AGI level stuff where it can do the
38:46 work of humans that should be licensed
38:48 just like you license planes and you
38:51 know radio wave you know frequency bands
38:56 and and [ __ ] like that like the the
38:58 really powerful stuff you license the
39:00 powerful Frontier models the big
39:02 Frontier models you regulate and then
39:04 let everyone who's using that [ __ ] do
39:07 what they do like Let There Be
39:09 Innovation down here but just
39:12 understanding that Nuance between you
39:15 know different levels of capabilities
39:18 and and the difference between you know
39:21 building a large language model that's
39:24 trained on 20,000 gpus over a 2-year
39:28 period with all of the data of the
39:30 internet that's a completely different
39:34 entity the little old Story vine who's
39:37 making a Content generation tool with
39:40 calls to open AI right like it's like
39:42 come
39:47 on and that's that's the problem when
39:49 they don't understand what it is they're
39:51 just like AI bad AI
39:55 scary label anything AI
39:58 don't push the button you can make
40:05 money and that's you know that's why
40:07 they're going to be able to pass [ __ ]
40:09 like this because you know people are
40:10 afraid of it they don't understand
40:12 it cuno sharing the live thank you so
40:15 much
40:17 cuno and Kyle should teach them well was
40:20 funny I I said to at some point I said
40:23 to Senator Rodriguez I'm like hey you
40:25 want AI lessons and he went
40:29 [Laughter]
40:33 he was like I think you've mistaken me
40:36 for someone who actually gives a [ __ ]
40:37 about
40:45 this but if they end if they do end up
40:48 putting some sort of AI commission
40:50 together like I'd love to be on that I'd
40:52 love to like you know go in and educate
40:55 some lawmakers and blow their minds and
40:58 you know show them what's coming because
41:01 I think what's coming is is profound and
41:03 I you know and I want legislators like
41:05 thinking about well what happens when a
41:08 G
41:09 gp5 Class large language model drops
41:13 that has agents and reasoning in it and
41:15 it can actually do people's jobs not
41:18 like augment their jobs but do their
41:21 jobs like yeah I want my legislators
41:24 having a [ __ ] clue about that
41:32 they
41:32 [Laughter]
41:32 [Music]
41:36 don't I mean honest to God the the looks
41:38 on some of the Senator's faces it's just
41:42 like it's like do we really need to deal
41:44 with
41:51 this all right so that was that was Mr
41:55 Smith Goes to Capitol Hill for for you
41:59 lunar stick I'm working on it oh good
42:01 yeah lunar stick lunar stick I I honest
42:04 to God how you did this so so lunatick
42:08 was Daisy Thomas was
42:10 a a uh a political figure in Utah I
42:14 think
42:15 right
42:17 um and she's good at this stuff she
42:20 knows this stuff well it is it is crazy
42:24 it's just it I'll tell you this was
42:26 another FAS ating thing about about the
42:29 day so our little group was a bunch of
42:33 you know people that were part of the AI
42:35 salon and the Rocky Mountain in uh you
42:38 know interest group and
42:42 um oh hang on
42:45 champ champ what do you want you want is
42:48 Mama going to give you cheese is that
42:49 what's Happening Here
42:52 Come m going to give you
42:54 cheese you go to Mama
42:57 he want
43:07 cheese so our group we had I don't know
43:10 we probably had in person I think we had
43:12 about a dozen people
43:15 testifying and most of our people were
43:18 just individual business owners right
43:19 either just individuals who were
43:21 concerned or individual business owners
43:23 and then
43:25 um but there were also also like like
43:29 the AFL uh CI wait the the the ALU was
43:34 there the Colorado Tech Association was
43:37 there so a bun there were a bunch of
43:39 interest groups and and lobbying groups
43:42 that that were also there and what was
43:45 fascinating was the
43:47 individuals did their passion
43:51 testimony but it was just of a
43:53 completely different level like the
43:54 Professionals in the room they knew
43:57 their [ __ ] the Professionals in the room
43:59 were like were like okay here's here's
44:02 the specific ways that these three
44:04 definitions are too broad here are the
44:06 here here are the the two
44:09 self-referential um you know Clauses
44:11 within the bill you know you've got to
44:12 fix this you've got to fix that here's
44:14 why this won't work here's the
44:16 implications of this like the
44:17 professionals knew their [ __ ] was really
44:20 interesting hey Kyle have you ever
44:22 caught the weekly lives that the
44:24 ideogram CEO does he did one today I
44:26 have not caught those Gavin um that's
44:29 interesting post those to I think you're
44:32 in the uh you're in the salon right post
44:34 those wait it's crazy that I love it so
44:36 much no matter how much I hate it too
44:39 that's lunatick talking about the
44:40 politics stuff I could you know you know
44:43 what's funny lunatick at the at the end
44:46 of the at the end of the night or at the
44:48 end of my night before when I left there
44:50 were these there were these high school
44:51 this High School AI Club was there and
44:54 they were actually um um testifying in
44:57 favor of it these high school kids were
44:59 they were [ __ ] articulate man it was
45:02 it was good the the senators were like
45:04 we're impressed with you it was it was
45:06 neat
45:08 um but
45:10 um what I realized is in my lifetime
45:14 like in my childhood and in high school
45:16 and in college like I never did the
45:19 field trip to my state capital right I
45:23 never sat in the room and watched this
45:25 [ __ ] happen as a as a young man so I
45:27 felt I felt like oh this is probably
45:30 what you do on High School field trips
45:32 you know when the civics class takes the
45:35 uh the field trip to the state capital
45:37 this must be what they do I felt very
45:39 much like a like a student just like oh
45:44 look they got big
45:48 chairs that was
45:50 [Laughter]
45:51 crazy oh man so anyway Gavin yeah post
45:55 post the link to that this the ideogram
45:58 CEO that would be interesting to
46:00 see yeah Mama give you cheese AFL CIO
46:04 yeah they were there who else was
46:07 there like just different employment
46:10 groups and you know people who have
46:13 interest like you know it's politics
46:15 right all these people have interest
46:16 they go come in they say their peace and
46:19 then the politicians do what they
46:21 do you've been heard you've been heard
46:24 here was the hearing you could show up
46:26 you could partic
46:28 participate so but I've never ever ever
46:32 in my life had any interest in
46:35 participating in uh the Civic
46:39 discourse so so so I've now done that
46:43 I've checked that little box for myself
46:45 it was not on the bucket list but it has
46:48 been
46:49 [Laughter]
46:52 checked oh man all right let's go let's
46:55 go look at some stuff here I want to go
46:57 find this Runway
47:00 thing
47:03 Runway any other thoughts or questions
47:05 on that ship before I move
47:08 on check the never to
47:11 return the salon. a yep I'll get to
47:16 that I was on a field trip it's
47:22 [Music]
47:24 true I didn't either I always wanted it
47:27 could couldn't afford the eighth grade
47:28 DC trip there you go yeah that was the
47:31 difference between you and me Daisy is
47:33 uh you wanted to go on the DC trip I was
47:36 like no I want to go on like the like
47:38 the garbage sorting the garbage sorting
47:40 Factory was more interesting to me than
47:46 politics I I I remember going to a uh a
47:51 chicken an an Egg
47:54 Factory did you know they had egg
47:56 factories
47:59 I went to the Penfield in
48:02 Pennsylvania the
48:04 Penfield Egg Factory I think was
48:06 somewhere in Lancaster County we took a
48:09 bus over there and we go in and they got
48:11 chickens everywhere and all these eggs
48:12 are rolling down and they got you know
48:14 people were like holding the eggs up to
48:16 lights seeing if they were I guess
48:17 fertilized or whatever um it was
48:22 crazy so I know more about egg factories
48:24 than I do about politics
48:30 um anyone from CU or enar enar sounds
48:33 familiar
48:35 pate and um Dan Murray Dan Murray from
48:39 the Rocky Mountain interest group they
48:41 meet at CU so so the the Rocky Mountain
48:44 AI interest group they they had a lot of
48:46 representation from Boulder and from CU
48:50 and and from the from the tech from The
48:52 Tech Community up there so that that was
48:55 decently well represented cuz I someone
48:57 mentioned enar I don't know if someone
48:59 was actually testifying from there but I
49:01 know I heard it
49:09 [Music]
49:18 mentioned okay let me go look at this uh
49:21 Runway
49:22 [ __ ] Runway
49:32 this is
49:34 it oh
49:37 wait oh maybe it wasn't
49:40 new
49:49 H I don't think it was
49:55 new now this is December 11th I just saw
49:58 this video for the first time all right
50:00 that wasn't as exciting as I thought it
50:03 was I mean we can look at the video
50:05 because I think it's
50:11 interesting what if the next Frontier in
50:14 the AI Revolution the models that will
50:16 truly usher in a new era of creativity
50:19 were AI models that think a lot more
50:21 like Reuben my dog
50:32 llms or large language models are
50:35 everywhere they can generate poems
50:38 essays or even movies they can help
50:41 brainstorm new ideas or find new
50:43 material in a nutshell these models work
50:46 by predicting the next token in a
50:48 sequence like words in a sentence so it
50:51 makes sense that the more sentences it
50:52 knows the better it'll be at generating
50:54 new
50:55 ones as they get better and better at
50:58 predicting they gain more understanding
51:00 of the world and the patterns in it
51:03 these large language models introduced
51:05 an entirely new paradigm to AI big
51:08 models plus Big Data equals greater
51:11 knowledge of the
51:12 world the thing is llms only know about
51:16 language what if we applied the same Big
51:19 Data big model approach to how we
51:22 generate videos well that's what we're
51:25 calling General world models
51:29 gwm's and we believe that they're the
51:31 next big
51:32 [Music]
51:34 thing similar to llms General World
51:37 models are given a large amount of data
51:40 but not just text it's also given videos
51:43 images and audio everything you'd need
51:46 in order to understand how the world
51:47 works and with all this information the
51:51 models build out sort of a mental map
51:52 for
51:54 themselves and this is where my dog
51:56 comes in hi
51:58 reuben reuben also has an internal model
52:01 of the world based on things he knows
52:03 and it looks something like this he
52:05 learned that if we go this way there's a
52:07 higher chance of us going to the park
52:10 and if we go that way there's a dog that
52:12 kind of looks at him
52:13 funny he knows that there's usually
52:15 sidewalk chicken scraps right here and
52:18 that these stores known as pet stores
52:21 that have the treats that he likes once
52:23 we get to the park this is where all the
52:25 butt sniffing happens
52:27 with all of this data the sights sounds
52:30 relationships of things Reuben has
52:32 figured out how to predict certain
52:34 outcomes and adjust his behaviors just
52:37 like the general World
52:40 model but the most incredible thing
52:43 about all this is this part these models
52:46 have the ability to generalize its
52:48 understanding to new and unseen data it
52:51 knows how to imagine the future based on
52:54 its knowledge of the world it's kind of
52:56 like how Reuben knows to avoid other
52:58 dogs that also look at him funny and how
53:01 he knows to drag us into pet stores that
53:03 he's never been to We Believe That by
53:06 trading these models to predict the next
53:08 frame or token in a sequence the models
53:10 will learn a much more detailed
53:12 understanding of the world including the
53:14 wise and the house than the large
53:17 language models so what does this all
53:20 mean it means that pretty soon General
53:23 World models will allow us to simulate
53:26 worlds that more closely reflect our
53:29 own or the ones Ruben built up in his
53:32 little
53:33 head so yes the next Frontier of AI will
53:37 act a lot more like this one right you
53:42 know nice piece so I I thought I saw
53:45 this for the first time today so but
53:47 this was from back in December so so not
53:50 super new but really good explanation of
53:54 large language models really good
53:56 explanation of of what they're calling
53:58 the what is it large World models
54:01 um conceptually this is very similar to
54:05 what open AI announced with Sora their
54:08 video model right that it actually
54:11 understands the world and so I just
54:13 thought this was this was an interesting
54:15 thing to see today to to see that you
54:18 know that they're they're thinking the
54:20 same way and and you know I I think
54:23 that if we think of GPT 5 just as as
54:27 it's going to be like
54:29 gp4 I I think these models are going to
54:31 get quite sophisticated quite quickly
54:33 where the the multimodal stuff where its
54:37 ability to see and hear and make audio
54:41 and make video and make pictures is
54:44 going to get a lot more
54:45 seamless and so the sophistication might
54:48 not just be on a on a language level
54:50 where it you know hallucinates less it
54:52 might be like it understands the world
54:54 better and better and better meh I know
54:57 I know Pate this is very this is very
55:00 much this is very much a marketing video
55:02 I get it cuz the other thing is if that
55:05 if you know if this is four almost five
55:07 months old um then you know like where's
55:10 the show us show us what you've built
55:13 you know sitting on top of this so but
55:16 we'll see you know maybe by the end of
55:18 the year Runway puts something out
55:19 that's really really something
55:22 good um okay it looks like I was
55:24 buffering there for a second someone
55:26 else said said something that I wanted
55:27 to do excellent explanation oh K9
55:31 Minority Report that was pretty funny I
55:33 got a notice from open AI that it can
55:36 share information between sessions yeah
55:38 I I saw a couple of posts on that as
55:41 well let's let me reload here and see if
55:44 it actually is
55:47 true all right there's 3.5 there's
55:52 four why is this not loading
55:58 ah yes all right we're learning this in
56:02 real time
56:04 people chat GPT now has memory across
56:09 chats
56:11 okay this is a big deal remember back in
56:15 the day when we were excited chat TPT 4
56:17 went multimodal yeah like four months
56:20 ago yeah I know okay keep the
56:23 conversation going chat GPT will carry
56:25 what it learns between chat
56:27 allowing it to provide more relevant
56:29 responses more helpful over time chat
56:32 GPT will become more helpful as you chat
56:35 continually remembering details and
56:37 reference or preferences manage what it
56:39 remembers you're in control review and
56:42 delete specific memories restart from
56:45 scratch or turn the feature off in
56:48 settings all right so let's go to
56:54 settings security wait
56:59 General
57:03 personalization oh something's loading
57:07 here the new version of idiagram will be
57:11 released I think if I remember correctly
57:13 in two weeks oh
57:17 cool custom instructions off failed to
57:20 get your
57:23 settings oh yeah they lost my [ __ ]
57:28 they lost my [ __ ] custom instructions
57:31 y mother you're son of a let me reload
57:38 here yeah this is brand new Hot Off
57:41 depresses from open AI
57:43 people yeah they're
57:46 having this is okay typical of when chat
57:50 GPT doesn't update the site's going to
57:52 suck for a couple of
57:55 days okay
57:58 settings
58:00 personalization custom instructions okay
58:03 it remembered them great
58:06 okay
58:09 memory chat GPT will become more helpful
58:12 as you chat picking up on details and
58:15 preferences to tailor its responses to
58:18 you learn more let's I I am not a manual
58:21 reader but this is a nuanced enough
58:23 feature that I actually want to read
58:24 about it
58:28 we're rolling out a small portion of
58:30 chat GPT free and plus users this week
58:35 to learn how useful it is we will share
58:37 plans for broader roll out soon so if
58:40 you don't have this that's cuz you don't
58:42 have it but go check
58:47 this Joker had it
58:50 yesterday and it's gone wait this was
58:53 updated a week ago what was from
58:55 February
58:58 I don't know what you're talking about
59:03 jocker everyone save your custom
59:09 instructions P can't even load the
59:12 settings super cool yeah no they're they
59:14 got their it's you know they're having
59:17 server upgrade
59:19 specialness we can tell it to remember
59:21 that it has internet access oh yeah
59:24 that's actually something that I wanted
59:25 to check is is um on my app when I asked
59:31 it to check the internet it brought back
59:32 browsing the internet that remember they
59:36 made it go away let's see um
59:38 search the
59:40 internet for the
59:43 latest nuggets
59:49 news no it's still not saying it all
59:53 right that's shitty I don't like that
59:55 they're
59:58 they're not
1:00:00 disclosing difficulty accessing the
1:00:02 current
1:00:04 news what is the
1:00:09 weather in Denver tomorrow
1:00:28 all right well so it appears to be
1:00:31 connected to the internet all right
1:00:33 let's go back to this here little L
1:00:35 little thing chat gbd can carry what it
1:00:38 learns between chats allowing you to
1:00:40 provide more relevant responses as you
1:00:42 chat with it it becomes more helpful
1:00:44 remembering details and preferences from
1:00:46 your conversations its memory will get
1:00:48 better the more you use
1:00:50 it I I'm curious if you can go back if
1:00:54 it remembers [ __ ] from like the past
1:00:56 [ __ ] in year and a half I've been using
1:00:57 you stupid
1:00:59 tool um you can teach it to remember
1:01:03 something new by chatting with it for
1:01:05 example remember that I'm a vegetarian
1:01:09 when you recommend a recipe oh that's
1:01:11 fascinating okay that's good it's good
1:01:13 we have examples
1:01:17 here it has no memory wait what are you
1:01:19 saying over a week ago that's the same
1:01:21 article from oh when they first had the
1:01:24 limited roll out
1:01:27 can you tell it to change songs I got it
1:01:32 yesterday all
1:01:35 right it can't remember kenet I don't
1:01:38 know you can control chat gbt's memory
1:01:42 you can reset it delete it specific or
1:01:44 all memories or use temporary chat so
1:01:48 that's interesting temporary chat so how
1:01:50 do we do a temporary
1:01:53 chat do I have to go into settings each
1:01:55 time
1:02:02 manage any
1:02:07 chat I just solved my puzzle
1:02:10 piece any chat with chat GPT wait as you
1:02:14 chat with chat GTP the details and
1:02:17 preferences it remembers will be shown
1:02:19 here
1:02:33 where did I leave off in my last
1:02:36 remember that I like concise
1:02:39 responses I wonder can I what what about
1:02:43 all my [ __ ] history
1:02:58 memory will be available when we roll it
1:03:01 out oh memory for gpts will be available
1:03:05 when we roll it
1:03:10 out custom instructions continue to
1:03:13 allow you to provide chat GPT with
1:03:15 direct
1:03:16 guidance does it remember sensitive
1:03:25 information all right
1:03:30 um I don't think I'm I'm not going to
1:03:32 play with it now just
1:03:35 because like this feels like kind of
1:03:38 thing I need to I need to play
1:03:40 with a bit more to to understand how it
1:03:43 works and to come up with examples so
1:03:45 I'm not going to I'm not going to do
1:03:46 that now I think that'll be boring as
1:03:48 [ __ ] tell it to remember your dog's name
1:03:51 log out and log back in oh that's good
1:03:53 okay so let me do a new chat
1:03:59 my
1:04:04 dogs dog's name is
1:04:09 Champ and he
1:04:13 sings whenever I play the guitar
1:04:21 period I'd like you to remember that
1:04:28 memory
1:04:32 updated got it I've noted it okay so I'm
1:04:36 going to do a new CH a new and I'll even
1:04:39 reload the page so we're we're
1:04:46 definitely what's my dog's name
1:04:56 [Laughter]
1:04:59 unable to load
1:05:01 history H something seems to have gone
1:05:04 wrong yeah no
1:05:07 [ __ ] try it
1:05:10 again chat gbd4 yeah this is [ __ ]
1:05:14 clunky as [ __ ] the their servers
1:05:17 are
1:05:20 buried or my browser is
1:05:23 crashing could be either
1:05:31 someone said before I hung so janky
1:05:34 updates you froze
1:05:37 it unable to load history yeah yeah
1:05:40 there they've got a mess going on right
1:05:44 now this this isn't going to be any fun
1:05:46 to play with oh oh this is cool look so
1:05:49 you can do a temporary chat so you you
1:05:52 pop
1:05:52 it oh that's very cool so a temporary
1:05:56 chat it won't go into your memory so so
1:05:58 if you want to you
1:06:01 know have a thing that won't go in there
1:06:03 you do it as a temporary chat and I
1:06:05 guess it'll do it on whichever model you
1:06:07 select but okay so now we're in new so
1:06:09 now I'll
1:06:11 say what's by my dog's name and what's
1:06:15 his special skill
1:06:29 the [ __ ] ain't working it's janky ass
1:06:33 [ __ ] the [ __ ] ain't working it's a poo
1:06:36 poo poo
1:06:42 poo who who who's the person that you'd
1:06:45 like to be on the government's AI task
1:06:48 force oh it's this
1:06:51 guy janky piece of [ __ ] this is all poo
1:06:54 poo
1:07:06 oh good lord people all right we're
1:07:09 going to go play with LTX Studio because
1:07:11 it's a [ __ ] ton more fun than this
1:07:13 stupid piece of
1:07:19 [ __ ] well in theory cat GPT now knows
1:07:23 Champ's name champy was the first first
1:07:26 thing that it learned all right so we're
1:07:29 going to
1:07:30 go to LTX studios Okay so if
1:07:38 you all right let me give you a little
1:07:41 bit of background on
1:07:44 LTX LTX Studio I don't think they're
1:07:47 more than like um 6 months old oh by the
1:07:50 way Pate well all of you but but Pate in
1:07:53 particular because Pate kind of called
1:07:55 this
1:07:56 so remember when Devon came
1:07:59 out and it was this agent that looked
1:08:02 like it was going to do all the stuff
1:08:03 and Pate was like I'll believe it when I
1:08:05 see it and then it turned out to be that
1:08:07 was he he made the right call that that
1:08:10 it it it wasn't all that real well
1:08:12 they're only a six-month old company but
1:08:15 apparently uh they just got invested in
1:08:18 for like14 million at A2 billion
1:08:21 valuation it's a$2 billion valuation for
1:08:25 for writing an a agent that chat GPT is
1:08:27 probably just going to be able to do 3
1:08:29 months from now
1:08:32 um anyway so Chachi or uh LTX
1:08:38 studio um was announced I don't know two
1:08:42 months ago this is a similar sort of
1:08:44 thing I think it's a very young company
1:08:46 uh it's a couple of Israeli
1:08:48 Founders
1:08:50 um making a lot of noise in you know
1:08:54 Hollywood circles because you B
1:08:55 basically put in a prompt and it
1:08:58 generates
1:09:00 [Music]
1:09:17 everything and then they they did this
1:09:20 whole oh they're doing a they're doing
1:09:22 an open call
1:09:24 $150,000 production thing um they did a
1:09:28 whole thing on some of the features that
1:09:30 it's got like when they were talking
1:09:32 about launching it
1:09:36 where it does consistent characters it
1:09:40 does scenes it does background music it
1:09:43 does effects it does titles it does all
1:09:45 this sort of stuff Yong ma what's
1:09:48 happening is I got access to LTX studio
1:09:51 so anyway it's pretty cool um if you
1:09:54 want to get access to it it is a pain in
1:09:57 the ass so here's what you got to do
1:09:59 you've got to join their
1:10:03 Discord so here I am on Discord LTX
1:10:07 Discord and then you basically just have
1:10:11 to go to announcements and then they
1:10:14 they basically do this they do batches
1:10:16 of like a hundred people they put out
1:10:19 like a Google form that you fill out and
1:10:22 they they basically just drop these at
1:10:24 some point in the morning and the form
1:10:26 is completely filled up within 10
1:10:28 minutes so I missed this for like 2
1:10:31 weeks and 2 days ago or was it
1:10:36 yesterday yeah it was yeah it was
1:10:38 yesterday it was it was when I
1:10:40 testified yesterday morning I missed the
1:10:44 first
1:10:45 three so so they did one at 700 a.m.
1:10:48 well no they did one at 5 a.m. and it
1:10:51 closed they they did it at 5:05 it
1:10:54 closed at 5:15 then they did one at
1:10:57 7:09 and it closed
1:11:01 at I don't know when and then they did
1:11:05 one at 9:15 and I got in on that one at
1:11:07 like 9:17 I looked up and it was live
1:11:10 and I was like [ __ ] and I jumped on it
1:11:12 and I filled out the form and I got it
1:11:14 so if if you want access to this thing
1:11:17 this is very much like uh if you've ever
1:11:20 [ __ ] around with nfts and you're
1:11:22 hanging out in Discord with the Discord
1:11:24 Bros or with the nft
1:11:26 Bros it's [ __ ] exhausting but anyway
1:11:29 I did that because a I wanted it but I
1:11:32 also wanted to be able to show all of
1:11:33 you what it's about so go do that all
1:11:36 right and then here it
1:11:37 is so welcome to LTX Studios join our
1:11:41 Discord their [ __ ] and then basically
1:11:43 type in a film idea or a full
1:11:45 synopsis um I did one for so if yall
1:11:50 remember
1:11:58 [Music]
1:12:01 I made this guy named KS
1:12:06 Dylan this fake dude I I am I am very
1:12:09 enamored of late with with doing
1:12:17 um with doing kind of documentary
1:12:21 storytelling of people that didn't
1:12:22 actually exist I don't know what it is
1:12:25 it f Fates
1:12:27 me the
1:12:32 [Music]
1:12:38 singularity well that's the the beat
1:12:41 poet version of me so I've been doing
1:12:43 these fake characters called k k period
1:12:46 S period and then some famous person's
1:12:49 last name so here's KS Dylan in
1:12:54 um in Brazil
1:13:00 thece
1:13:04 Shadows she was the
1:13:09 arti
1:13:11 oops what
1:13:14 [Music]
1:13:18 happened
1:13:21 deep yeah that was you Kelly Kelly camp
1:13:26 in her ABS a harvest I could
1:13:32 not
1:13:36 if and you'll see here in a second he
1:13:39 sing he sings in
1:13:42 Portuguese
1:13:45 never if I
1:13:47 never words wouldn't be
1:13:51 lost written in measured the
1:13:55 [Music]
1:14:02 [Music]
1:14:07 I thought that was when you were young
1:14:09 do I really look that old it's [ __ ] up
1:14:12 if that's when I was young then I gota I
1:14:15 got to take more
1:14:18 vitamins so
1:14:20 okay so then what I put I I just put in
1:14:23 a a prompt that was like
1:14:27 um New York City Folk musician KS
1:14:32 Dylan um in Brazil like you know meets
1:14:37 someone and like questions whether he
1:14:39 moves back to New York City or something
1:14:41 like that so it
1:14:43 generated
1:14:47 this and so and so what it
1:14:52 is see if I can get it small enough
1:14:56 so what it generated was all these
1:14:58 different scenes and so there's scene
1:15:00 one scene two and then within the scenes
1:15:03 there
1:15:04 shots so and then each each shot has a
1:15:07 description about it the the shot type
1:15:10 like is it a closeup is it a you know a
1:15:14 wide angle things like
1:15:16 that um and and let me let me zoom in a
1:15:21 little bit so you can see them so it's
1:15:24 like you know
1:15:27 here's the here's the shot and then you
1:15:29 can go to the shot
1:15:31 editor and that will show you the prompt
1:15:34 so here's the actual prompt for that
1:15:36 editor and then the shot type is a is a
1:15:39 wide shot but they have all your
1:15:40 different camera angles here and then
1:15:43 scale and you know motion and so so
1:15:46 there's all sorts of presets for for
1:15:49 basically a prompt to an image generator
1:15:52 that that is I'm guessing this is stable
1:15:54 diffusion and it's some shitty version
1:15:57 of stable diffusion it's not super well
1:15:59 trained um and then they they're
1:16:01 probably using stable video diffusion
1:16:03 for the video and you can say make this
1:16:05 3 seconds or 4 seconds or 5 seconds so
1:16:07 every shot you've
1:16:10 got um you know that kind of
1:16:13 detail and and uh you can also put in
1:16:17 sound effects for each shot so I could
1:16:20 put
1:16:21 Street Sounds
1:16:26 and it should find me
1:16:32 some New York City traffic City Street
1:16:35 downtown traffic pedestrians walking
1:16:39 by Europe Cafe streets signaling cars
1:16:43 pass by
1:16:45 [Applause]
1:17:11 anyway we don't have to do all this now
1:17:12 okay so let me just keep telling you
1:17:15 okay
1:17:16 so then you've got so the
1:17:20 scene so here's your overall stuff about
1:17:23 the scene and then you've got a voice
1:17:25 over here and if you click on the voice
1:17:26 over here's the script for the voiceover
1:17:29 so it writes the
1:17:31 script um I suppose you could leave that
1:17:34 blank if you wanted to so you just
1:17:36 basically give it a prompt and then it
1:17:38 gives you all this [ __ ] so you can give
1:17:40 it a fairly detailed synopsis from what
1:17:42 I can understand you cannot I asked them
1:17:45 if I could upload one of my screen plays
1:17:47 and would it just figure that [ __ ] out
1:17:49 and no it can't which program creates
1:17:51 the most realistic images that would be
1:17:53 mid journey by far
1:17:56 um especially if you want photographic
1:17:58 um but ideogram is really really good
1:18:01 right now as is um if you're really
1:18:04 geeky and you want to install software
1:18:06 and don't mind doing that [ __ ] stable
1:18:08 diffusion is a whole ecosystem where you
1:18:10 can do all sorts of crazy stuff um if
1:18:13 you if you like uh lots of switches and
1:18:15 dials for control then use leonardo.da
1:18:27 and then Dolly within um within chat GPT
1:18:31 is really good too or inside uh
1:18:33 Microsoft co-pilot okay
1:18:36 anyway so and then the other thing it
1:18:39 does is it makes characters so here's KS
1:18:44 Dylan and you can do a face swap face
1:18:47 switch so I uploaded a version of my
1:18:52 which one did I use I did oh I did the
1:18:54 cowb boy I did the the sexy cowboy dude
1:18:58 the sexy AI version of me so it looks
1:19:01 sort of like me and it looks sort of
1:19:03 like the the ks Dylan we had before but
1:19:06 you know it's got KS Dylan as a
1:19:08 character and then it's got Ana silver
1:19:11 I've got to edit her because I put an n
1:19:14 in her name so let's change her n to an
1:19:17 N option NN gives you an Ana Sil or
1:19:22 Silva all right so there's Ana Silva
1:19:27 and then wild the Silva is the is the
1:19:30 wise old musician that I guess I guess
1:19:33 is the uncle of Ana now that I think
1:19:36 about it all right so it makes these
1:19:39 characters and then within the script
1:19:42 you can call a character KS Dylan with a
1:19:45 mix of awe and
1:19:48 curiosity takes in the colorful sights
1:19:50 right so so it knows that character so
1:19:53 it puts that character in that scene you
1:19:56 know with all the shot stuff and then
1:19:58 you can preview it and it'll Stitch all
1:20:00 these shots
1:20:03 together and is this ready to go oh it's
1:20:06 building
1:20:13 it and I haven't tried doing a full
1:20:15 render yet cuz I was still working on
1:20:17 the
1:20:18 script I came to Rio in 1972 on a lark I
1:20:22 figured i' stay for a week or two but
1:20:24 she had other plans who was she her name
1:20:28 was Ana and she stung me we didn't talk
1:20:30 much didn't need to the vibrant streets
1:20:33 of Rio became our stage and our Melodies
1:20:36 echoed through the city as our love grew
1:20:38 I faced a choice returned to the life I
1:20:41 knew or Embrace this new rhythm in the
1:20:43 end I found my home in the pulsing
1:20:45 heartbeat of Brazil where our music and
1:20:47 wait wait wait wait look at the arm look
1:20:49 look at the arm wait this is so
1:20:53 good it's coming off for her right
1:20:55 shoulder and that goes around his
1:21:01 neck oh that's good ana ana is
1:21:05 anatomically
1:21:07 challenged she is beautiful but her arms
1:21:10 they were like
1:21:14 pretzels she seemed to grab me from
1:21:16 everywhere I never knew where her hands
1:21:18 were coming from I didn't know how many
1:21:21 many fingers she had
1:21:25 so it is it is a very janky
1:21:28 model
1:21:30 um okay oh [ __ ] I just closed it yeah
1:21:34 dang it uh uh uh yes I'll accept your
1:21:37 cookies your cookies are
1:21:40 fine your cookies are tasty tasty little
1:21:45 cookies
1:21:48 fantastic um but like the script it
1:21:51 wrote was pretty shitty and I I was
1:21:53 working on the script and then I I
1:21:55 changed the order of scenes and you can
1:21:58 take things out of scenes if you
1:22:04 want so like you know and this this one
1:22:08 like bustling Rio
1:22:12 streets that's not a bustling Rio Street
1:22:15 is
1:22:17 it oh yeah so so check this out this is
1:22:20 kind of cool so if I go to the shot
1:22:23 editor here's the thing I
1:22:26 discovered um let make this
1:22:30 bigger frame text
1:22:34 image oh cool okay so I can probably I
1:22:38 can I can drag an image in here so I
1:22:41 could go make something in mid journey
1:22:43 and bring it in here that's good to
1:22:46 know and then text what's this Frame
1:22:51 edit generative fill remove objects
1:22:54 camera angle
1:22:56 but there's there's the seed here so
1:22:58 like that doesn't look like a bustling
1:23:00 Rio Street to me so you can just hit the
1:23:03 the re reload the
1:23:07 seed and it'll change it up and so I've
1:23:10 done this a couple of times like that
1:23:11 looks better that looks more like
1:23:19 Rio we definitely want
1:23:22 streets that's kind of cool
1:23:25 but I like the one with the people in it
1:23:27 more people in
1:23:29 it there you go that's cool that looks
1:23:31 kind of like 1972 Rio
1:23:36 right all right cool so so like that's
1:23:38 that's already better and then I can go
1:23:40 back to the
1:23:42 storyboard and so that's that's that
1:23:44 there's him let's see let's go that
1:23:47 one's okay but let's do this as a
1:23:49 instead of a closeup let's
1:23:52 do from above
1:23:58 closeup of a street
1:24:00 musician and again the hands pretzel
1:24:03 hands weird weird
1:24:11 [Laughter]
1:24:13 [Music]
1:24:15 [ __ ] the images are so bad there there
1:24:19 he is he's he's got two fingers up here
1:24:22 on the fingerboard and two fingers is
1:24:25 here he's got he's got to
1:24:27 [Laughter]
1:24:36 really all right you know what we're
1:24:37 gonna do no Kyle what are we gonna do
1:24:39 we're gonna go over to Mid journey is
1:24:41 what we're going to
1:24:46 do
1:24:48 mid
1:24:51 joury and we're going to go
1:24:55 we're going to go create and I'm going
1:24:57 to go
1:24:58 [Music]
1:25:00 um Koda
1:25:06 Chrome
1:25:08 photo
1:25:11 from
1:25:13 1972
1:25:15 of a
1:25:19 street musician in
1:25:23 Rio from a
1:25:26 above and then I'm going to
1:25:30 go took weirdness down but stylization
1:25:33 I'll do decent and we're going to do 9
1:25:36 by 16 because that's what aspect ratio
1:25:38 this thing shot
1:25:40 in all right and then let's see what we
1:25:45 get
1:25:56 wow science fiction would work both
1:25:58 would be good to use too
1:26:02 L I tried feeding it the lyrics of a
1:26:05 song 10 quatrains it got lost maybe my
1:26:09 song yeah I again this is janky is as
1:26:14 janky does that's exactly right why is
1:26:16 this why are these not
1:26:20 photographic but they're already better
1:26:22 aren't they that one's kind of cool
1:26:28 says that
1:26:31 one but it looks like
1:26:34 trash that doesn't so let's
1:26:38 see
1:26:43 [Music]
1:26:46 um let's see let's see
1:26:50 uh
1:26:53 um Pro
1:26:59 professional
1:27:02 documentary
1:27:10 photograph
1:27:12 on Koda Chrome
1:27:16 film from 1972 of a street musician in
1:27:22 Rio shot
1:27:30 From A
1:27:32 Balcony
1:27:35 above the
1:27:40 street and then I'll put
1:27:45 um people are happy and
1:27:51 dancing all right let's try that
1:28:05 [Music]
1:28:10 yeah I know I could use chat GPT to do
1:28:12 the prompts I you
1:28:19 know we're in a we're we're in a weird
1:28:23 we're in a weird place right right now
1:28:29 where like
1:28:32 there's as these tools get better
1:28:35 there's part of what they're going to do
1:28:37 where it's going to it's going to
1:28:38 offload a bunch of work for us but then
1:28:41 I think stuff like this it's going to
1:28:43 make new things possible where like
1:28:46 instead of saying like oh it'd be cool
1:28:48 to make a YouTube video sometime we're
1:28:51 now going to be saying things like oh
1:28:52 you know what maybe I'll make a feature
1:28:54 film this
1:28:56 weekend so like I just I think that our
1:28:59 brains are going to be like electrified
1:29:01 because it's just things are going to
1:29:03 get crazier and crazier and crazier
1:29:06 these do not look like photographs to me
1:29:08 what the [ __ ] is going
1:29:14 on that's kind of cool but it's it's not
1:29:17 really
1:29:21 right all right let's go to Chad GPT
1:29:25 I'm going to go temporary
1:29:33 chat oh look temporary chat won't be
1:29:36 used in model training the chat won't
1:29:38 appear in history use memory or be used
1:29:40 to train our
1:29:42 models for safety purposes we may keep a
1:29:45 copy for up to 30 days for safety
1:29:47 purposes
1:29:49 yeah don't trust any of these guys
1:29:56 write me an image
1:30:00 prompt that can be used to generate an
1:30:03 image in mid
1:30:07 Journey that
1:30:10 describes a photograph of a street
1:30:16 musician in Rio in
1:30:22 1972 shot from above with with people
1:30:24 dancing around
1:30:28 him
1:30:30 period also include a hat on the street
1:30:34 to collect
1:30:40 [Music]
1:30:52 money that's a lot of [ __ ] prompt
1:30:54 chat
1:30:59 GPT all right fine fine do that go make
1:31:07 that can you try ideogram yeah let's go
1:31:10 try ideogram as well all right um
1:31:17 idog
1:31:19 a if you haven't tried ideogram you
1:31:21 should try ideogram it's quite good
1:31:25 I don't I think it's going to eat that
1:31:27 oh no it took it took that whole thing
1:31:29 okay so we're going to go N 16
1:31:32 by9
1:31:37 yeah um we're going to do
1:31:41 photo and
1:31:44 cinematic and then
1:31:48 generate I had 45 of my students
1:31:51 download chat GPT this week for an
1:31:53 assignment oh that's awesome that's
1:31:55 super cool Allan very nice love
1:31:58 it that's not bad there that image all
1:32:01 right let's see what mid journey is
1:32:06 doing 10% complete what's going on Mid
1:32:14 Journey looks like it
1:32:16 hung reload
1:32:23 reload oh it ate that prompt
1:32:26 interesting guess it didn't like me
1:32:36 leaving oh no it's still going all
1:32:40 right see what idrs got going
1:32:56 that one labeled
1:33:07 it look the floating
1:33:09 [Laughter]
1:33:12 hats this stuff's all so [ __ ]
1:33:17 janky here we go this is that's more
1:33:19 what I was thinking all right
1:33:26 that one's a bad
1:33:31 guitar that one's kind of
1:33:33 cool let's do we'll do very strong on
1:33:36 that one we'll make a whole pile of
1:33:39 variations of this one we'll just grab
1:33:42 one of
1:33:43 those and we'll go back over to our
1:33:45 happy little our happy little but you
1:33:48 can see where this is so what's cool
1:33:50 about LTX studio is you can take a basic
1:33:53 idea it'll it'll it'll split it into all
1:33:56 this stuff and then basically you've got
1:34:00 an image and video generation project
1:34:03 for every shot in the film which is what
1:34:05 you would do anyway right but this just
1:34:08 makes it much more apparent you know
1:34:10 that you know telling linear narrative
1:34:13 stories is it's a lot of work it's tough
1:34:17 all right
1:34:33 let's grab that
1:34:37 one
1:34:39 download going back to
1:34:42 LTX all right this we're going to
1:34:47 replace
1:34:50 with whatever the [ __ ] I just downloaded
1:34:55 that it
1:35:01 yeah camera
1:35:04 motion I don't know what scale
1:35:09 does we'll do natural wait balance
1:35:12 mixture of subject motion with camera
1:35:14 motion subject motion with only a little
1:35:17 camera motion yeah that's what we're
1:35:20 going to do
1:35:25 and then 3 seconds and sound effects
1:35:28 let's add a sound
1:35:29 effects
1:35:31 um Street
1:35:36 guitar no audio results found so we'll
1:35:39 do street
1:35:42 [Music]
1:35:45 performer
1:35:46 no uh let's do
1:35:51 um uh
1:35:59 uh
1:36:01 Samba
1:36:05 guitar no audio results found can I
1:36:08 upload something
1:36:12 here sound effects um let me
1:36:16 do uh
1:36:20 Street let me just see what that comes
1:36:22 up with City Street
1:36:26 Park Birds City Street
1:36:38 [Music]
1:36:54 dear
1:36:58 dear
1:37:04 crowd people
1:37:09 talking all right that's good let's do
1:37:11 that one all right so let's generate
1:37:13 this let's see what we get
1:37:16 generate okay so now this is going to
1:37:18 make a
1:37:21 video and
1:37:25 I don't how I don't know how it stitches
1:37:27 them together but let's see what it
1:37:29 let's see what it
1:37:33 does this is this is going to be
1:37:37 like this is [ __ ] hours
1:37:41 right I mean first of all it's going to
1:37:43 be hours just learning it right and then
1:37:45 it's going to be hours like oh try
1:37:47 busing yeah busking would have been
1:37:51 good Becky R surgery done hey
1:37:55 congratulations had a 10day checkup tiny
1:37:59 setback I'm tripping on these AI videos
1:38:02 I think that might be the the drugs
1:38:04 you're on Becky but yeah these videos
1:38:06 are pretty cool
1:38:07 [Laughter]
1:38:08 too okay let's play this scroll to
1:38:12 selected how do
1:38:16 I okay that's a [ __ ] up piece of
1:38:19 interface nonsense there we go
1:38:26 oh come on
1:38:31 really oh that's pretty good like just
1:38:34 as a
1:38:36 shot all right let's let's see now let
1:38:39 me go back to the
1:38:41 storyboard so we now
1:38:49 have let's
1:38:52 go so it says Reen generate video so I'm
1:38:54 going to I'm going to go preview and
1:38:56 Export and it in theory yeah see it's
1:39:00 got a little play icon here for that for
1:39:03 that shot so it should be static image
1:39:07 static image and then Street musici
1:39:20 playing Becky Ru haha life is good I was
1:39:23 teaching my lead surgeon about AI he's
1:39:25 amazed that's good you want to keep your
1:39:28 lead surgeons happy keep them engaged
1:39:31 just tell them not to have chat GPT um
1:39:34 do the surgery just yet we need we need
1:39:37 another iteration or two before it can
1:39:39 actually do that for
1:39:43 us oh that's really funny yeah I hope
1:39:46 you feel better soon
1:39:48 okay I came to Rio in 1972 on a lark I
1:39:53 figured I Stay for a week or two but she
1:39:56 cool okay she cool so let's let's go
1:40:00 back let's go render the first two
1:40:03 scenes so let's hop into the shot editor
1:40:06 here like I think that's a fine
1:40:10 image wide shot let's
1:40:14 do natural a balanced mixture of subject
1:40:17 motion and Camera motion so let's get
1:40:19 the camera moving I don't know what
1:40:20 scale means but we'll move that up a bit
1:40:25 and let's do 4 seconds
1:40:29 here and we've got Oldtown Street
1:40:33 Ambiance generate all
1:40:40 right Alan look in AI Show and Tell for
1:40:43 my student work very
1:40:51 cool did you do it in images
1:40:55 images oh yeah here we
1:40:58 go continued oh that's
1:41:01 cool nice
1:41:03 piece
1:41:05 wow here are some of my architecture
1:41:07 student results from prompting chat GPT
1:41:10 for the first time that one's flipping
1:41:15 gorgeous
1:41:19 damn I love the mix of blueprint and
1:41:23 rendering
1:41:27 I mean can you imagine if you wanted to
1:41:29 do something even remotely like that
1:41:31 with traditional rendering tools like
1:41:33 you'd have to learn 3D modeling you'd
1:41:35 have to learn Photoshop
1:41:38 compositing you know texture mapping
1:41:42 lighting holy
1:41:44 [ __ ] we are just in a whole different
1:41:47 category of
1:41:52 life that that's
1:41:56 perfect and then let's go to let's go to
1:41:59 our boy KS Dylan that's a little close
1:42:03 can we back back that camera off oh it
1:42:06 says it's a medium shot all right so
1:42:08 let's go fix that let's just go flip
1:42:10 through some seeds here but what's cool
1:42:13 is again it says the you can't really
1:42:15 see it but it says at KS Dylan so KS
1:42:19 Dylan is one of the characters so if you
1:42:21 want a character in the shot you just do
1:42:24 it like that all right that's a medium
1:42:26 shot so let's
1:42:31 do from above From Below Birds ey view
1:42:34 establishing
1:42:37 shot all right let's try a wide shot
1:42:41 cuz that closeup is a little
1:42:51 close doesn't really look like him but
1:42:53 let's flip through the seeds a little
1:42:55 bit he's supposed to have longer hair
1:42:57 than
1:43:04 that that's good that actually kind of
1:43:06 looks like that other scene so let's do
1:43:08 this
1:43:09 one and then we'll do we'll do what's
1:43:14 extreme stronger camera motion for more
1:43:17 Dynamic scenes now we'll do natural so
1:43:20 some camera some some him
1:43:28 and again all of this stuff is stuff
1:43:30 that you could do right if you've got
1:43:32 storytelling skills right you could do
1:43:35 this this is all stuff that you could do
1:43:37 in cap cut or you could do in Final Cut
1:43:39 Pro or you could do an iMovie by making
1:43:42 images in mid Journey putting them over
1:43:44 here but it just this is like a nice
1:43:47 organizing tool that puts all the [ __ ]
1:43:50 together so that's pretty cool and the
1:43:52 fact that you can make IM is outside of
1:43:55 outside of this
1:43:57 tool makes this a much more viable thing
1:44:00 to to not have just horrible
1:44:02 images let's see how bad this
1:44:06 is oh that's pretty
1:44:13 good all right let's go let's go back to
1:44:15 our
1:44:16 storyboard we'll do preview
1:44:26 the ideogram website has a new design
1:44:29 does
1:44:33 it I think it's the same since I've seen
1:44:36 it but maybe it's
1:44:49 different yeah IDE ideogram is a really
1:44:53 a really nice simple I like IDE a lot
1:44:57 all right let's look at our first three
1:44:59 shots of our first
1:45:01 scene I came to Rio
1:45:08 in I came to Rio in 1972 on a lark I
1:45:12 figured I'd stay for a week or two but
1:45:14 she had other plans who was she her name
1:45:18 was Ana and she stung me we didn't talk
1:45:21 much didn't need to the vibrant Street
1:45:23 of Rio became our stage and our Melodies
1:45:26 echoed through the city as our love grew
1:45:28 I faced a choice returned to the life I
1:45:31 yeah the script is horrible got to fix
1:45:34 the
1:45:36 script um all
1:45:40 right so that's LTX studio um it
1:45:47 is for for as early a piece of software
1:45:51 as it is it's very well thought out
1:45:54 right and the girl changes yeah listen
1:45:58 this this [ __ ] is going to be janky
1:46:01 janky is as janky does right
1:46:04 so I think that's just we're going to
1:46:07 just have to accept that it's going to
1:46:08 be it's going to be ugly for a while and
1:46:11 you know at some point maybe they'll
1:46:12 allow you to do things in other tools
1:46:14 and instead of the tool they
1:46:17 used
1:46:19 um but yeah but it you know it kind of
1:46:22 works
1:46:24 it kind of
1:46:26 works I wonder can you in the voice
1:46:31 over no you set the narration and it you
1:46:35 pick a you pick a voice and it narrates
1:46:38 I was wondering if you could upload your
1:46:40 own audio I think you probably
1:46:42 could there's also a soundtrack you can
1:46:45 upload a file for the sound track
1:46:49 so over in
1:46:52 yudo I was playing around with making
1:46:57 some some Brazilian kind of guitar
1:47:01 [Music]
1:47:06 music let's go grab this
1:47:10 [Music]
1:47:16 one that's good that kind of sounds like
1:47:18 1972 Brazil doesn't
1:47:21 it oh voiceover
1:47:33 narrator oh okay yeah voice over okay
1:47:37 soundtrack upload
1:47:48 file
1:47:52 okay preview and
1:47:58 Export just posted the urban community
1:48:01 designed by my architecture students in
1:48:03 Show and Tell yep that was super cool
1:48:05 Allan thank you for sharing that
1:48:09 Samba tap the screen to give Kyle those
1:48:12 likes thank you Becky Ru that's very
1:48:14 sweet of
1:48:19 you I want it now I know Cosmic lover
1:48:22 it's amazing
1:48:23 actually I I know it no it listen this
1:48:27 is very very cool like there there's
1:48:29 something in incredibly powerful in this
1:48:31 all right let's see if we got our our
1:48:33 soundtrack
1:48:34 now I came to Rio in 1972 on a l I
1:48:39 figured I'd stay for a week or two but
1:48:41 she had other plans who was she her name
1:48:44 was Ana and she stung me we didn't talk
1:48:47 much didn't need to the vibrant streets
1:48:50 of Rio became our stage and our Melody
1:48:52 it's so funny
1:48:54 he's talking about the vibrant streets
1:48:56 of Rio this looks like this look this
1:48:58 looks like uh like some some [ __ ] you
1:49:03 know blown out downtown in in like a
1:49:07 medium-sized Florida Florida town
1:49:10 outside of
1:49:12 Orlando it doesn't look like Rio at
1:49:18 all that's really funny actually in fact
1:49:21 let me go fix that one cuz that one that
1:49:23 one just bugs me shot editor Let me just
1:49:26 I'm just going to flip through some so
1:49:28 the evening Globe bathes a street corner
1:49:30 as the music so wait so the evening glow
1:49:34 bathes the Rio Street
1:49:44 Corner that's kind of cool but we want
1:49:49 people as the music captivates the
1:49:54 um
1:49:56 people p opal walking
1:50:07 by sh
1:50:12 [Music]
1:50:30 oh that one's kind of cool I should have
1:50:32 gone back to that oh
1:50:38 well the evening glow let's
1:50:41 see bathes the let's do bustling Rio
1:50:46 Street
1:50:49 busling Rio Street we need more bustle
1:50:54 it's a boring [ __ ]
1:50:56 City why why did KS Dylan stay
1:51:01 here this is like a Mall of America
1:51:04 nightmare come on people it's got to be
1:51:07 Rio we want some
1:51:10 life all right maybe the problem is the
1:51:13 wide shot let's go with a medium shot
1:51:26 oh that's good all right so let's do
1:51:27 that and then let's do
1:51:30 um we'll do extreme camera movement I
1:51:35 don't know what scale means but we'll
1:51:37 [ __ ] with
1:51:40 it sound effects let's go plus we'll go
1:51:43 Street
1:51:54 oops what did I
1:52:07 [Music]
1:52:09 do yeah that's good we'll generate
1:52:13 that so we did extreme camera movement
1:52:16 here so we should get some big camera
1:52:18 movement that'll make this more exciting
1:52:21 because this is boring right now
1:52:32 Rio and an upscale
1:52:35 feature and using pi with my Mac and I
1:52:37 absolutely love Pi yeah Pi's amazing
1:52:40 Pi's amazing idram is also working in
1:52:42 updating the image editor so that
1:52:45 subscription users have access to it
1:52:47 nice Copa Cabana definitely not Rio no
1:52:50 it's definitely not Rio I think I might
1:52:52 have to put Rio de Janeiro you
1:53:02 know oh that's kind of cool that kind of
1:53:06 shaky camera
1:53:08 movement for like a documentary of KS
1:53:11 Dylan like that's the kind of shot you
1:53:13 would get all right let's let's go back
1:53:15 to our
1:53:17 storyboard let's preview and Export
1:53:20 let's see what that shot looks like in
1:53:22 context I bet it look
1:53:25 better down down
1:53:28 down down down
1:53:43 down all right here we go people
1:53:54 I came to Rio in 1972 on a lark I
1:53:57 figured I'd stay for a week or two but
1:53:59 she had other plans who was she her name
1:54:02 was Ana and she stung me we didn't talk
1:54:05 much didn't need to the vibrant streets
1:54:08 of Rio became our stage and our Melodies
1:54:10 yeah that's kind of cool right as our
1:54:13 love grew I faced a choiceed streets of
1:54:15 Rio became our stage let's let's I want
1:54:18 to go hang on calm down people calm down
1:54:22 people
1:54:25 let's go let's do let's
1:54:28 see I don't give a [ __ ] about those all
1:54:31 right let's generate that
1:54:34 video Let's generate that
1:54:39 video that song's boring isn't it yeah
1:54:43 let's see if we got something
1:54:47 [Music]
1:55:02 that's more
1:55:03 [Music]
1:55:10 interesting uh uh soundtrack trash it
1:55:14 upload a new
1:55:16 one grab that
1:55:19 one all right did we render
1:55:26 oh so cute and then what's Ana doing
1:55:30 over
1:55:32 here wait what happened to her
1:55:36 hair oh the wind blew like it okay
1:55:39 [Laughter]
1:55:48 beautiful this is cool [ __ ] man
1:55:53 this is
1:55:54 something we can fix that
1:56:00 exactly
1:56:05 okay I came to Rio in 1972 on a lark I
1:56:10 figured I'd stay for a week or two but
1:56:12 she had other plans who was she her name
1:56:15 was Ana and she stung me we didn't talk
1:56:18 much didn't need to the vibrant streets
1:56:21 of Rio became our stage and our Melodies
1:56:23 echoed through the city as our love grew
1:56:27 I vibrant streets of Rio became our
1:56:29 stage and our melod vibrant streets of
1:56:32 Rio became need to the vibrant streets
1:56:35 of Rio became our stage see I think I
1:56:37 think rather than going to that shot you
1:56:39 got to go from because they're saying
1:56:41 the vibrant streets of Rio became our
1:56:43 stage that's not a bad line you go from
1:56:46 that shot to another shot of them like
1:56:47 walking hand in hand or something like
1:56:49 that I don't know we'll see anyway all
1:56:52 right I'm going to get out of here
1:56:55 um that's LTX Studio it's it's pretty
1:56:59 slick for for early stage software it it
1:57:03 does not suck nearly as much as it could
1:57:05 and
1:57:07 should um and you know again I go if if
1:57:12 this company can survive for a
1:57:15 year the underlying tools are going to
1:57:17 get so much better and so's like this
1:57:19 Tool's going to get more sophisticated
1:57:23 and then but but so then a really
1:57:26 interesting thing starts to
1:57:28 happen
1:57:32 um while this tool improves while LTX
1:57:35 Studio
1:57:38 improves so to is Sora is going to come
1:57:42 out Runway is going to come
1:57:46 out and probably about the time if you
1:57:49 go a year from today 12 12 month wants
1:57:53 out we're probably at the point where
1:57:57 you could just say make me a movie in
1:58:00 Rio Brazil and this you know you
1:58:02 basically just put the prompt into some
1:58:05 chat GPT like thing and it will squirt
1:58:08 out the whole movie now what you
1:58:11 potentially lose there is any control
1:58:13 but you may not need the control right
1:58:15 so the the downside of something like
1:58:18 LTX studio is also its strength right
1:58:21 your ability to be anal retentive and
1:58:23 fix all the [ __ ] and and Master all your
1:58:27 shots but we're probably going to have
1:58:29 tools come along that you won't need to
1:58:31 do that
1:58:33 so this the weird like this and and you
1:58:37 know
1:58:39 like that needs to be a consideration
1:58:42 for people like if you're thinking about
1:58:43 starting a company you got an idea for
1:58:46 something don't assume that whatever the
1:58:50 limitations are today are going to be
1:58:53 there a year from now which is kind of
1:58:55 [ __ ] up like where do you start what
1:58:56 do you do I don't know there's not a
1:58:58 good answer for
1:59:00 it all right uh please follow the
1:59:04 channel subscribe to the lives pick up
1:59:07 the video series there if you want to
1:59:10 support the channel um keep coming back
1:59:12 to these night after night I'm here come
1:59:15 back we'll show [ __ ] like this um I
1:59:18 should be getting the rabbit I'm
1:59:20 guessing in the next week or so so I'll
1:59:23 definitely do the unboxing
1:59:25 live um and we'll play like from from
1:59:29 what I can see it's not horrible like it
1:59:31 doesn't suck um you can make mid Journey
1:59:33 images with it like they like in in the
1:59:36 launch version of it they've got mid
1:59:38 Journey set up so you just push the
1:59:40 button and tell it to make images and it
1:59:43 makes them on Mid journey and then
1:59:44 populates your your rabbit screen with
1:59:46 your mid Journey images that's pretty
1:59:49 slick that's pretty slick so we'll see
1:59:53 um all right um thank you to my
1:59:56 subscribers thank you regulars thank you
1:59:58 mods yeah good modding tonight I
2:00:00 appreciate you know pulling pulling
2:00:02 stuff up as I was demoing cuz whenever
2:00:05 I'm demoing I get completely lost in the
2:00:07 comments um utto thank you very much uh
2:00:11 Becky Ru welcome back glad you're glad
2:00:13 glad you're on the mend and you've got
2:00:16 your sense of humor and wonder that's
2:00:19 awesome um Tam Z good to see you Allan
2:00:22 good to see you Cosmic lover
2:00:30 Joey Angelique Gavin biddy Papa
2:00:41 JJ all right I'm out of here have a good
2:00:44 night what day is today today is
2:00:46 Thursday tomorrow's Friday night date
2:00:48 night I don't think I have anything
2:00:50 going on but if I do I'll make a video
2:00:54 um I got a busy day tomorrow so I'm not
2:00:56 going to be able to pay much attention
2:00:58 to what the [ __ ] going on with AI but
2:01:00 hopefully something cool
2:01:02 drops we'll see but all right good
2:01:05 talking to y'all hope you have a good
2:01:07 night bye