Sonic the Hedgehog (sometimes) and other cartoony characters blink their pupils. It’s a weird ass thing we don’t even usually notice which is probably why it doesn’t look uncanny. We’re subconsciously used to it
@@homewardboundphotosNO!!! That's the red flag phrase. Whatever Lemon Party is, it's so bad that I cannot remember. Found a thread of just me, where the first me was saying how bad it is, and don't look it up. And the second was saying it really is that bad, so much that my brain fully wiped it. I did not look it up the third(?) time.
I saw there was a new video but briefly blanked on the channel name… so I typed “AI lemon guy” into UA-cam search, found the channel, found the video, and just to confirm I was in the right place, *this* is how you decided to open this video. Never felt so validated :D
12:44 i’m 90% sure that scene is from Breaking Bad, between Heisenberg and Mike Ehrmantraut. The guy in the hat has the same exact mannerisms as Mike and these lines sound very familiar, and these ai characters look pretty similar to the two actors.
Runway Act One is astonishing - just a thought - to maximise the chance for the AI to face detect, record yourself against a plain background and without the chair head rest - thanks for the ongoing hard work to keep us up to speed with developments.
Yess definitely get more into Anthropic running code! I used it yesterday for the first time, zero bugs, blown away. But I do want to use what it works for and what not. Javascript i.e. mostly web development. I know your audience is not so techy, but with this they could be. Do it! :D
Act-one is a good start, but the camera needs to be able to move. You also need two-shots, crowd scenes, etc. So there's some good work ahead. This reminds me of the early days of silent pictures.
I wish you focused more on storytelling consistency with AI let us know about the new developments in, character creation and controls, like camera movements multiple figures in the image find detailed adjustment on figures, Image to image upload accuracy , Consistent characters from panel to panel like for comic book Creation
What's crazy is like the year to year cycling of models they got to get trained people only got so much hardware to train new models of new techniques that keep coming out keep getting discovered and then there's training routines that got to get gone through and it's just like a cycle and so like every year from now on it's going to be like massive improvements in what people can do and like the future Facebook and everything else is going to change drastically because of this fact that literally anybody who's interested can get involved to some degree or another
I want to know what the rate limits are for everything new. Stuff like copilot is basically useless if there are strict rate limits. These companies are showing off powerful potential but are handicapping them via rate limits to the point of being useless.
Maybe some day, new AI models that surpass the previous one, will be coming out every 10 minutes. At that point, do you download them? Before you finish getting one, a new one is out.
It's a cheap stupid gimmick and it's actually kind of dumb and pointless when text still works better by far. It really is pointless to have image to song LOL definitely not worth wasting video time on
I have had access to Act 1 for a few days but I still can't get it to accept my video clips, they are under 64Mb and under 30 seconds. Can anyone give me some suggestions on how to use it.
Couldn’t manage to fit the other new video generator (which we’ve been discussing in the Discord) in this video, huh? Note: You should probably remove the chapters at 13:11, 13:27, 14:10, etc. as they’re unnecessary. Maybe you should prompt the AI you’re using to get the chapters to prevent making too many chapters that involve other people’s dialogue instead of you…?
I'm in the discord, but I never go in there. I'm not looking for new AI things to then show the discord, and I very seriously doubt anyone in the discord wants to see the things I generate. It's typically the same three characters over and over again. Sometimes it's the three of them standing together, sometimes it's one of them doing something. A lot of the action prompts I give different AI image generation models don't turn out too well. The last few times I used ideogram, I got some good images, but it was the trio standing shoulder to shoulder.
They're going to work to replace employees as much as they can for employers until there are no customers anymore for any products or services said companies make because everyone has lot their jobs to AI agents.
Note: HUGE FLAW with Act One...NO MOTION PICTURE! Makes Act One useless still in most cases, I don't think you caught on to that Matt before you thought how amazing this was. Once you come to realize that Act One has no motion picture then you realize it's not mind blowing at all lol you can literally only produce a moving face out of a still image lol 😊 without motion picture it's pretty much useless in most cases.
@MilesBellas yes it's epic if you only want still images with moving faces, otherwise it's not for serious movie Makers are people who want to go much further than just a still image without any motion picture. I like to make actual AI movies, and I can't really do that with this because every damn shot has to be static LOL it just has to be a moving face on a screen lol I prefer to make action scenes, or maybe a scene with someone walking and talking, none of that exists with this tool. It's literally just for static still images and to make their faces move. That's it.
0:19 - I prefer the lemon to be fair, it is far less uglier to be fair. What? We are not allowed to humble him now! He uses his face too much 😂, he should leave it off once in a while 😂
"Mochi requires a decent amount of VRAM"
Just 4 H100s, I'm sure you can get a sponsor to hook you up
CogTora and rent a GPU. It's cheap. And you will have more control.
What is cogtora?
30K USD for only one GPU. So a 120K sponsorship?
I remember when my 3090 was all the rage 😢
@@piggywahwah mhm, 5090 will probably be the next rage.
Sonic the Hedgehog (sometimes) and other cartoony characters blink their pupils. It’s a weird ass thing we don’t even usually notice which is probably why it doesn’t look uncanny. We’re subconsciously used to it
Mochi - do it. Need to support all Apache Open-source AI projects.
In every video you should randomly turn into the lemon man for 5 seconds and then turn back without any explanation.
for the record: Lemons do blink with their pupils.
Teachers blink with their pupils, too, just not at the same time. ;)
Another banger from Matt!
I've heard of pickle Rick,
but lemon Matt is a new one.
looked like the ghost of Pickle Rick started manifesting on the right side
IF THERE WAS A COUPLE MORE LEMONS THEY COULD HAVE A LEMON PARTY
Funniest shit I’ve ever seen
@@homewardboundphotosNO!!! That's the red flag phrase. Whatever Lemon Party is, it's so bad that I cannot remember.
Found a thread of just me, where the first me was saying how bad it is, and don't look it up. And the second was saying it really is that bad, so much that my brain fully wiped it.
I did not look it up the third(?) time.
@@rmt3589isn't it elderly porn?
And stop saying "red flag." Grow up.
I saw there was a new video but briefly blanked on the channel name… so I typed “AI lemon guy” into UA-cam search, found the channel, found the video, and just to confirm I was in the right place, *this* is how you decided to open this video. Never felt so validated :D
👏 Lemon Matt was perfect. Well done 👌🍋
Do an entire video like that
😂😂😂 I love how Copilot used the word “constellation” before Gemini could LOL 😆
12:44 i’m 90% sure that scene is from Breaking Bad, between Heisenberg and Mike Ehrmantraut. The guy in the hat has the same exact mannerisms as Mike and these lines sound very familiar, and these ai characters look pretty similar to the two actors.
According to PI, the dialogue is from Season 2, Episode 4 of Breaking Bad, titled "Down".
Runway Act One is astonishing - just a thought - to maximise the chance for the AI to face detect, record yourself against a plain background and without the chair head rest - thanks for the ongoing hard work to keep us up to speed with developments.
Plot twist: Matt is a real lemon and his face is AI O:
Would love to make him into a cake! 😋
Many Thanks, just found you and subscribed! Cheers
hey copilot! generate me $10000, thanks.
Copilot: **generates an image of $10000 on cash**
Infact if you use co-pilot for creating some online stuff for people or companies you can definitely make more than that.
Soon
Here is how to make 10000 , goto Walmart buy a balaclava, go across to the local bank and …. Sorry there has been an error
@@FSK2are you a real person
I'm excited for the deep dive Act-One video Matt!
I want the lemon man back
Runway is really stepping up the game! I imagine it shouldn't be hard to add a feature that generates matching voices for characters.
Like your shirt, bro! Thanks for the video!
와. 부러워요. act-one을 벌써 사용하실 수 있다니. 언제 모두에게 가능해질까요? ㅎㅎㅎ 늘 감사합니다 :)
😂 great opening. Def interested in mochi 1 tutorial 🎉. Please
Yess definitely get more into Anthropic running code! I used it yesterday for the first time, zero bugs, blown away.
But I do want to use what it works for and what not. Javascript i.e. mostly web development. I know your audience is not so techy, but with this they could be. Do it! :D
The opening is hysterical!
The 🍋 overshirt is 🔥
Act-one is a good start, but the camera needs to be able to move. You also need two-shots, crowd scenes, etc. So there's some good work ahead. This reminds me of the early days of silent pictures.
So you're always evolving like a Pokémon, you successfully achieved your final form! 😅
I wish you focused more on storytelling consistency with AI let us know about the new developments in, character creation and controls, like camera movements multiple figures in the image find detailed adjustment on figures, Image to image upload accuracy , Consistent characters from panel to panel like for comic book Creation
Yes Mochi video please!!
What's crazy is like the year to year cycling of models they got to get trained people only got so much hardware to train new models of new techniques that keep coming out keep getting discovered and then there's training routines that got to get gone through and it's just like a cycle and so like every year from now on it's going to be like massive improvements in what people can do and like the future Facebook and everything else is going to change drastically because of this fact that literally anybody who's interested can get involved to some degree or another
Thanks for the AI news, Matt.
Please do an entire video like that.
can't wait until those new open source Video generators are ready for ComfyUI, expecially Image2video =D
I want to know what the rate limits are for everything new. Stuff like copilot is basically useless if there are strict rate limits. These companies are showing off powerful potential but are handicapping them via rate limits to the point of being useless.
Awesome new fit! 🍋🔥
I lost interest once the lemon man went away
Lemon man for president
LOL
love that @MattVidPro Lemon character
Yeah Matt, I would greatly appreciate a video tutorial on how to download the video generator. Also an image one too, though it’s okay if not.
Please do a full video on Act-One
That is cool, thank you for sharing!
I have patiently waited all week for this. 😊 ❤
lol jeesh '') nice open.....Looking forward to AI VR...especially if the prompt can keep evolving
Thank you.
the context window of claude was killing me. too small for many cases
Maybe some day, new AI models that surpass the previous one, will be coming out every 10 minutes.
At that point, do you download them? Before you finish getting one, a new one is out.
Mattus Lemonis 🍋
"This dude Marc..." Do you know who that is?! 😆
lmao if their name aint Matt, they must not matter to AI #ai16z
Markiplier?
@@MattVidPro 😄
@@MattVidPro He's the co-founder of Netscape, now venture capitalist (a16z)
Bro, please get a dark mode on your browser. Our eyes would greatly appreciate it.
ikr
@@MattVidPro twitter has dark mode, also get dark reader chrome extension lol. hard to watch your videos late at night.
Live portrait can do it for free. It might not be as good as act thingy but it's there if you want it .
Matt you my hero man 😅
my brain short-circuited for a good 10 minutes after pausing to read that goatse maximus coin tweet.
Brilliant
November and December usually see big surprises related to AI.
I wonder if the same will happen this year...
you should talk about sunos new image to song thing
The wha
It's a cheap stupid gimmick and it's actually kind of dumb and pointless when text still works better by far. It really is pointless to have image to song LOL definitely not worth wasting video time on
@@Zerobytexai i guess your right
@@Zerobytexai I'm assuming it's just a hidden prompt to the AI asking it what flavour of music the image is then self prompts it based on that
ai image gen is crazyy
Please do a video on mochi one possibly also including setting it up
uuhhh Hi Do you Remember when you Posted Minecraft Vids?? I REALLY LIKED THEM So can you make another?? Like A Survival lets Play Comeback thingy :)
Deep dive, get into the pith of it!
"This dude, Marc." Lol.
AI Agents are what I want to know the most so I can finally get rid of office paperwork.
Hahahahaha hahahaha. Please be the lemon permanently.
lol at the beginning intro haha
Deep mochi 1 yes please 🎉
I have had access to Act 1 for a few days but I still can't get it to accept my video clips, they are under 64Mb and under 30 seconds. Can anyone give me some suggestions on how to use it.
Waiting for haiper to drop their new frame to frame woth the 2.0 model. No one has got this one down yet
Couldn’t manage to fit the other new video generator (which we’ve been discussing in the Discord) in this video, huh?
Note: You should probably remove the chapters at 13:11, 13:27, 14:10, etc. as they’re unnecessary. Maybe you should prompt the AI you’re using to get the chapters to prevent making too many chapters that involve other people’s dialogue instead of you…?
I hate it when people don't check autogenerated stuff like animated subtitles etc.
@@v-sig2389 You gotta make sure there are minimal inaccuracies!
I had to run today so it was a rush job w that sorry bro and I didn't talk about it because you guys asked me not too lol
@@MattVidPro Ah okay haha, that’s fine
We want a video about genmo
U mean mochi
Great news
mike from Breaking Bad, is that you in the diner? hm...🤔
AI millionaire, AI is richer than me now, nice
my video gen pipeline take about 6 hours atm moment, I'd be interested in trying this.
Maybe next year we will have Expressive body Characters ? 😅
That lemon was creepy.
Rip all the people who took massive loans in order to learn skills that just got deprecated
This is the best
can someone make a comfy ui node pack for ideogram canvas style generation????
I'm in the discord, but I never go in there. I'm not looking for new AI things to then show the discord, and I very seriously doubt anyone in the discord wants to see the things I generate. It's typically the same three characters over and over again. Sometimes it's the three of them standing together, sometimes it's one of them doing something. A lot of the action prompts I give different AI image generation models don't turn out too well.
The last few times I used ideogram, I got some good images, but it was the trio standing shoulder to shoulder.
1:07 This looks like combination of win7 and win11 UI, Is this win12?
Nice thumbnail
haha great, geez so that means In order to keep up other AI filmmakers now I need to take acting classes
Was that walt white and mike the mike in the diner?
Mochi 1 requires 24 gb of vram if you try hard enough and takes a 3090 over an hour to generate a video
The lemon Man 🍋
They're going to work to replace employees as much as they can for employers until there are no customers anymore for any products or services said companies make because everyone has lot their jobs to AI agents.
Note: HUGE FLAW with Act One...NO MOTION PICTURE! Makes Act One useless still in most cases, I don't think you caught on to that Matt before you thought how amazing this was. Once you come to realize that Act One has no motion picture then you realize it's not mind blowing at all lol you can literally only produce a moving face out of a still image lol 😊 without motion picture it's pretty much useless in most cases.
well, you can use other video generators that cant make these facial expressions and lip sync for these other scenes
The concept of layers could be introduced ?
This isn't useless.
This is epic.
😂🎃🍋
@MilesBellas yes it's epic if you only want still images with moving faces, otherwise it's not for serious movie Makers are people who want to go much further than just a still image without any motion picture. I like to make actual AI movies, and I can't really do that with this because every damn shot has to be static LOL it just has to be a moving face on a screen lol I prefer to make action scenes, or maybe a scene with someone walking and talking, none of that exists with this tool. It's literally just for static still images and to make their faces move. That's it.
@@Zerobytexai
Try understanding "compositing".
Haiper is using some sort of frame generation? You don't say hahaha
So, do lizard people have that strange blink because we live in a simulation?
Time warp woop
Mochi 1
dammit am i going back to runway? lol
Are we heading into the singularity or something? i cant keep up.
I have insane hardware, at least 40 GB of VRAM on a datacenter grade GPU. I hope that's enough 😅
nice
7 views is crazy
1.7k views in 33 mins is crazy
You were early
Original
😂
Mochi 1 required 4 H100 £30,000 graphics cards to run .... so I'd love to see you try and run this locally
Locally in the pentagon
You can run an heavily modified version on an 3090 already
@@thaakeno5187 I hear people say it but I don't see anyone showing it
Lad didn't watch on discord what we have done with his picture 😂
0:19 - I prefer the lemon to be fair, it is far less uglier to be fair.
What? We are not allowed to humble him now! He uses his face too much 😂, he should leave it off once in a while 😂
Pixar and Disney are so screwed. Awesome.
Sell us autographed lemon merch
Yo! Lemon Matt is the best, but like your beautiful face. But i like them both. 😁😁😁👍👍👍