OK. Now I'm Scared... AI Better Than Reality!

[@vi6ddarkking]

Honestly this got so much attention in part because of the total disaster that was SD3. With Stability seemingly going screwing the pooch, the Open Source throne was empty for a new king. And the fact that most of the team are Stability alumni, is just the cherry on top.

[@somnaut]

I'm sure that's not a coincidence. Bet you anything those folks left Stability because of the direction they were heading.

[@haroldsmith5761]

This basically just a Flux commercial

[@generalawareness101]

All the channels are this way over the last week. Sad.

[@DJVARAO]

And?

[@NoidoDev]

​@generalawareness101 Well, it's good, and the fast basic version is free (open source).

[@LAALLEN-dg1mk]

Pretty much

[@BanXxX69]

Yes and I‘m glad I‘m shown that this exists😊

[@jerzyukasiewicz1468]

Flux.1 - schnell requires about 80GB VRAM GPU for fast operation (schnell). Time to take 1 photo in FullHD resolution on CPU i9-13900KF with 256GB RAM is about 9.5 minutes! Don't tell stories that this is a solution for a home laptop... The Flux has the same finger count issues as other underdeveloped models. The Flux has also problem with a text. This is a very interesting model, but it is far from perfect.

[@andrewpriest09]

Who told you that?? i dint think there is any truth in that statement ,i run it on motherboard graphics I7 4600 proc. generates Ultra high def images in about 1.5 mins i have no problem with text, Where are you using this model

[@jerzyukasiewicz1468]

@@andrewpriest09 Noone, i didn't ask. Just checked it out.

[@MAJEPHTIC]

4:16 "If you really start to scrutinize these images, you'll notice some telltale signs that they're AI generated." ***shows a picture of a dude with a donut in place of his head and 6 fingers on his left hand***

[@phen-themoogle7651]

I saw that dude getting a donut before. He’s real, 6 fingers and always has a giant donut stuck to his head. He even killed Inigo Montoya’s father too.

[@zipauthorzipauthor7867]

I thought a person with a donut for a head was a big give away 😂

[@gordonbrinkmann]

@@phen-themoogle7651 😂😂😂 I love Princess Bride references

[@indiffident]

many people have a completely distorted idea of ​​reality, especially in the world of photography, due to wild postproduction.

[@adriantaner7950]

5:11 “and it’s not hard to see why”. Notice how the why lacks inflection, it’s a constant tone. The voice is AI generated

[@danboud8135]

Yup, this is not a real person speaking.

[@MarcelGomez-e3l]

Yeah, I am a video editor. It is a default voice in almost all edition softwares.

[@jesussaddle]

@@MarcelGomez-e3l Such as which ones, by which companies?

[@Be_Nice_To_Dogs]

I wouldn't be surprised if it's all AI. So many channels are. For example, the channel True Crime Cases Studio is written, compiled and narrated by AI, and all its imagery is AI. There is literally no human involvement. Despite the fact it's passed off as 'true crime' yet is 100%T AI generated fiction, YT are happy to allow it continue despite me reporting it and providing evidence it is fake. Go figure.

[@TomHendricksMusea]

But this problem is not just recent ai developments! Remember. YOU READIN THIS ... , you can push a button and sing with better pitch than anyone who's ever lived .... but its all fake. It's also an industry standard. Pitch monitoring software shows it. AI won THE ENTIRE MUSIC INDUSTRY ! REAL HAS BEEN REPLACED! Whos fixing all Billboard vocals till they're fake? Proven fact. Pitch correction or autotune ruin everything and it's being covered up by the musicians, radio, and music media. What Happened to Music? 1 Records have fake ai autotuned vocals 2 Sales have crashed since 1999, that's 24 years straight. 3 The Big Three Labels, 3 old men CEOs, control the industry. Warner, Universal, Sony. 4 No hits Push a button and you can sing with better pitch accuracy than any singer ever.... but its all fake!

[@valuemastery]

OK, so the benchmark is reality. If it meets reality, it's as good as it gets. How can it possibly be "better than reality"? That would mean that it somehow differs from reality, which per definition would be worse than reality.

[@Eddierath]

Thank you for this... These people are digging their own holes with what they promise I swear.

[@matusfrisik3887]

Reality is sometimes disappointing.. or just boring.

[@Zanroff]

Better than reality would mean reality without all the pain and misery.

[@aaronmarchand999]

Take an attractive girl for example, if AI can make an image more attractive than than the hottest of the hot in real life then in a sense it's better than reality, ie. reach 100% realism but then create things beyond what is possible in reality

[@WESKER_294]

Yapping

[@Cyba_IT]

Is it me or did all the Midjourney pics look better than the Flux ones in this video? I've seen other vids where the Flux pics look amazing but this vid didn't show that, probably because it's AI.

[@mirek190]

to get good skin you need setup flux guidance 2 ... om the video seems be default 3.5 where skin looks awful

[@jesussaddle]

"We're already seeing concerns about how AI-generated images could be used to create fake news". Actually I was expressing concerns like this in the 1990's and early 2000's. Philip K. Dick was around Burbank talking about stuff & that really brought it home - Total Recall plot becoming real. No one listens. Frustrating!

[@redpillnibbler4423]

News is largely already fake.

[@CommanderKOfficial]

in a few years or decades we wont know if actual real people in front of you are real or just a computer

[@FutureSyncAI]

Personally I'm not going to have that problem

[@NoidoDev]

@@CommanderKOfficial Maybe, but most robot girlfriends will rather be optimized for looking supernaturally beautiful.

[@Theguywhowanders]

We are fleshy robots tho 💀

[@CommanderKOfficial]

@@Theguywhowanders I suppose

[@Miniflower25]

Man that's scary 😢

[@feasterfamine836]

Zero flux given

[@zankfrappa93]

"each pushing the boundaries with their updates" your ai voice messes up "their" and its also just sounds like a really unlikely thing for a normal human to say. so much for the AI revolution

[@mahiaravaarava]

That's some seriously mind-bending AI tech! As someone fascinated by the potential of artificial intelligence, I'm both excited and a bit nervous about where this is heading.

[@8singularity]

Ironically, this entire video, including the script and scenes, appears to be AI generated.

[@jelefaz]

And it´s only just begun...

[@outcastcwd]

AI is better than reality... To be fair, reality hasn't been setting the bar very high, lately....

[@MikaParker-z7e]

Ai can't get pregnant

[@RS-Amsterdam]

No AI Karens

[@davetaylor2088]

The Running Man movie is now real life.

[@xXx-lfg]

Your AI voice is awesome, is that Elevenlabs?

[@ereceeme]

Have any of these apps been tried on the new snapdragon laptops with their neural engine ?

[@duck-headedllama9991]

This video is perfect to show everyone that Midjourney is still the king of image generation AIs.

[@NorbertEnders]

It’s good to see another contender on the market. Especially as we see now more balance worldwide. The US still is ahead in many ways, but followed closely by China (Kling), UK (Stable Diffusion) and now Germany (Flux & Sota / Black Forest Labs). User oriented development needs competition.

[@1Infeqaul1]

Reality with out natural compassion, would be the hell of hells.

[@getivan]

There's nothing proprietary, here, though. Grok 2.0 is the best I've seen on text, right now.

[@udiorockmeamadeus]

And yes I'm using LORA on huggingface.. It can't draw the multiple Grammy winner of the 70s who has the distint look and the flugal horn.. He is popular for a song called "Feels So Good".. HE creaated the music for canonball run 2..

[@Hobnockers]

Can you use flux with image to image?

[@DiyEcoProjects]

Can i ask something please ~ how do i make sure what im downloading is NOT a virus? is there a way to tell please?

[@ericerian1811]

You can follow or, you do follow me from the very beginning, Means, be scared! The UE 4, and 5? At the first, its the 'somehow' wizard thing within it that really works beyond my imagination.

[@jsivonenVR]

AI “art” is a misnomer. “Art = the expression or application of human creative skill and imagination, typically in a visual form such as painting or sculpture, producing works to be appreciated primarily for their beauty or emotional power.” Pls refer to them as AI pictures 😊

[@SeraphicPhantasm]

It's generated by a human. Ofc it's a human expression.

[@CERISTHEDEV]

Dude people generally say "ai art" because is short and it rolls out the tong better people tent to abreviate words even if the shotened version makes less sense as rhe longer one Stuff like yall for you all or cant for can not Like in my language people say "Imagem ia" but this phrase breaks grammatical rules but people say it anyway because its easier to say

[@kellyblack2010]

It still an art… because the human expression comes from something that only humans have..imagination. image generation just allows you to conceptualize what you thought about and turns it into visual representation. Also you can Make it look any way you want or better than you imagined it in your head. But what will truly be impressive is when Ai generated video becomes more viable to where the is no warping or distortion and it indistinguishable from an actual recording,that would be where it becomes interesting. Take for example how live action films of novels are usually adaptations done by someone else other than the original author, sometimes they can be good other times it could have been done better…. But what if the original author could basically generate the entire movie and just hire voice actors. They could craft every scene exactly like they want, detail the characters specifically to how they envisioned them …everything down to the scenery. Even the style of the movie.

[@jsivonenVR]

@@kellyblack2010 nearly everything you described makes “art” even more of a disposable commodity. Entertainment, not art. Kinda what happened to pop music in recent decades. Which might be good outcome since art with passion, craft and true creativity stands out from the slob. 👌🏻

[@rev.jimmywonko9615]

Yet a collage is considered art. Humans made AI. You can disagree with creation process, but it IS art.

[@DavidVargas]

How did you manage to create such a great video??

[@EdL-m3b]

The best is know that you can paint images using Photoshop or even paint/draw with pencil/oil paint/air brush but now you have Ai as part of your arsenal. Sure, there are slew of 3D tools too. The best Ai artists are those who have art & design training. Real artists.

[@steve6533]

The best AI artist won't be able to compete against non-artists using the best AI.

[@francoissoulier5849]

It look incredible, yes. And it's even difficult to imagine concequences in only 3 to 4 days.

[@tmarq4655]

What is shows me is that it's a pain in the butt just to get it. This is like asking permission from the Pentagon to go to the lunch room. Seriously, a quick start button would help get moving quickly on using and trying it.

[@chriz9959]

its not really that hard to set up comfyui with lora models and get workflows and test. i got stunning results in 10 hours from installing to rendering images

[@tlatosmd7074]

@@chriz9959 An install and setup process taking 10 hours is certainly not "easy" or "fast". 99% of the potential target demographic locked out by default.

[@Professorke]

I created my own AI photo generator by using pieces of code from various open source and supplemented by my own code. On my PC, I7-9700K/64GB and RTX3060/12GB, my AI spits out nice pictures between 4 and 6 seconds, and this completely uncensored, I can generate anything. I named it Nancy, after an ex sweetheart, and I have already given many examples on X. I do not put the code online, because in current form it is too dangerous, because Nancy can generate just about anything with existing AI models. Nancy still makes mistakes, yet every day I improve the code. Sometimes I have to make mistakes on purpose, because sometimes the result can be shocking. It is a hobby and should not become more and do not want to damage people. I only do it to learn from it, nothing more!

[@BxPanda7]

That is really interesting did you create a completely new backend or a custom webui ? Also 4-6 seconds !! With how many steps and at what resolution ? Did you grab the optimizations from Forge or is it something entirely different ? With forge I can make images in 20steps 1024x1024 in around 10-11 seconds on a 3070ti (8gb).

[@austinrusso2178]

I use Open Art AI for any new models now. I enjoy the convenience.

[@CharlesKyle97]

Tell the AI to give you an image of itself.

[@NScherdin]

Everyone wants to point out how "it's not perfect". My rebuttle is "yet".

[@petercroft9895]

Like your spelling. Rebuttal :-)

[@udiorockmeamadeus]

Its not as real as you think, someone compared it to midjourney, and it was so so.. But it is very realistic on photos,

[@flouserschird]

There needs to be someway to regulate watermarks on AI images to detect real vs fake

[@helifynoe9930]

So far, no luck. Attempting to get into the "Let's Talk" conversation, taking place within this video as suggested. So far, the communication is still unidirectional, and thus we have not been able to get even a single word in. We are aware that to do so, we have to alter the past, thus literally reach back in time and in turn alter the video itself.

[@lastchildz8567]

And just like other AIs that tried to compete with midjourney, Flux will lose because it is not easy to learn and the token system limits prompts.

[@mal2ksc]

Wrong! The only place it is even helpful "speak bot" is on the negative prompts, and only because the things it does that you would want to suppress are so specific. Signatures, watermarks, and bad hands are the three I always start with. Otherwise, you can talk to it like you were describing a scene to a set director, and it will do an excellent job of interpreting that. If you want two characters dressed differently, you say "Two women. The woman on the left has blonde hair and red eyes and is wearing a blue dress. The woman on the right has red hair and green eyes and is wearing a white dress." And that's what you're going to get. Of all the prompting systems I've seen, only SD3 and DALL-E can match its ease of use, but Flux comes without the built-in nanny filter. This gets a HUGE thumbs up from me. It is exceptionally approachable if you can speak coherently, no more having to learn bot-speak to get your images!

[@Rde-enterprises]

I recently switched to Flux as my go-to image tool!

[@PeachtreeGuitar]

What mic are you using? It sounds great

[@screenstory]

There is no microphone: it’s an AI-generated voice.

[@buntnik]

Wow! Very impressive. Where do I buy it? Is there an image maker that’s any better?

[@GenoG]

Yeah, not much "info" about how-to, etc for Flux... Was hoping to get some tips. I've been using the Flux BNB NF4 version on ForgeUI and it's been pretty cool!! Weird not to have any neg prompts and CFG stays at 1. But, detail and capability are all showcased really well... It pretty much does what you tell it to! Prompts are "real language" story problems, so that's different than SD1.5 which has been my path to get here. No XL or Pony (for God's sake :-D), so I don't have too many bad habits as my SD1.5 prompts seem to drop right in to Flux and work as is... certainly room to "learn the lingo".

[@babyfox205]

Which of these can be used for stylizing transparent PNG game assets? Does any of them with with transparent background?

[@mal2ksc]

Try generating anything related to archery. You'll have a gigantic fail. This is also a severe weakness of Stable Diffusion right up through V3.0. Second flaw I've noticed: if a face is upside down, it isn't recognized as a face. Good luck generating images of people doing acrobatics, or certain lewd positions.

[@MichaelLeeOne]

I love Flux. It is awesome. Still waiting on an affordable AI that can generate video content, though.

[@hurdygurdyguy1]

"...let's not get bogged down in the doom and gloom just yet..." let's give it a few days!! 🤣 Funny, I tried the online Flux Schnell with the prompt, "female hippie with bare breasts dancing underwater" and got exactly that, bare boobs...but, trying again I got female hippies with bikinies, every single time ! It was on to me! 🤣

[@denlillaekorren]

01:41 the promising future

[@ex0stasis72]

This video sounds like it was both written and voiced by AI. Not that it's bad. It iist has a certain ChatGPT signature feeling to it.

[@iparkedmycar]

I mean as a Midjourney users since it's launch flux isn't at the head of the pack.

[@MyrLin8]

Realistic is good, too realistic is not, and mostly not necessary. just because you can tell it's AI? sweet! it's AI :) good enough. The key folks is Open Source. :) that's the ticket :)

[@AINewsUSAupdate]

Very good video, lots of things come to know. Thanks for the information

[@Carlo-h2b]

I am waiting for AGI at 2029 narrow ai is impressive but not human like

[@QuicksilverSG]

The Pinokio install for Mac is bogus. After downloading and installing the dmg, you're supposed to launch patch.command. This runs a script which requires a password. Without that password, the Pinokio app won't run.

[@Davidsavage8008]

The fingers and toes gives it away.

[@cyborgmetropolis7652]

Let's also comment on the AI script and the AI voiceover

[@beardordie5308]

This narration is great but there are some tells that it's AI. Not many, though. Is it elevenlabs?

[@xxredshiftxx]

I need a more powerful AI , bring it on 👽😁

[@qubit0002]

how could we not be in simulation when it only took humanity couple hundred yrs to reach this point while universe is billions of yrs old...

[@funkymunky]

16 gigs of RAM, i7 core, NVIDIA GPU - that enough? Also: can I use Flux to edit visuals uploaded from my laptop?

[@Wyn3e]

Nvidia GPU with 12 to 24gb of vram same with AMD if ur running Zluda on windows or ROCm 6 on linux and 32gb of ram is pretty much needed and idk what u mean by edit uploads

[@funkymunky]

@@Wyn3e Thanks! That's within my laptop's capabilities. I'll rephrase: can I use Flux to work on visuals uploaded directly from my laptop? Or is it limited to native creation?

[@BxPanda7]

@@funkymunky The model itself is over 16gigs so you don't have enough ram to load the model on your machine. You can still use it using a cloud service though.

[@funkymunky]

@@BxPanda7 Cheers!

[@priestpega]

I tried it and honestly, I don't see much difference from the others. On some images I got extra legs or missing limbs just like the others.

Hey, so cool, when i saw the title i thought this also about my edited pix, because i uploaded my edited pix on some stock photo site. I really think A.I. used some of my edited pix to create many woman pix for example on Pinterest. So i like the title of this video.

[@thatguy7085]

Well… just a change. You know now… nothing can be trusted online. Nothing.

[@LexLex-tn7by]

Lets talk about the Artificial Intelligence in the human body.

[@jameswilkinson150]

Yeah, but how many artists and photographers have they screwed over to get these?!… Sorry, I mean, how many websites have they "scraped", aka how many images have they basically stolen?!!! I think the answer is probably ALL of them.

[@Yattayatta]

And those artists in turn screwed over the inventor of the Camera and the people that improved photography before them you mean? Or it is ok for humans to look at art and get "inspired" (aka copy with minor tweaks) but not for AI to do so?

[@NoidoDev]

Are people making such comments part of a campaign, or is it the video recommendation algorithm?

[@jameswilkinson150]

@@Yattayatta nope, completely different analogy. You are referring to a type of technology, these A.I. basically hoover up AND USE ALL images then just spit out bastardised versions. It is blatant theft really. Nothing like the comparison between artists and photographers.

[@jameswilkinson150]

@@Yattayatta you clearly understand nothing about copyright. Inspiration is not actually using other peoples work. If you actually USE other people’s work (as these ai’s do) then that is copyright theft.

[@Yattayatta]

@@jameswilkinson150 But they don't use other peoples works, an AI can't remake an image it was trained on. It just has weights of parameters in it's "memory" not the actual image, just like if I ask you to draw the monalisa you could, but you wouldn't get it exact. Clearly you don't understand how the AI works?

[@TsarHare]

safe as well sane

[@BikashDutta-n1o]

Are you using AI voice generator or real person's voice?

[@davidpacheco5501]

I really want to know this too

[@CERISTHEDEV]

Currently ai voice generators cant change tons mid speech so i would say its the real guys voice

[@tlatosmd7074]

@@CERISTHEDEV You're talking about TTS. STS aka voice conversion certainly can change pitch or tone.

[@CERISTHEDEV]

@@tlatosmd7074 my bad i completely forgot about sts

[@budekins542]

You need a computer with a Massive memory to do Flux but yes it is the new King of A.I picture generators.

[@NoidoDev]

Regular gaming PC.

[@ghost-user559]

Running it on a Mac Mini with 16 GB. You just do less steps and tile the encoding. If I offload the model from Ram, I can use the Neural Engine and Cpu and it uses only 1 GB at the very end if the render. It takes a while. Or you can do 20 steps Dev or 6 steps Schnell, and it uses 10 GB ram for the fp8 models. PC idk.

[@cbnewham5633]

Schnell doesn't require very much in the way of hardware. A decent laptop will run it.

[@ghost-user559]

@@cbnewham5633 You can also do Dev at 512 and then upscale like the old 1.5 workflows. It does really good even at 512 and it takes a fraction of the resources. Then you can decide how many steps based on your hardware. You can also use the embedded guidance to cut the time in half.

[@ghost-user559]

@@cbnewham5633 You can also run Dev at 512 pretty easily

[@rg-cb2wd]

th ep is gonna be so good oml

[@fonziebulldog5786]

Best of the best by this date ... Runway Gen-3 AI.

[@eyerock36]

2:15 Does everyone here know what the FBI has identified this symbol as representing? Hopefully a coincidence the company chose this symbol and not reflective of a hidden agenda or preference.

[@EricKingOfWithers]

They should totally ban AI generated images and videos because people are abusing it. They can still have ai chat bots but

[@TSOMTSOMTSOM]

Good old Flux. They are obviously able to create adverts to spam youtube too.

[@JOnline-im2wz]

Flux IS Stable Diffusion

[@ccvvxxbbbbxxvvcc7541]

Is it just me or does the face on Benjamin look a little more bitter than before? The problem with AI generation or randomization in general is that you can shuffle a deck of cards only so many times before you have unshuffled it completely. At what point can you say with certainty your AI isn't copying an already existing person or creating one from an alternate reality.

[@DJVARAO]

Flux takes 800s to create one image in my pc... but the results are impressive.

[@dishcleaner2]

I WANT IT TO BE INDISTINGUISHABLE FROM REALITY, EVEN BY A GAN. I WANT TO LIVE IN THE MATRIX AND TAKE THE BLUE PILL TODAY RIGHT NOW

[@Paul1239193]

AI stuff is going to be perfect and it will put in imperfections on purpose to make them look more human--more human than actual humans. Here we go...

[@YamashitaPhilippines]

where to download flux? i install pinokio

[@User-aplnk]

Is that named after Aeon Flux perhaps??

[@jelefaz]

Or the flux compensator in "Back to the future"

[@screenstory]

@@jelefaz Isn’t it called a “flux capacitor” in BTTF?

[@mikehobbs4280]

Believe nothing you hear, and half of what you see..

[@chopitblaze]

But we can already make anything look like real. So what if now AI will be able to do it? Photoshop masters were doing this for decades.

[@udiorockmeamadeus]

AI has a tendency to make stuff too perfect, and that is a sign its not real

[@TheCoolBrooo]

What's definition of reality

[@MichaelForbes-d4p]

Everything that actually exists and everything that actually has happened is happening or will happen.

[@martiddy]

Morpheus quote

[@yourlogicalnightmare1014]

Reality is what god looks like from a given perspective

[@BxPanda7]

"Everything you experience" Because it doesn't matter if you're the only one experiencing something, it still is real to you, otherwise your thoughts wouldn't count as real.

[@FoulPet]

Whatever our future AI masters tell us

[@senaprasena168]

nice... gonna check it when I got home.. wait.. where's the link?

[@johto]

AI narrator, right?

[@toututu2993]

These so called AI oversaturated market is a huge insult to humanities greatest achievement. Whether is about inventions, cultures, representations, ideas, values, histories, expressions, achievements, creativities, lifestyle, and countless more that these corporations have 0 clue nor respect of

[@udiorockmeamadeus]

FLUX can't draw a picture of Chuck Mangione..

[@cedricweber4141]

Your "Midjourney" Logo ist the Ideogram Logo (7:44)

[@torycrawford8552]

Don't be scared Resistance Is futile We are the BORG, Skynet is watching!

[@you-are-the-tharpsters]

You can’t do advertising with flux unless you ask the people who made it. People will surly be very responsible and respectful and not use this AI for revenge or deepfakes right?

[@trinidad111]

If you have this rookie crap, imagine what the government has

[@ShahrukhParvez]

Is this real voice or Ai generated voice?

[@IM2awsme]

The scary thing is I can't tell if this video is AI or not? 😅

[@SKGFindings]

“I visualise a time when we will be to robots what dogs are to humans, and I’m rooting for the machines.” -Claude Shannon

[@Verschiedenes]

And it‘s only the beginning…

[@DanielHedrick-w9h]

What is reality? Simulators?

[@barrynotlarry]

I have played with it in Nightcafe

[@justinwright33]

better than reality? what does that even mean?

[@NoidoDev]

@@justinwright33 Prettier?

[@CERISTHEDEV]

Its a therm that people have been using since the invention of 3d graphics now it just means something that looks like extrapollated version of a real thing

[@artofgarduno]

Better shadows better contrast more dynamic lighting. Basically the perfect photo that's is incredibly hard to do in real life.

[@rraaddff]

purdy soon we will not need movies we will be able to make are own