Easy Guide To Ultra-Realistic AI Images (With Flux)
HTML-код
- Опубликовано: 10 сен 2024
- Playing around with FLUX.1 and LoRAs to get some ultra-realism.
Discover More From Me:
🛠️ Explore thousands of AI Tools: futuretools.io/
📰 Weekly Newsletter: www.futuretool...
🎙️ The Next Wave Podcast: / @thenextwavepod
😊 Discord Community: futuretools.io...
❌ Follow me on X: x.com/mreflow
🧵 Follow me on Instagram: / mr.eflow
Resources From Today's Video:
fal.ai/models/...
Let’s work together!
- For brand & sponsorship inquiries: tally.so/r/nrBVlp
- For other business inquiries: mattwolfe@smoothmedia.co
#AINews #AITools #ArtificialIntelligence
When using runwayml with an image, you should mostly be giving the prompt direction instead of describing the image. Like for your beard guy image it would be like 'man holding microphone with one hand, speaking to audience, other hand moving as he talks, shot using handheld camera'.
Great! Now use it as fake competition for OF models or findom and make a lot of 🤑🤑💸💸💸
@@emanuelmma2Won’t allow NSFW photos anyway so yeah good luck with that lmao
I created the video around the 2:23 mark (Ago-ra) though I failed to notice that her pinky was missing lol. I really enjoyed your video and am looking forward to watching many more! Thanks and keep up the great work!
The lighting on the ponytail guy is amazing. I would say it's real ten times out of ten.
He better than real .. and same goes for many of the ladies.
Well then you're not looking very hard
Things are getting crazy, and I’m loving it!
So glad you explained how to use Fal and everything. I can't keep up with all the new models that drop
For a beginner also say that your videos are really very friendly, thank you very much. Because of my professional needs and the high learning threshold of flux, I've been using mimicpc to run flux before, it can load the workflow directly, I just want to download the flux model, and it handles the details wonderfully, but after watching your video, I'm using mimicpc to run flux again finally have a different experience, it's like I'm starting to get started! I feel like I'm starting to get the hang of it.
You could try it in Kling, which is often awesome and much cheaper than Runway. I am thinking of dumping Runway and just using Luma, Kling and Pixverse (new Pixverse 2 is erratic but like Luma or Kling) with Pika for lip sync. If Runway doesn't get it the first time, it's like flushing money. I'm already assuming Sora is going to be priced for Hollywood and am no longer excited about it.
I feel like Luma is taking more and more time to process, yesterday I waited for 4 hours or so to get the video
Kling is great when it works, but it also wastes a lot of tries. And without a motion brush it makes directing real scenes impossible because you are just at the mercy of whatever it decides to do. Runways motion brush is currently the best thing about it. A combination is definitely the way to go at the moment.
I can't believe we have good looking AI images before GTA 6
I am so glad you explained you get $2.00 free on fal... Because I got freaked out "before" I saw this video they were going to send me a bill.... I realized I was racking a bill after I generated some images.
Love this short nerdy informational videos - great job Matt
That's 12 Billiom Parameters just imagine if it was 24 - 70 Billion
Love your videos, Matt. Thank you for keeping us on top of all things AI.
Great job on another awesome video, Matt! Keep them coming!
THANK YOU, MATT!
skin a little too shiny in still image: "nah, i don't like this"
whole hand fucked up in video: "This is actually really good!"
:D
Its getting realy freaky very fast!!!
Let's hope Mr angrypenguin will add Lora soon to his awesome tool
It's intriguing to see how AI can mimic the subtleties of texture and light in images. The mention of a 'plastic shininess' is spot-on-how can we refine this further to achieve more natural results?
This guide is super helpful! I’ve been experimenting with different tools lately, like CGDream, and it’s amazing how much more realistic the outputs can get with the right approach
To me it's only helpful if I can run it locally on my own machine.
Definitely super realistic. But when they made videos of it, those seemed clearly animated by AI to me.
i've been showing these kinds of things to family members, especially older ones who might not realize if they are being misled
Great stuff as always. Thanks Matt, we appreciate you. Go Padres!
Stability has severely screwed up the potential of SD3. Looks like inside job to me.
How are you able to generate the same image at @9:12 without using the seed no or referencing the image. Can you share the full prompt?
I think you should make more content (maybe even shorts) where you flame technology.. The first few seconds were hilarious
You are my no.1 source of tech news... But I didn't know you had a funny side, and since you're EXTREMELY informed, you would have very hot takes
Definitely impressive. However, Flux in ComfyUI is - at least on my Win11 16 Gb RAM pc - too slow to really be practical at over 15 minutes per roll on the dev model. Looking forward to how this develops. Thanks for the post!
Any place to train our own images on flux for generation?
Jeez SD3 looks like it was taken out of the oven to soon . Been giving Flux a shot on hugging and it works really well.
Hi Matt, i am flshed by Flux as well - but: I have very bad results showing real proms like actors etc. Can you confirm that? And any countermeasures? Thanks!
Thanks Matt!!!!
Hi. Great introduction. BTW is this how they are making the cat fashion walks and cat car wash videos that are flooding TikTok etc. ?
Yes it’s a good model 🎉good video Matt
Great video 💗💗
thanks for all your hard work
Nice!! Now use it as fake competition for OF models or findom and make a lot of 🤑🤑💸💸💸
Seems like it won't be far off before AI is used to make entire films, then ultra realistic video games, then eventually virtual reality sims
great video
With these improvements and that recent Deep Live Cam (deepfake live video!), how exactly can we, as a species, not question _everything_ . Thought it was scary a year ago. Great stuff Matt. Thanks as always. 👍
The images I generate in Flux are exceptionally realistic - except, on occasion, for the fingers. Invariably there is an extra digit that is impossible to remove. I don’t find this fault so prevalent in other models. Removing and replacing backgrounds in the Flux app is a joy, so fast compared to Photoshop. But, hey! Who could have dreamed this a year or two ago.
Hey, I am a day one follow and appreciate all your content so much! One thing I would wish and still think it comes one day is a tutorial how to use AI to make a website beginning to end without all these subscription services. Like real one where pay for domain name rent only
The first few images are not AI generated non-sense. It's the AI making a promise, that we will look like that, if we keep messing around.
🤡
How do u add these to be uncensored and not limited by credits on your own machine?
you were using schnell models mate. No waxy stuff in the DEV and DEV pro.
GLIF does NOT work. Everytime, it just gives error: Failed to call FAL API.
I've heard many people are running this on their local Nvidia gaming GPU's. Can you make a tutorial on how to do that instead? Having to pay services for just a few images isn't worth it when there's already free workflows if you have a GPU.
Is there somewhere you can upload a still image of a girl and say "Make her dance to K-Pop with music" ?
Where do I get flux? Download it?
since Flux will run locally (if you have a stout machine like you probably have for video editing your RUclips content), is there a reason you didn't simply download it and add the LORA that your examples were using?
Is there a link to download this realism Lora?
Why does it feel like the beginning with ai all messed up is the future lol. 😂
How we can create great prompts to give it to Flux ?
What happened to Playground AI? It had it all - now it's reduced to a shadow. Very concerned. Nobody in that community has commented. They just stopped publishing anything. What happened??
Any thoughts on the best apps / process for taking old photos and animating them (and/or colourizing them)?
Thank you
So, how close are we to AGI
making video games? Do we need AGI for that, and how close are we to achieving it? Can AI agents already reason, code, program,
script, and map? Can AI break down games, create art assets, and handle long-term planning? With better reasoning, could AI
eventually develop a game rather than just writing out ideas? Could
it also put those ideas into action? I wonder if ChatGPT-5 will be able to create games with agents or even remake old, closed-down
games like Dawn of the Dragons if all the artwork and data are available in the wiki.
There are more than 15,000 games releases every year and I suspect there will be some zero's after that when Ai creates games with the push of a button. Whatever you do, make it's unique and something that stands out among the million others, else it's game over for you.
That can possibly happen
@Noblific cool, I hope so. Thanks. I am so hoping ai or agi will recreate old games.
@user-lu7np4cg1t I'm not in it to make money only, hopefully recreate some old closed down games. Rpgs
@@kellymaxwell8468 ah yes, sorry... thought you meant to create your own game. agi will easily be able to recreate what's been done in the past. patience... it' coming. listen to the new lex fridman + elon musk podcast that was just released for more details. enjoy!
An easy guide to creating ultra-realistic AI images with Flux involves leveraging its advanced features for generating high-quality visuals. Follow the guide to master techniques for enhancing image detail and achieving lifelike results with ease.
What's the benefit of this advancement? What value does it bring?
dopepics AI fixes this. Ultra-realism with FLUX and LoRAs.
Прикольно, Боярский сменил шляпу)
Thanks for the great video as always. I have a question however. Can I run Flux (Schnell) in Invoke UI? Could you maybe test that at some point?
tangent question, but are there any decent offline/local solutions yet for image to video?
stable diffusion 1.5 seems to have the most resources for that, but the rendering takes a loooooong time.
12:50 did your hand just AI big?
Convinced the government have been using this AI for the last 30 years
Amazing
Ok that tan girl with the ninja turtle hands kinda has me feelin a little mixed up right now.
Mixed up?
I’m checking out the Fal Flux Pro Image to Image model and I am not encouraged
2:27 The girls right hand has only four fingers
ok thats exactly how many fingers we have
@@indian_crocodileRight Hand
Flux is great -- if you have a super-expensive graphics card. The rest of us will have to wait until something else to come along.
Don't know why but I was able to run full flux1 dev and schenel in comfyUI on my Alienware 4060 8Gb VRAM with 32Gb ram.
LETS GOO, ty i did not know that it was working already@JayZone80
@JayZone80 Thanks for the info
It available online, and the GPUs are standard gaming GPUs.
Okay so how we gonna know what is real or fake 😢
I didn't get U$2 in credits after signing up at Fal AI :(
I am running the flux dev model locally using the webui for stable diffusion. Can anyone share a screenshot of their local webui settings for great output? Thanks
Use comfy for better results
Haw to get Flux?
SD3 is Deep Floyd IF quality 😂
Some people are "concerned" again.
would've been better to use a new example indicating that it actually works
The video stuff still needs a lot more time to cook. Even the good models are still pretty bad and surreal looking. Flux is pretty damn remarkable, though, especially for an open source model. I'm really interested to see how it develops with fine-tuned models and LORAs over the coming months.
Disappointed you didn't test it with Will Smith eating spaghetti...
.. and no mention of the elephant in the room? ... Kling? ....
trying to use forge on pinokio but kind of struggling to use flux on it. it works quicker but have more buttons and settings and is overwhelming
After gen 3 alpha sora ai hype is over
Hi!
Has anyone tried to use an API on Fal?
in distant future there would be people looking like this and will wonder why they were doing this and when did it happened
strange creatures Ai 😅 😂😂😂
People who actually does the science won't like the way you are criticizing these models.
what an intro 😂😂
Plot twist - this whole video was AI generated.
How to know if its AI? Is it pixelated? If yes then its a real photo. Simple as that. We don't have good enough phones for sharp images, and they can't be blown up. Also if you zoom in and its not pixelated it AI.
The animated vids still looked a little AI generated.
plastic looking, shinny skin? what are you on about??, it looks perfect...
Yes but can you monetize them?
you can sell it to your granma 😂
No one is gonna buy AI generated 💩
It's opensource, so... yes.
@@Ǿœ0œǾ Not by itself but with literature blended in, yes.
@@whyc2094 who needs monetization when we are all grass eaters
can't wait for an AI that can actually generate full-body characters in a consistent manner. That would be really game changing!
Plot twist its just stable diffusion reskinned as paid a polished paid service 💩
If you don't use flux pro, your generations will be a mess and you must pay for the pro
😅
At what point can we no longer trust our own eyes? It’s unfortunate that these models can run locally. With cloud services, you could at least encode something to track its origin maybe?
So only governments, companies and criminals would have it. Hmm, no.
For the information you are providing. I find your video way too long. I am coming because you wrote "guide to ultra realistic ai image with flux" and I end up with one simple information (that i already knew) which is that you can use the Realism lora. 13 minutes for an information that could have been shared in 2 minutes.
Hello clown. 🤡🖐️
🤣 Jesus
lol