AI News: Amazing New Tools You Can Use NOW!
Вставка
- Опубліковано 29 чер 2024
- Here's the AI news you might have missed this week. Check out Future Tools: futuretools.io/
Discover More From Me:
🛠️ Explore thousands of AI Tools: futuretools.io/
📰 Weekly Newsletter: www.futuretools.io/newsletter
🎙️ The Next Wave Podcast: @TheNextWavePod
😊 Discord Community: futuretools.io/discord
❌ Follow me on X: x.com/mreflow
🧵 Follow me on Instagram: / mr.eflow
Resources From Today's Video:
Luma AI Video Generator: venturebeat.com/ai/luma-ai-de...
Sora by OpenAI: openai.com/index/sora/
Luma Dream Machine Creations: lumalabs.ai/dream-machine/cre...
Pika Image-to-Video Model: x.com/pika_labs/status/180059...
Stable Diffusion 3: stability.ai/news/stable-diff...
Stable Diffusion 3 Medium: huggingface.co/stabilityai/st...
Max Woolf on Stable Diffusion 3: / 1800921802765717754
Leonardo.Ai Homepage: app.leonardo.ai/
Midjourney Model Personalization: x.com/midjourney/status/18006...
Google GenType: labs.google/gentype
Suno Audio Input Feature: / 1800932487633207599
Adobe AI Training Policy: www.theverge.com/2024/6/10/24...
Apple Intelligence Introduction: www.apple.com/newsroom/2024/0...
OpenAI and Apple Partnership: openai.com/index/openai-and-a...
Apple Paying OpenAI: appleinsider.com/articles/24/...
Market Cap Rankings: companiesmarketcap.com/
Marques Brownlee on Apple Data: x.com/MKBHD/status/1800340862...
TechCrunch on Elon Musk: techcrunch.com/2024/06/10/elo...
Musk Drops OpenAI Suit: www.cnbc.com/2024/06/11/elon-...
OpenAI Welcomes New Execs: openai.com/index/openai-welco...
Microsoft Copilot Pro Update: www.windowslatest.com/2024/06...
Hello Qwen2: qwenlm.github.io/blog/qwen2/
Photographer Disqualified from Contest: petapixel.com/2024/06/12/phot...
Sponsorship/Media Inquiries: tally.so/r/nrBVlp
#AINews #AITools #ArtificialIntelligence
Time Stamps:
0:00 Intro
0:20 Luma Dream Machine
6:25 Pika Labs Update
6:59 Stable Diffusion 3
10:06 Leonardo's Custom Model
12:35 MidJourney Personalization
14:22 Google Gentype
15:31 Suno Audio Input
19:11 Adobe Terms of Service
19:40 Apple WWDC Recap
25:30 OpenAI and Apple Partnership
26:31 Elon Doesn't Like The Apple Deal
28:00 Elon Drops OpenAI Lawsuit
28:42 OpenAI New C-Level Execs
29:14 Microsoft Removes GPTs
29:49 Qwen 2
30:55 Real Photo Wins AI Art Contest
31:29 Final Thoughts For The Week - Наука та технологія
Your ability to break down complex AI developments into understandable segments is truly a talent. Looking forward to more updates like this!
def he uses AI to make it understandable.
He’s promoting something that will literally end humanity for billions
@@nickc1425that's the plan
@@MoMoGammerOfficial 100%
@@nickc1425that’s what your grandparents said with the internet
I don't think the only concern about Adobe's TOS was training on AI. I think many were also concerned about Adobe saying it can review all your content and block it if they deemed it inappropriate and that they could share your work with any 3rd party tool.
And not giving you the option to access your content unless you accepted their terms.
Why is blocking what they deem inappropriate concerning? Yeah, it sucks, but it's their shit and there are tons of other options. Be glad they don't have a monopoly on it.
@@nickgirdwood3082You have to be trolling
Probably a bit of each
@@nickgirdwood3082 yup, you have got to be trolling or extremely stupid. Choose one.
Excellent content! Love your upbeat delivery of news on AI!
i have a friend who is working in a nursery. a while ago i told her about chatgpt and said how interesting it is. she did not think it something useful as also she was not techy minded.
now she says she can't do without it.. she creates fun activities, stories, questionnaries, games and etc with it. all her struggles for creativity to please kids are over, she says. she now know more activities than ever before
Do people actually have that much difficulty thinking of things without software training-wheels for their brain? It's funny, but back in the day my Mother would just either say, 'I'm busy, entertain yourself', or she'd come up with something on the spot for us to do. What you're describing sounds like a lack of the ability to think.
@@dmwalker24 i think you are oversimplfying her role.
nurseries are competitive environments. you need to be useful there. being creative and funny all day, everyday is not easy. everyone has expectations. managers, principals, parents, kids.
they also have some personal matters.
you can try but you can't give all your emotional energy to your work, everyday.
i'am an 80s kid. i had to do things otherwise teacher could give me a smack.
they can't do this anymore. times are rapidly changing.
imagine your co-worker is coming up with more solutioons than you.
how you will respond?
@@dmwalker24 you can do surprisingly lots of different things whit it not many if any humans are a expert in all fields but ai can be altough it is wrong sometimes.
@@dmwalker24 Ironic that you made a comment like that on the comment section of a video about AI. Yet, here you are.
Seven hours? You kids. When the first program on the Amiga computer generated a cool looking ray traced graphic it took 30 days. And you prayed the power didn’t shut off.
So apply that to everything, go ride a horse in rush hour traffic
30 days! Luxury. When I were a kid, our computer was a potato and a few bits of copper wire. We dreamt about 30days for rendering!
Tell that to kids these days, and they won't believe you.
Amiga! I still have my A1200. It's my favorite computer by far.
@@chekote The Amiga was so cool and so far ahead of its time. I had (first) an Amiga 1000. I had gone back to college to get a degree in computer science. When I would talk about it in that IBMMS-DOS yellow-green type on a black screen world, people would say things like why do you need color? Or Stereo sound? That had no idea.
I wish I still had my old Amigas.
@@billbuyers8683 LOL
When you played the music, your persona went into a different bpm and key. It was quite lovely, like a bob ross of AI. You were being gentle like Mr Rogers and it was quite refreshing. The point of AI is that we are going to be able to slow down and only do the things that we really want to be doing with our lives and not be caught up in the hustle of what people call the real world. You are an incredible youtuber, and there is a potential that if you slow down and take advantage of that key in G you may get even more followers. You are great and wonderful Matt!
Bob Ross of AI? You ppl are soo we Todd did.
MUCH BETTER THUMBNAIL! TY FOR GIVING ME HOPE
These videos are my favorite where you show all the newest and coolest features that have come out and give demonstrations. The always take me hours to get through because I pause and try as I watch.
Thumbs up.. thanks Matt and team!
Recently discovered your channel, glad I did! Thank you for all your work, much appreciated!
I love your coverage, thank you!
Thank You Matt ..Love your latest content on AI .
The Leonardo image gen model is insane. 2nd favourite to Ideogram right now
Thanks for this video.
cpt4o Transcript:
- Talks about one of the most fun weeks in AI. 0:00
- Starts with the introduction of Luma AI and Dream Machine. 0:17
- Mentions competitors like Sora, Veo, Cling, Pika, and Runway. 0:25
- Talks about the generation times and issues faced with Dream Machine. 0:50
- Describes the initial 7-hour wait times and error outputs. 1:13
- Discusses text-to-video failures and shows some examples. 2:03
- Talks about Sora's text-to-video examples. 2:39
- Explains that image-to-video performs better. 3:53
- Shows some image-to-video examples. 4:10
- Explains how Luma excels at image-to-video. 5:06
- Mentions Pika Labs' image-to-video model updates. 6:27
- Announces the release of Stable Diffusion 3. 7:03
- Shows examples and tests of Stable Diffusion 3. 8:11
- Discusses Leonardo's new Phoenix model. 10:08
- Shows features and examples of Leonardo Phoenix. 11:01
- Talks about MidJourney's model personalization feature. 12:35
- Explains the process of creating images based on personal preferences in MidJourney. 13:12
- Provides information on Google's GenType tool. 14:24
- Shows examples of creating letters in different styles with GenType. 14:57
- Talks about Suno's new audio-based song creation feature. 15:31
- Shows examples of new audio features in Suno. 16:01
- Mentions Adobe's new terms of service stating they won't train AI on customer work. 19:09
- Gives a brief overview of AI features announced at Apple's WWDC event. 19:45
- Talks about AI integration in Apple's devices and new features. 20:22
- Explains new AI-based features added to Siri. 24:09
- Mentions Elon Musk's reaction to the Apple and OpenAI partnership. 26:29
- Talks about Elon Musk dropping his lawsuit against OpenAI. 28:05
- Provides information on OpenAI's new C-level executives. 28:47
- Mentions Microsoft's removal of GPT features in Copilot Pro. 29:17
- Discusses the performance of the new open-source model Quinn 2. 29:49
- Talks about a photographer winning an AI art contest with a real photo. 30:53
- Wraps up with a summary of AI tools and developments of the week. 31:29
Thanks you , lots of useful information in this video.
Thx so much Matt. I feel like I’m getting left behind while trying to keep up. I’m not sure how to jump back in?
Nice as always
We just discovered your channel. Great video, thanks for sharing that content 🙌
Great Vid, Matt! Thank you 😍
This Wolfe guy has the coolest job on the internetz. I'm so jealous! 😝
little tip with luma.. if you grab a screen shot of the last frame and reload it.. and use the same prompt it should carry on from where it finished.. though there may be a little jump i did this with 4 videos.. 3 of them where consistent then on the 4th it changed to a manga video.. but i am still impressed..
You should have tried the extended prompts that Leonardo generated in sd3 to have a fair model comparison.
Hello Matt, thanks for the extensive video review
Good stuff all around sir, good stuff.
🧐 Cant wait to see where we will be next year 🔥🔥
Great summary, thanks
Great as always!!!!!
Your video is outstanding, a truly great watch indeed.
i wish Goro Penguin could be introduced into Mortal Kombat 1, Netherrealm make it happen! I've used Leo for some time now, in fact, I believe it's because of you that I found it. In the beginning it was meh, but over time it has become a very powerful generator and now that I am a Creative Partner with them, I find myself using it as my go to for images. The amount of tools built into it as well as models, loras via elements, and features like Universal Upscaler has solidified it as my go to for image generation. Thanks again for another great video my friend, I'll always tune in for all the latest in AI tools.
My experience seems to be the opposite of yours on Dream Machine. Image to video always ignores my prompting and mostly just pans across the image. But straight text to video has been pretty impressive for me, although it gets confused if you prompt for multiple characters.
Turned you into a steve harvey news team 😂
thanks Matt!
amazing videos bro ..thanks !
Video Gen AI might struggle with fingers, but to be fair: when you dream and look at your fingers, they also look weird and wobble around. Looking at your fingers is actually a great way to check if you are dreaming or not. So even our brain video generator finds it hard to display fingers
if you dream and YOUR fingers wobble around, you should either reconsider if you arent AI or maybe have some condition haha. mine are perfectly fine. only thing i do is imagine i have 6 fingers on each hand BEFORE looking at them, to find out if im having a lucid dream or not.
i can understand if others fingers and faces are a mess, but yours? you got some weird dreams :P
Whenever I used copilot I never saw an ad or link to make gpts, that why it might not have got used, if it wasn't for this video I wouldn't even know it existed
Yay Phoenix is AMAZZIIING
I honestly love the honest title Matt, cheers
LEo AI text is pretty good except it puts in a lot of the prompt words into the picture not just the quoted words. Of course I had "advertising poster" in the prompt, so it may be taking the extra text as sub-text for the poster design.
I keep using SUNO almost daily it's really cool specially when it can effectively use all the languages for lyrics.
Thanks for the update, I feel plugged in. Who needs an AI implant to stay in touch when we have your videos?! I particularly love the [actual] flamingo story.
This was entertaining. I was wondering what was the response to your AI and robotics video? Are you going to do more?
Super hyped!
I tried Stable Diffusion 3 twice and it timed out both times. Do I have to stay on the page and wait for it? It's like waiting for grass to grow.
I just hope SD3 is good for training, that and tools like controlnet and ipadapter are what make it valuable to me.
No, SD3 is badly nerfed and peeps are saying it's not worth trying to finetune it, it's too gimped and censored to the point of stupidity.
It's terrible.
I can imagine an AI creating a short video on a thin flyer. You open it up, and it shows you other clips of things, such as for travel or tourist attractions. I don't know how far we are with thin bendable interfaces. But imagine such a thing, being unique to each person, but also true to the advertisement.
Printing a video seems like the real challenge 😂
17:40 - It seems like a lot of Suno songs have the same electronic sound, making them identifiable as Suno songs. I've seen lines lyrics in other people's songs that it also gave me in mine.
Thanks for these! I love AI and am trying to grow my channel with!
Crazy news! Mind-blowing news! OMG!!! Shock!
Literally every fucking title regarding ai news since 2023:
It would be interesting to see the leonardo prompts in stable diffusion3
Thanks Matt, we thought we would see our film in yours. Seems like it’s breaking the internet rn 😂
Thank you.
Thanks!
So good!
30:51 Great, but don’t let quality suffer. you should have added non-abbreviated test names to this slide. it is an easy LLM task- long time subscriber.
IMO, Luma is the best image to video i've tried so far. Even better than pika. This is the only one I've tried that doesn't keep morphing peoples faces and mostly keeps them consistent throughout the frames. Also, thanks for another great video!
The sound is much better then when i started watching like a year ago 🤙
Matt, here's what I would say would be best to focus on.
Which tools CAN BE used professionally (no rights issues) today.
In what context can they be used - what can and can't you enter as text (as far as intent). Obviously it has to be legal content, but regarding whether or not a prompt could even include a celebrity's name in the prompt (I'm assuming not... but people don't state it). Or someone says "somewhat close to Matt Wolfe, but younger, bettter looking" - can they still use it if the image doesn't really look that close to you?
Scrap all the drama around the companies - the people.
Just the tools...
Thank you for this video. Most of the news i didnt expected, as they are so close to me and i use these them a lot :)
Thank you for being so honest. I tried dream machine and I was so disturbed and disappointed. Love your honesty
Hi Matt. I am a music producer and have been wanting to try and make an AI music video for an artist for some time now. I check your videos to be up to date with the best tools for it but i still couldn't realize what the best text to video generator would be. One of the reasons is that the song is a very 80s dance beat and i am trying to generate dancers. Do you have any suggestions? Have you ever done anything like this? Thanks!
Just a thought on the links for your videos, maybe toss a section on your site that is searchable as well that has your daily video links in it? Might help tie the channel even closer to the site, just a thought.
1:37 consider it beta version . Solve for now , which engines are usable for multiengine use .
Is it me, or did everyone else really like the first and second songs about "stars are singing lullabies" that Suno Audio Input created?
Aside from the Gen-Z style, auto tone voice, it was pretty impressive.
It would be good if you could use an image generated model to make a story board with consistent characters then a image to video model that stitches all the pictures together with movement
Give it a few months 😄
Agree with you that SD3 results on the raw are quite underwhelming, I get it, it's just released and not yet model-refined but I expected more.
best Ai news reporter
The Suno update about the wavefile is kinda of weird. Once I download the wavefile stills sounds like an MP3 , they might just been doing a conversion from mp3 to wave which is not the same thing as creating the audio as wave and then letting download that actual file.
I actually agree with this, I think you're right.
Though just yesterday I've jumped over to Udio just to try it, sounds good but I don't like having to do things in 30 seconds chunks unless there's something I'm missing.
18:02 this is actually impressive
Guys you can learn AI beside news , best AI channel 😍
SUNO IS A BIT LAME AT VOICES BUT INSTRUMENTALS WOULD BE WILD WITH the complex specific audio words and some nice hq samples, i.e. abstract music is semi synth semi AI, totally nonphysical audio effects it;s pretty fascinating. abstract ambient idm melidic weird sfx etc... try it.
QUESTION! II just wrote the first draft of my first novel (no help from ai, yet) and I was wondering if chatgpt or maybe another you know of can proof read and fix tenses, so I can get to final draft easier.
Here you go, mate. Edit this with your content and specific instructions.
Proofreading Request: Tense Consistency Check
Document Title: [Insert Title Here]
Author(s): [Insert Author Name(s) Here]
Document Type: [e.g., Academic Paper, Novel, Business Report, Blog Post]
Total Word Count: [Insert Word Count]
Section(s) for Focus: [Specify if the entire document needs review or only certain sections]
Background Information: [Provide a brief background of the document, including its purpose and target audience. Mention any specific aspects that you think are relevant for understanding the content or context.]
Specific Instructions:
Overall Readability: Please assess the document for general fluency and coherence. Ensure that the ideas flow logically and that the language is clear and concise.
Tense Consistency:
Primary Tense Used: [Specify the primary tense intended for the document, e.g., past, present, future.]
Highlight and correct any instances where the verb tense does not match the primary tense designated for the document. This includes checking narrative consistency in literary texts, tense alignment in academic or research works, and appropriate tense usage in professional or business documents.
In sections meant to deviate from the primary tense (such as flashback scenes in a narrative or references to prior research in academic works), ensure these shifts are clear and logically justified within the context.
Contextual Tense Usage:
Identify and correct any mismatching tenses that disrupt the chronological flow or clarity. This includes ensuring that tense usage accurately reflects the timing of events, actions, and states described.
Grammar and Syntax: While the focus is on tense consistency, please also correct any grammatical errors encountered during the review, paying close attention to subject-verb agreement, the use of passive vs. active voice where appropriate, and sentence structure.
Suggestions for Improvement: Offer suggestions for enhancing tense clarity and overall document coherence. This might include rephrasing sentences, restructuring paragraphs, or noting areas where additional detail or clarification is needed.
@@a_changedworld Wow! THanks!!
@basedcommodore No worries, good luck with your novel. I expect a mention at the front when it's published 🤣🤣🥲
@@a_changedworld np :) its called 'Jacked In' by Michael North
15:32 Interesting. note: Google doesn’t get fonts.Image generated letters should allow for proper kerning.
i like the Suno song!
Adobe says it has wrote itself an "any use" license for all your files and review all your files( to keep your art and pc safe from cuss words and nudity). Then they say they wont use your personal files for training but can use it in any way that improves there business which includes selling and trading of your art, deleteing your art, morphing your art and of course using it as training data improves threir business. Them trying to clarify that we dont understand the words does not help. We just read that Adobe has made an agreement with itself to make itself a license to co-own and use your art. Therefor You have no personal art and to continue using your art, You must keep up with all your subscription bills
Thanks for the audio and video breakdowns again. The inclusion of the (cool) Apple announcements in a video about tools you can use NOW is a bit strange though… given nothing will be released until earliest end of year.
SD 3 can only be used for personal use; it's T+C are insane right now.
When I tested Luma it took some minutes, maybe they have a long queue now
That google font generator thing is 100% new, its gotta be
Why must you always use "A wolf howling at the moon" as your prompt, can't you use "Matt Wolfe howling at the moon" instead?" Hahaha! ❤❤❤
Suggestion, start with the parts that work well and leave the bad ones until the end.
It was a fun week, if only for the lulz. SD3 and Luma "MemeMachine" footage has been hilarious.
Just glad to see someone else that uses the word “janky”
Boom!
One of the most fundamental things about us humans is that we have a set amount of attention.
Give us Neuralink, great, lots of input to enhance our senses but we can only concentrate on one thing at a time.
We can easily be overloaded then go into a trance-like state with a glum look on our faces going ummm.....
Crash our cars, wee on our feet etc.
Summaries of what we need to know will be great but I wonder if that ummm factor will be shortened?
canvism is a fun tool to use as well
Thanks for another great video Matt. I too am a Suno Junkie.
Matt you need to listen to the Nvidia CEO last interview to get a sense of how we have be playing checkers but are about to switch to chess.
Ceviche Cheesecake are two words I never thought I'd see together.
I did the prompts you did and had mild success.
Im gonna try to refine the prompts to be specific and use image to video option.
I dont know if it beats SORA to be honest.
Whatever horsepower OpenAI has behind the scene is serious business.
This technology years from now is going to be insane.
This alone will end Hollywood
Hey Matt. Throw those guitar songs up on Spotify and show us how you made $ with them! PS: I love Suno. It 's what I use for all my UA-cam Channel A.I. videos/animations. Suno support has been very good to me as well.
I've been using ChatGPT to write game-specific intro and outro songs for my Let's Plays, and then I throw them into Suno a few times, take the song, make a video with some game footage, and bam, intro video without using a song I technically lost legal access to years ago. I have made a few outro videos with footage, but then I got lazy and started to just put it with a title card-type image for said game.
I even use it for adult games which I do Let's Plays for on the Hub. And then the Hub refuses to monetize them because they think I'm not in the video when my camera is on literally the entire time. I think their employees work with one hand at all times, which is why they're incompetent. Or it's because it's a Canadian company and they hire people right off the plane. I criticized them, saying they should get people instead of robots to review videos, and get the typical modern response of being told I'm being abusive. People seriously think being told to do their job properly is abusive.
i died lauphing while u were explaining the lumia morphes, weird hand and all tansparently.
I never manage to get Suno to sound like I'd want it but Udio does much better. Their upload feature works about the same.
Would love to see any of them come up with a model where you can hum a melody and tell it to keep the basic just of the melody but make it genre X.
I agree with you about having fun with music generation! I've spent 1300 tokens at Udio in 3 weeks.... 👀
Btw, did you know about the AI Song Contest? Submissions close at the end of the month!
udio rocks. they should expand to video. whatever their algo magic is would be awesome with video
What are you generating? I don't understand complaints against Suno. I've generated amazing songs with Suno. Are you trying to get something other than rap music?
@@nickgirdwood3082 epic cinematic soundtrack music is pretty fine on Suno, but metal sounds a bit off to me, especially the lyrics
Suno is king! 😤⚡️
Every week is the most crazy AI news. And it never works as advertised.
I feel like you dont understand Stable diffusion, also it was stable diffusion 3 not SDXL that you mentioned at 12 minutes 30 seconds. Stale diffusion is superior because its not like the other platforms, its open source you can combine it with controlnet, you can fine tune it5, you can use loras, perhaps checkout civitia....Stable diffusion is like the entire plate of food not just the side dish you comparing it to.
Did you see the article on Gizmodo about AI detectors causing job loss? Thoughts?
12:24 You said we got SD XL instead of SD3
Wish I hadn't already watched 10 videos on this stuff
(29:30) Maybe GPTs would have caught on if the people making them were actually paid for their HOURS and HOURS of work? Like somebody said they would do back in November?
With the way SD3 is trained, now you have to be an esay writer instead of prompt engineer.