7:45 I just signed in and voices can no longer be cloned for free (which is evident at 7:53 where at the bottom of the features listed for the free plan, the phrase, "voice clones" is crossed out).
It's such a brilliant concept. I forgot to mention it in the video, but it almost reminds me of camera direction in storyboards. Fairly genius move here!
Yay, finally a major breakthrough making the focused object move in a controlled way. Motion brush is great for backgrounds. Now we need an AI editor that can handle both and also possibility with adding a layer of a focused object on top of a background image and merging them (light, style etc) in the process. 🤔
Yup, we’re almost compositing, or doing old cel school animation techniques in a way. I’ve seen some stuff on the horizon, like AI motion graphics ala After Effects, so I think there’s an infrastructure to bring everything together once it’s ready. We’re pretty close!
Just wanted to say that I absolutely love your channel. You somehow always cover all the AI developments that I’m interested in, and you do it in a professional (though not as professional as that AI version of your voice 😂) and friendly manner. Keep doing what you’re doing, and I hope your channel receives the success it so richly deserves. 🙏🏼❤️
Thank you so much!! And 100% will do! I've got some interesting videos coming up pretty soon-- a little outside the box, but I think you'll find it pretty interesting! Don't worry though, the "usual" show will continue as well! And man, I wanna hire my own AI Voice to do some work for me!
"...that offness... is professionalism." I really hope to use that excellent line one day, lol. Great video, mate (I'm aware that it's a number of months old as I write this, but it came in clutch). I've been into AI-assisted creation (mostly gaussian splatting and making stock photos with Stable D), just beginning to get into audio and video with AI tools and this helped a ton. Cheers!
Meh, I feel like I've been waiting for AI video to get good now for about a year, and it seems to be going nowhere fast. There still doesn't feel like there's much movement in the videos and you have literally no control over what you get, I'm going to call this technology dead on arrival
I'd caution the DOA...I've totally stopped doing that just because you never know what's right around the corner. I'll agree, Video has not had its "Midjourney Moment" -- and in those terms, what we're playing with here is maybe early v3? But, MJ v3 was super cool and exciting at the time. It'll be baby steps and then suddenly one big leap. I think that'll happen this year.
You're a gold-mine of information - I do my own research, but you do great work and I'm always coming back -- isn't this space so entertaining? Constantly upgrading with insane potential. The decentralization of entertainment is here.
I think we’ll see it for sure, if not sooner than 3 months. I always kind of suspect that when we see one of these big breakthroughs that others are working on something similar. Great minds and all. There’s also an interesting point that they announced it so early? Almost like they’re trying to beat someone to the punch?
Totally. I think no matter what, this shows the idea works at the very least. If there's one thing I've noticed covering all these white papers, it's that they're ALL reading/tracking them. So, I wouldn't be surprised if the underlying ideas weren't incorporated into other models soon enough. There's also a part of me that keeps thinking there must be a reason they released this paper so early (2 to 3 months out?)-- I'm wondering if they weren't trying to beat someone to the punch.
Good stuff, as always! Would love to see some more on magnific - I remember you were pretty enthusiastic about it, but I don't see people reporting on it these days.
Same! I still use it all the time, in fact this thumbnail was a lo-res screenshot run through Magnific. When they’ve got a new update, I’ll cover it, but I always have to brace myself for the onslaught of “price is out of control” comments. Hundreds of them.
Not sure if you remember me mentioning this in one of your past videos but, Magnific - you shot yourself in the foot with the business plan. They basically encouraged competition to decimate them, which will happen in the coming few (human) months. The Boximator tool could really change things. Great video. Thanks. 💯 edit: Oh, invideo is awesome, but the voice clone was not quite right imo. Better than many of the inbuilt ones, but if the main idea is to personalise, I would not have known in advance that was you. It'll get better I'm sure.
I do remember that! And...yeah, Javi was for sure stirring up a hornets next there. On the one hand, I totally agree with you, on the other I always appreciate that kind of business showmanship-- that kind of Orson Wells type bravado. I mean...It is (if nothing else) entertaining!
Haha, someone else in the comments was like, "you should-- then when you see someone pretty, you can pull it out like a magic trick!" super slick. ...although it would probably get a restraining order on me. And my wife wouldn't be too happy about the whole thing. haha
Spidey blew me away! Haha, I and keep thinking: In 2 to 3 months, we'll be saying the same about the NEXT thing! This is going to be a big year for sure!
Wow, very cool stuff. A lot of good teases here and stuff to look forward to. Wish the Leonardo tool is available for pro users. That looks more satisfying and affordable than Magnific.
Oh, that’s true! I mean, that’s where you can really get a “have the man in the red shirt sneeze, and the man in the blue shirt run away” kind of prompt. Hmmm. Maybe by next year?
I am wondering why Boximator even requires human box and path drawing. Text-to-Image tools or even mutlimodal LLMs can identify objects in pictures and create boxes for them. So I guess it's more a question of the processing pipeline. The prompt "Man lifting arm drinking coffee." would have to first identify the object of the movement ("coffee"), then identify the coffee in the picture and draw a box around it, then identify the movement path by general direction ("lifting" = up) and identifying the target of the movement ("drinking" = mouth), again finding the position of the mouth. All of thast should be solvable by current Generative AI.
While I do think the boxes serve a purpose, I kinda get your point-- particularly considering the size of those soft boxes. The boxes likely serve as an additional level of visual control, likely more for the human user than the bot. I can see it evolving out to something like PS's magic selection tool at some point, where you just mouse hover over an area and it auto selects the character or limb. But man, combine boxes with Sora? Wheeeeeeee...
You need to get sponsored, you have a great mastery of grabbing the intent of a program and displaying it's capabilities within a short span. This is a difficult skill to come by
Oh man, thank you so much! It's funny, I once got a comment for some tutorial I was doing that said something like: "They didn't have a manual, so you just made one for them..." I still laugh about that one.
Seriously…tough to keep up! I drop a video and the next day something massive breaks! But. It is interesting to think when something like boxes hit Sora as well. All this tech will eventually converge. And that is wild to think about!
It’s brilliant. As just a UI for movement? Brilliant. I’ve been saying, you often use boxes in storyboards to indicate camera movement, so this fits right in!
I currently use Magnific, but I can’t wait for Leo. I have a paid sub there too. How do I access the Alpha? Or is it just best to wait for the release?
First off, can Boxinator get Will Smith to eat spaghetti? And second, the main problem I have with these previews of new AIs shows the crazy cool stuff they can do is that by the time they are released, the already existing ones (Pika, Gen2, etc.) have a good chance of having these same features or even better. So it looks now like they are so much better than the competition, but the competition isn't showing us what they are going to have available to us in 2 to 3 months. They're showing us the great things already available to us.
That’s a very good point, and I have taken note of the fact that Pika has been very quiet lately. Like, curiously quiet. My bet is they aren’t napping and we just might see Will Smith eating spaghetti coming soon!
(1:38) "...it takes a lot of rerolls to dial something in." Think about the massive amount of time and electricity that is begin wasted for the simple reason that the Python crowd refuses to rethink their strategy by dividing images into layers in order to manipulate individual subjects rather than their current "whatever so long as I don't have to type more code" workflow.
Their AI model understand 3D space accurately right? so it means that maybe in the future it might even be able to be experienced in real time in virtual reality HMD with 3D depth (of another AI model that was trained to split single images to two different angles to stream it in the VR HMD), how crazy is that going to be? to be able to type some text describing a fantastic scene and be able to then right away experience it in full 3D and even be able to move our head/rotate it in its space ETC? the future potential of this tech is so crazy and incredible.
Totally! No more slow-mo boxing matches that look like they're taking place underwater! Although, now I'm thinking about a Rocky remake, only they're scuba divers...hmmm, that seems like a good AI Movie!
I'm sure they're working on it, or something similar, right now! I'm curious about what Pika has been up to. They've been pretty quiet lately...Like, suspiciously quiet!
Yeah, so that was a disaster from trying to train too many styles with Midjourney, BUT-- it actually came out cool! Nothing like what I was aiming for, but that's ok!
If this is really how it works when it is publicly available…. OMG…. hopefully Midjourney‘s (soon?🤷🏻♂️) text to video will have something like it in store 😎
Video AI = ho hum 😔 Creative upscalers = 🔥 That said, I'm all in favor of advancements in AI video. What about cubes instead of boxes....... in this way you could hypothetically control the X, Y, Z axis. Now that'd be something to raise an eyebrow over!
So, I forgot to mention it, but it almost kind of feels like how you indicate camera motion in storyboards. I think it's a really cool idea for an interface, and really beats any current system of "Prompt and Pray!"
Thanks to InVideo for Sponsoring this video, please give them some love here: invideo.io/i/TheoreticallyMedia
The only youtube channel of AI-related I liked is yours ..👍
7:45 I just signed in and voices can no longer be cloned for free (which is evident at 7:53 where at the bottom of the features listed for the free plan, the phrase, "voice clones" is crossed out).
I just saw Open AI's SORA ......... finally I'm starting to be a believer in video AI
Boximator! 🤯.. simply wow!! A real glimpse of the sort of tools we’re going to see coming throughout this year.
It's such a brilliant concept. I forgot to mention it in the video, but it almost reminds me of camera direction in storyboards. Fairly genius move here!
@@TheoreticallyMedia But I can not wait 3 months for that...Are they trying to destroy me?
Yay, finally a major breakthrough making the focused object move in a controlled way. Motion brush is great for backgrounds. Now we need an AI editor that can handle both and also possibility with adding a layer of a focused object on top of a background image and merging them (light, style etc) in the process. 🤔
Yup, we’re almost compositing, or doing old cel school animation techniques in a way.
I’ve seen some stuff on the horizon, like AI motion graphics ala After Effects, so I think there’s an infrastructure to bring everything together once it’s ready. We’re pretty close!
Just wanted to say that I absolutely love your channel. You somehow always cover all the AI developments that I’m interested in, and you do it in a professional (though not as professional as that AI version of your voice 😂) and friendly manner.
Keep doing what you’re doing, and I hope your channel receives the success it so richly deserves. 🙏🏼❤️
Thank you so much!! And 100% will do! I've got some interesting videos coming up pretty soon-- a little outside the box, but I think you'll find it pretty interesting!
Don't worry though, the "usual" show will continue as well!
And man, I wanna hire my own AI Voice to do some work for me!
"...that offness... is professionalism."
I really hope to use that excellent line one day, lol.
Great video, mate (I'm aware that it's a number of months old as I write this, but it came in clutch). I've been into AI-assisted creation (mostly gaussian splatting and making stock photos with Stable D), just beginning to get into audio and video with AI tools and this helped a ton. Cheers!
Wow I can’t wait to try Boxanimator! Thanks for the previews as always great stuff!
Meh, I feel like I've been waiting for AI video to get good now for about a year, and it seems to be going nowhere fast. There still doesn't feel like there's much movement in the videos and you have literally no control over what you get, I'm going to call this technology dead on arrival
I'd caution the DOA...I've totally stopped doing that just because you never know what's right around the corner. I'll agree, Video has not had its "Midjourney Moment" -- and in those terms, what we're playing with here is maybe early v3? But, MJ v3 was super cool and exciting at the time.
It'll be baby steps and then suddenly one big leap. I think that'll happen this year.
Appreciate it!!
You're a gold-mine of information - I do my own research, but you do great work and I'm always coming back -- isn't this space so entertaining? Constantly upgrading with insane potential. The decentralization of entertainment is here.
Keep doing your regular voice - it’s warm & relatable. Use Tim-clone for reading the news or when you need gravitas (Tim-clone’s voice is deeper).
Animation using boxes looks amazing - I hope we will be able to use this soon.
I think we’ll see it for sure, if not sooner than 3 months. I always kind of suspect that when we see one of these big breakthroughs that others are working on something similar. Great minds and all.
There’s also an interesting point that they announced it so early? Almost like they’re trying to beat someone to the punch?
This looks pretty wicked. If it works as well as they are showing, it would totally change AI film making. 🎬
Totally. I think no matter what, this shows the idea works at the very least. If there's one thing I've noticed covering all these white papers, it's that they're ALL reading/tracking them. So, I wouldn't be surprised if the underlying ideas weren't incorporated into other models soon enough.
There's also a part of me that keeps thinking there must be a reason they released this paper so early (2 to 3 months out?)-- I'm wondering if they weren't trying to beat someone to the punch.
Boximator looks amazing. It’s giving me Peter Gabriel Sledgehammer stop motion vibes 😍
+10 for the Peter Gabriel reference! And totally agreed! Boximator looks pretty brilliant as a way to direct video motion!
Good stuff, as always! Would love to see some more on magnific - I remember you were pretty enthusiastic about it, but I don't see people reporting on it these days.
Same! I still use it all the time, in fact this thumbnail was a lo-res screenshot run through Magnific.
When they’ve got a new update, I’ll cover it, but I always have to brace myself for the onslaught of “price is out of control” comments.
Hundreds of them.
Not sure if you remember me mentioning this in one of your past videos but, Magnific - you shot yourself in the foot with the business plan. They basically encouraged competition to decimate them, which will happen in the coming few (human) months. The Boximator tool could really change things. Great video. Thanks. 💯 edit: Oh, invideo is awesome, but the voice clone was not quite right imo. Better than many of the inbuilt ones, but if the main idea is to personalise, I would not have known in advance that was you. It'll get better I'm sure.
I do remember that! And...yeah, Javi was for sure stirring up a hornets next there. On the one hand, I totally agree with you, on the other I always appreciate that kind of business showmanship-- that kind of Orson Wells type bravado. I mean...It is (if nothing else) entertaining!
Who carries a rose in the pocket? YEEEs shenanigans :) as always man, love your sense of humor 😄🤣
Haha, someone else in the comments was like, "you should-- then when you see someone pretty, you can pull it out like a magic trick!" super slick.
...although it would probably get a restraining order on me. And my wife wouldn't be too happy about the whole thing. haha
Boximator looks worth waiting 2-3 months to try it! Amazing results. Especially Spiderman.
Spidey blew me away! Haha, I and keep thinking: In 2 to 3 months, we'll be saying the same about the NEXT thing! This is going to be a big year for sure!
'That offness was professionalism" lol - My friend we love you professional or otherwise
Haha, thank you for suffering all my "ums"!!
Is that a rose in your pocket or are ya happy to see me?! Great stuff Tim, Boximator is RAD!
haha, thank you so much! I presume you saw this: ruclips.net/video/oDHS_04AcGU/видео.htmlsi=lM3KGO7zmbxi5K4G INSANNNNE!
I have now!@@TheoreticallyMedia
Wow, very cool stuff. A lot of good teases here and stuff to look forward to. Wish the Leonardo tool is available for pro users. That looks more satisfying and affordable than Magnific.
It should be out soon! This was just the Alpha for some of us that test the newest features. It'll be in your hands within a week or so, I'm sure!
That box thing is sick you could easily automate that with LLM vision models depending on how thats set up
Oh, that’s true! I mean, that’s where you can really get a “have the man in the red shirt sneeze, and the man in the blue shirt run away” kind of prompt.
Hmmm. Maybe by next year?
I simply cannot wait! Great vid!@@TheoreticallyMedia
Always keep a plastic rose up your sleeve so you can do a magic trick for a pretty girl in the grocery store.
You have just won the "Smooth Move of the Month" award!
I am wondering why Boximator even requires human box and path drawing. Text-to-Image tools or even mutlimodal LLMs can identify objects in pictures and create boxes for them. So I guess it's more a question of the processing pipeline. The prompt "Man lifting arm drinking coffee." would have to first identify the object of the movement ("coffee"), then identify the coffee in the picture and draw a box around it, then identify the movement path by general direction ("lifting" = up) and identifying the target of the movement ("drinking" = mouth), again finding the position of the mouth. All of thast should be solvable by current Generative AI.
While I do think the boxes serve a purpose, I kinda get your point-- particularly considering the size of those soft boxes. The boxes likely serve as an additional level of visual control, likely more for the human user than the bot. I can see it evolving out to something like PS's magic selection tool at some point, where you just mouse hover over an area and it auto selects the character or limb.
But man, combine boxes with Sora? Wheeeeeeee...
Sweet! Always got the latest and greatest info. Thanks for great content!
Haha, it can be a grind, but a comment like yours makes it all worth it!
I repet myself Tim. Thabks for everything. Thanks to you and to Wolfie. Keep it up!
That’s good company there! Thank you for supporting and commenting!
You need to get sponsored, you have a great mastery of grabbing the intent of a program and displaying it's capabilities within a short span. This is a difficult skill to come by
Oh man, thank you so much! It's funny, I once got a comment for some tutorial I was doing that said something like: "They didn't have a manual, so you just made one for them..."
I still laugh about that one.
Boxinator is awesome, wonder why they don't add frames for smoothing?
Pretty early in, I’m sure it’ll look a lot better as it comes closer to release.
Appreciate the updates! The serpent tree was a happy lil accident
I dug it! It's funny how AI hallucinations work-- Once I saw it, I can't UNsee it from the original!
Note that Invideo's voice cloner is NOT available on the free plan :(
Oh, it isn’t? My bad- I thought it was! Uhhhh, whoops?
And then came SORA...
Seriously…tough to keep up! I drop a video and the next day something massive breaks!
But. It is interesting to think when something like boxes hit Sora as well. All this tech will eventually converge. And that is wild to think about!
Has Boximator or its equivalent become available on any platform yet? Thank you.
Thank you, Tim 🤗
If Sam can raise 7 trillion... that's 3x more money than exists in America.
It's totally insane! It hit the cutting room floor, but I had a whole bit on how that's James Bond Super Villain Money!
Boxes look SO cool!
It’s brilliant. As just a UI for movement? Brilliant. I’ve been saying, you often use boxes in storyboards to indicate camera movement, so this fits right in!
invideo shows free voice cloning on their pricing tab.
but when i sign up it is crossed out?
anyway i can test this before buying a subscription?
I think you can, you should be at least? I’ll check in with them on that.
Damn this stuff is crazy. When are you making us a music video with suno ai 3 and this stuff?
Haha, as soon as I get the chance to breathe!!
I currently use Magnific, but I can’t wait for Leo. I have a paid sub there too. How do I access the Alpha? Or is it just best to wait for the release?
I think you have to be part of the test group right now for Leo, but really, it’s like days away at this point. It’s really good as well!
@@TheoreticallyMedia Ok thank you! Keep up the good work. I learn a lot from your videos.
First off, can Boxinator get Will Smith to eat spaghetti? And second, the main problem I have with these previews of new AIs shows the crazy cool stuff they can do is that by the time they are released, the already existing ones (Pika, Gen2, etc.) have a good chance of having these same features or even better. So it looks now like they are so much better than the competition, but the competition isn't showing us what they are going to have available to us in 2 to 3 months. They're showing us the great things already available to us.
That’s a very good point, and I have taken note of the fact that Pika has been very quiet lately. Like, curiously quiet.
My bet is they aren’t napping and we just might see Will Smith eating spaghetti coming soon!
Can you please provide the link of Box Amator?
Ah, I forgot to add the link in the description. I’ll pop that on as soon as I get back to my machine!
@@TheoreticallyMedia Thanks dear! You're doing amazing. ❤❤
Man, you are the best! :)
(1:38) "...it takes a lot of rerolls to dial something in." Think about the massive amount of time and electricity that is begin wasted for the simple reason that the Python crowd refuses to rethink their strategy by dividing images into layers in order to manipulate individual subjects rather than their current "whatever so long as I don't have to type more code" workflow.
Lets go Tim 🔥
Ayyyyyy! Good to see you here!! How's your project coming? Can't wait to check it out!
Thanks Tim.
1000%! And thank you for watching!
Their AI model understand 3D space accurately right? so it means that maybe in the future it might even be able to be experienced in real time in virtual reality HMD with 3D depth (of another AI model that was trained to split single images to two different angles to stream it in the VR HMD), how crazy is that going to be? to be able to type some text describing a fantastic scene and be able to then right away experience it in full 3D and even be able to move our head/rotate it in its space ETC? the future potential of this tech is so crazy and incredible.
Awww yeah! One step closer to making some fight scenes! 🤜🤛
Totally! No more slow-mo boxing matches that look like they're taking place underwater! Although, now I'm thinking about a Rocky remake, only they're scuba divers...hmmm, that seems like a good AI Movie!
Good job 👍🏽
Thank you so much!!
Great. Now AI can drink coffee. We're done.
Haha, here I am thinking: “well me and AI are best friends now!”
Thank you
Absolutely my pleasure!
How long will take Runway implement this??
I'm sure they're working on it, or something similar, right now! I'm curious about what Pika has been up to. They've been pretty quiet lately...Like, suspiciously quiet!
@@TheoreticallyMedia well now everything changes again with SORA 💥🫠
Sora:Jajaja
Can you change item in box
You can change the location of the box, but you can't change the image. You'd have to inpaint elsewhere and re-input the image.
10:08
I really like this image!
Yeah, so that was a disaster from trying to train too many styles with Midjourney, BUT-- it actually came out cool! Nothing like what I was aiming for, but that's ok!
Love that channel
Thank you so much!! Really appreciate it!
Thank-you for your videos brother!
1000%!! Thank you so much for the support as well!
No voice clone of the free option
Ahhhh. That’s my bad! I’ll see if I can edit that out. Sorry!
If this is really how it works when it is publicly available…. OMG…. hopefully Midjourney‘s (soon?🤷🏻♂️) text to video will have something like it in store 😎
I can't WAIT to see what MJ has in store for us with Video. That's going to be SUPER interesting.
when and where we can start with that ?
Amazing.
It is seriously nuts!
ecstatic for Boximator. Pika and Runway better keep up or they'll get left behind. Because Boximator is 100x better
I keep saying that Pika has been suspiciously quiet. I don’t think they’re napping, I think they’re quietly building something big.
@4:23 seal kissed by a rose
Haha, now I just hear Seal saying: BAYYYY-BE!
Video AI = ho hum 😔
Creative upscalers = 🔥
That said, I'm all in favor of advancements in AI video. What about cubes instead of boxes....... in this way you could hypothetically control the X, Y, Z axis. Now that'd be something to raise an eyebrow over!
So, I forgot to mention it, but it almost kind of feels like how you indicate camera motion in storyboards. I think it's a really cool idea for an interface, and really beats any current system of "Prompt and Pray!"
Make video on open ai sora text to video model it's absolutely revolution in ai video
Gotcha! ruclips.net/video/oDHS_04AcGU/видео.htmlsi=lM3KGO7zmbxi5K4G
It's been a bust week!
This will be huge *if true!
We’ll see in a few months, but I’m fairly confident we’ll not only see it, but it’ll look a lot better by then!
Amazing! E
It's SUPER Wild!!
Professionalism 😂
AI me clearly doesn't show up to work drunk! So, it has that going for it!
not enough options,
On the boxes?
Me me me 😅
Same! Same! Same!