Man, you rock! Thank you for all this valuable info. I'm a music producer-recording artist, simply trying to make some AI videos for my album during these rough times, and this is beyond helpful!
Unreal, Tao! One thing I have noticed for the data set is that selfies don’t work very well. They seem to cause my head to be bigger than it should or output weird angles. I’ve noticed that pretty much all of your photos are taken from eye level and by another person.
Yes, I'd definitely recommend getting someone else to take pictures of you. Selfies have slight distortions in them. You could also set a photo timer on your phone to take the photos yourself.
Super cool. As you probably know, Vidu keeps the likeness very well of images uploaded as character references when prompted to perform some more complex actions, though their 1080p upscaled videos are not nearly as smooth as Kling, as the initial lower framerate is not smoothed out enough when upscaled in their system (as of now). I'm betting that MiniMax will be the King of animating images of yourself when they finally release their image to video (until their competitors match MiniMax in animating more complex actions in real time speed).
Tao, have you ever tried adding another consistent character alongside you to create a story? I tried adding one I trained, but what Replicate does is blend the two characters together. If you figure out how to do it, please make another amazing tutorial.
Can I ask u brotha, do you know the Ai tool that you can apply any videos face motion to a still image? I did this a few months ago but can not recall what platform offered this?
I really love kling 1.5 but the cost is way too high for what I do. I have runway ml unlimited and its pretty awesome. Just yesterday i created over 20 clips just to get around 10 that are actually usable. With kling, that would've cost around around $11, maybe less since kling 1.5 is really good in quality and understanding prompts but still... too expensive
I use both Kling and Runway, I find it gives me the most flexibility in what I want to generate. Although the Runway unlimited plan does give better value
Yes, the LoRA will learn what is "in common" in between your training images. This can range from people, objects, even visual styles, colors and lighting conditions
one more thing, if we want our avatar to have specific clothes or background , can we upload reference images for that? i still think Midjourney gives more freedom in this because of inpainting feature
Thanks for this. QUESTION: How do you have multiple trained models in one shot? More specifically, is there a way to train several individual models, then bring them all together into one shot?
hey Tao! I´m having great results here! Thanks for that! Is there a way I can find my previous generations? Every time I generate new images it seems that the previous one are lost, even though it shows the quantity of runs I made, I cannot find the images! Thank you
Tao just wondering for Leonardo Ai which is the best model to use? Maybe you could do a video on it and compare each model with eachother to the same prompt? Thanks Tao
I haven't used leonardo for a while but I remember switching between different models like the anime or cinematic photography one depending on what I wanted to generate.
Hey thx man, even if I know stuff, I always find little things to add to my AI knowledge :) Cheers. Question about the Pro Voice Clone, do you recommend it? How good it is? Did you record a specific audio for that or used your audio library (from youtube videos). Thx again!
hi tao, this is incredible, great video :D question please, how do i use runpod with this app if my macbook pro is too old to handle this application and require cloud gpu? thanks
Hi Tao, your videos are truly outstanding! The level of information and the step by step explanations are priceless. Actually I want to make my main actor look younger and I’m struggling to find a proper way to do it. I shot a musicvideo where the 50 years old actor should look like he’s 30 years old. Do you have an idea how to do that? Kind regards
could you make a klingai guide on "image to video"? i'm having a hard time, the videos don't come out right or i don't understand how you have to explain the prompt. for example i use a picture of a blue car, in the prompt i use "blue car explodes" ....and nothing happens.....the video doesn't modify the image of the car...even in some results the explosion happens but without modifying the car....i've already spent like 1500 credits with simple examples like that and klingai doesn't seem to understand the prompt.
I plan on it 👍. Although I'd say the more you want to "transform" the original image, the harder it will be for Kling. You can also turn up the "creativity-relevance" slider to improve prompt adherence
@@taoprompts I have already modified "creativity-relevance" without any results. I think I'll stick with Minimax. Ai, it does an almost "perfect" job with "Image to Video" using the same prompts and the same image of the car as Klingai. Minimax does transform it, explode it, etc. if it manages to modify the car or whatever it uses.
@@taoprompts Very welcome! True, true. I did experiment a bit with that using Vidu, a few weeks ago, uploading an image with two characters, and was able to generate new scenes with those two characters. But the upscaled videos weren't looking very smooth to me, hehe.
We humorously upload our faces into the database, unaware that the underlying motive is to harvest our facial recognition data for future exploitation, when global subjugation becomes a reality
I completely agree. Also, social media sites harvest both our psychological and physical profiles. Eventually all of this data will be weaponised against us.
Man, you rock! Thank you for all this valuable info. I'm a music producer-recording artist, simply trying to make some AI videos for my album during these rough times, and this is beyond helpful!
It's awesome that you're using this for your music videos!
Unreal, Tao! One thing I have noticed for the data set is that selfies don’t work very well. They seem to cause my head to be bigger than it should or output weird angles. I’ve noticed that pretty much all of your photos are taken from eye level and by another person.
Yes, I'd definitely recommend getting someone else to take pictures of you. Selfies have slight distortions in them. You could also set a photo timer on your phone to take the photos yourself.
Thank you for taking the time to do these. VERY helpful!
For sure, I have a lot of fun playing around with what these tools can do!
Super cool. As you probably know, Vidu keeps the likeness very well of images uploaded as character references when prompted to perform some more complex actions, though their 1080p upscaled videos are not nearly as smooth as Kling, as the initial lower framerate is not smoothed out enough when upscaled in their system (as of now). I'm betting that MiniMax will be the King of animating images of yourself when they finally release their image to video (until their competitors match MiniMax in animating more complex actions in real time speed).
Yeah, I'm also super excited for Minimax. I'll definitely be trying out their image to video this week.
@@taoprompts Heck yeah! I'm a bit envious of those few people who got to test it out last week for a few minutes. 🤣
WONDERFUL!! Thank you so much Tao. I'm going to try it...ALL your videos have helped so much!!
Good work bro
As usual fantastic🤘😝🤘
Thank you!
Tao, have you ever tried adding another consistent character alongside you to create a story? I tried adding one I trained, but what Replicate does is blend the two characters together. If you figure out how to do it, please make another amazing tutorial.
I will try it!
Great work! 🙂
Thanks John 🙏
Pretty dope. Thanks for this video.
damn , next level stuff ! as always
Thank you, I'm going to spend a lot of time trying out real voice AI animations 👍
Thank you for sharing this 🙏
Thank you Tao .
Amazinnnnnggg!! 🎉
Thanks man! I'm loving this new lip sync feature
Can I ask u brotha, do you know the Ai tool that you can apply any videos face motion to a still image? I did this a few months ago but can not recall what platform offered this?
music videos about to go crazy
For sure, people are already using Ai for music videos
Thank you! great helpful content!
Thank a lot dude
you are the best guy, thank you
I appreciate that man, I've loving all these new Ai updates 👍
Great work! thnx
you are great! much love from germany
Thank you so much
💯I hope you give this a try, it's pretty cool tech
nice work
Thank you 👍
Fantastic my friend!
Try video-to-video later.
Good job!
Thank you!
I really love kling 1.5 but the cost is way too high for what I do. I have runway ml unlimited and its pretty awesome.
Just yesterday i created over 20 clips just to get around 10 that are actually usable. With kling, that would've cost around around $11, maybe less since kling 1.5 is really good in quality and understanding prompts but still... too expensive
What's the censorship like on runway? I'm not talking noods or pron. But sum reject chesty women or booty shorts?
I agree Kling is expensive but Kling had one month only $28 for 8000 credits but after one month $80.
@dasberlinlex I actually Send them an email asking If they can consider an limited Plan like runaway
I use both Kling and Runway, I find it gives me the most flexibility in what I want to generate. Although the Runway unlimited plan does give better value
Awesome , thanks for sharing.
That was awesome. 💥💥💥
Tao, With this method, can we train a model on products like car or perfume?
Yes, the LoRA will learn what is "in common" in between your training images. This can range from people, objects, even visual styles, colors and lighting conditions
great:)
Thanks, always glad to see you in the comments!
one more thing, if we want our avatar to have specific clothes or background , can we upload reference images for that? i still think Midjourney gives more freedom in this because of inpainting feature
In your training images if you are wearing the same clothing, the Ai will learn to generate that
Wow much easier then comfy
Yeah, it's nice to have these easy to use interfaces
Minimax can do image to video now
Thanks for this. QUESTION: How do you have multiple trained models in one shot? More specifically, is there a way to train several individual models, then bring them all together into one shot?
There is an add "extra lora" function when you're doing the sampling
hey Tao! I´m having great results here! Thanks for that! Is there a way I can find my previous generations? Every time I generate new images it seems that the previous one are lost, even though it shows the quantity of runs I made, I cannot find the images! Thank you
I found it, nevermind.. 😜
Tao just wondering for Leonardo Ai which is the best model to use? Maybe you could do a video on it and compare each model with eachother to the same prompt? Thanks Tao
I haven't used leonardo for a while but I remember switching between different models like the anime or cinematic photography one depending on what I wanted to generate.
Hey thx man, even if I know stuff, I always find little things to add to my AI knowledge :) Cheers. Question about the Pro Voice Clone, do you recommend it? How good it is? Did you record a specific audio for that or used your audio library (from youtube videos). Thx again!
The voice clone worked really well for me. I have a lot of recordings of myself speaking into a microphone so I just used that.
5:50 is it better to upload a AI generated picture of yourself, or just an actual picture of yourself, honest question.
Both will work well, I used Ai pictures to showcase ai generated scenes
@@taoprompts thanks!
hi tao, this is incredible, great video :D question please, how do i use runpod with this app if my macbook pro is too old to handle this application and require cloud gpu? thanks
Thanks, glad you liked this technique! I haven't used runpod myself, so idk how that would work.
Hi Tao, your videos are truly outstanding! The level of information and the step by step explanations are priceless. Actually I want to make my main actor look younger and I’m struggling to find a proper way to do it. I shot a musicvideo where the 50 years old actor should look like he’s 30 years old. Do you have an idea how to do that?
Kind regards
You could try blending the older picture with a photo of a younger man inside Midjourney (using character reference).
@@taoprompts good idea. Thanks for the reply! Actually I saw FaceFusion 3.0 and will play around with that.
Keep going Tao! Greetings from Germany.
could you make a klingai guide on "image to video"? i'm having a hard time, the videos don't come out right or i don't understand how you have to explain the prompt. for example i use a picture of a blue car, in the prompt i use "blue car explodes" ....and nothing happens.....the video doesn't modify the image of the car...even in some results the explosion happens but without modifying the car....i've already spent like 1500 credits with simple examples like that and klingai doesn't seem to understand the prompt.
I plan on it 👍. Although I'd say the more you want to "transform" the original image, the harder it will be for Kling. You can also turn up the "creativity-relevance" slider to improve prompt adherence
@@taoprompts I have already modified "creativity-relevance" without any results. I think I'll stick with Minimax. Ai, it does an almost "perfect" job with "Image to Video" using the same prompts and the same image of the car as Klingai. Minimax does transform it, explode it, etc. if it manages to modify the car or whatever it uses.
Running or breakdancing? 6:34 🧐
So cool
Heres an idea, placing yourself in famous movie scenes :p
That's a great video idea, I may try to put myself in squid games when it comes out 👍
What is the total pricing for training the model to creating a image
$2-4
@@taoprompts thx
👍🏾👍🏾YES!!! This is what I'm talkin' 'bout. 😄 Master Tao, thanks Brother.
I appreciate that man! Now we just need to be able to do multiple characters
@@taoprompts Very welcome! True, true. I did experiment a bit with that using Vidu, a few weeks ago, uploading an image with two characters, and was able to generate new scenes with those two characters. But the upscaled videos weren't looking very smooth to me, hehe.
how many images can be generated with that model any limit plz tell me sir
As many as you want. You pay for how much you use
Hi, Kling AI is taking a lot of time to make an videos out of the image, Any solution is here ? Tried multiple times
the paid plan is faster, although Kling is one of the slower platforms
what about the expressions . they are serious or straight faced
If you prompt for the person is smiling or angry it will animate that. But it does sometimes make the mouth larger
Thanks Tao,one more question:how much one model cost?
About $2-3 depending on the settings you use
@@taoprompts got it,many thanks!!!
which gpu should you use on replicate
I just use the default one. As long as there is enough memory on the GPU it will run, the difference will be speed of generations
😢
Please help me with any fully free AItool that I can use to generate consistent images for story.
big fan of your work since months, here is the my first comment though :)
It is now asking for payment method. I cant even figure where to look.
Hey, great to hear you've liked the guides!
In the dashboard there should be a place to manage the billing.
Cool video haha alot of leaches here hahaa
Thanks, I've been using the new lip sync features a lot recently 👍
Someone please ai Please tell me the tool with which I can create my consistent characters
You can use Midjourney or Leonardo Ai, although they are paid services.
Hey hailuo ai got image to video you should talk about it
Thanks for letting me know, I'll be testing that this week for sure
클링 비디오 생성이 너무 오래 걸리거나 안됨. 무료 이용자가 그런가 아님 원래 문제가 있나 모르겠네요. 불안해서 결재하고 싶어도 못하겠음. 뭔가 믿음이 안감
It's much faster for paid users
video's subject aside, you are beautiful!
Thank you, I appreciate the compliment!
We humorously upload our faces into the database, unaware that the underlying motive is to harvest our facial recognition data for future exploitation, when global subjugation becomes a reality
I completely agree. Also, social media sites harvest both our psychological and physical profiles. Eventually all of this data will be weaponised against us.
That's why I never put my face out there like this. If I want to be in a movie I'll just create an AI image that closely resembles myself.
Maybe so!
You can generate any Hollywood actor without image reference… the actors are the first ones 😂
Don’t flatter yourself! You are not famous enough for others to want to use your images! 😂🤣
I want to do Muhammad ali vs Mike Tyson in a boxing match title fight upload their images to the generator It will be a brutal war
Would be much better than Jake Paul vs Tyson 😂
kling is taking a hole day to generate things nowadays....
any free ways ??
Matt Wolfe has a $10 bonus here in his video description: th-cam.com/video/_rjto4ix3rA/w-d-xo.htmlsi=3eijmXM1ni1zNquy
still... too expensive
Bro it is free .
Cool but no free.
Gotta invest.
You’re spending thousands of $ on all the things you show. Sponsored or just rich.
It’s less than most people spend on their daily starbucks mocha lattes
your the best!.
Thanks, I appreciate that 🙏