Amazing - thank you Tao! I was on Luma from 1 to 3:15 this morning trying to get some videos generated, but the workflow was long. But I did notice this camera motion control...I had the same results and feeling that it still needs work. Thank you so much again!!
this is the usual inconsistencies we are al still facing among all these tools.. soon som day someone will make it through the next level. You have done a great honest job!
Thanks! I'm glad to see them trying to stay competitive with the other Ai video tools, hopefully this makes the other video generators also add in some sort of camera motion control
awesome updates -- I'm on Luma and started playing with Kling. Kling gets the motion nearly spot on -- e.g. on some vid generators I got a race car moving backwards (image to vid) while Kling got it right & even the steering wheel direction/motion matches
I've found that Kling can animate the largest variety of motions with the highest quality right now, other Ai generators can do decent motions but with a more narrow variety.
Thank you for updating. I think this new update from Luma is great, although there're still some limitations. However, based on the price and the quality of the videos created, I think Kling is still a better choice.
It's good that Luma is making an effort to try and add useful features, even if it's not super good. I hope this will motivate the other platforms like Kling to also update their camera motion controls
Zoom is where the image gets bigger/smaller because the field of view is changing. Push and pull is when the image gets bigger/smaller because the camera is moving. When the camera is moving objects will shift based on paralax effects, but during zoom everything is the same relative to everyhing else.
Thanks for these experiments! My hunch / question is whether these are possibly just text suggestions-basically autocompletions-more than actual new capabilities. Have you tried comparing the effectiveness of these commands vs. plain language? E.g. "The camera zooms in on a statue" vs "camera push in, statue," or "The camera rotates around a statue" vs "camera orbit right, statue"? That would be informative…
I think you're right that these are essentially text prompt suggestions. It's possible they slightly updated their model with refined data with camera motions labels attached to them. Plain language also works for panning and push/pull, also rotation for bird's eye views.
I'm struggling with keeping images still let alone adding motion! I watched your video about the loop feature (which was great thank you) but using terms like fixed static shot, fixed camera, tripod shot don't seem to work... do you have any other tips that might work?
Tao, do you have any issues with Kling? After I stopped using the Pro account, the process stopped at 99%. I started it 24 hours ago, and it’s still pending. I’m wondering if this is a bug or if they are trying to push people to purchase the paid account.
Same here, many processes stopped at 99% for days. They always give excuses about overuse and huge traffic, while there are free unlimited AI generation websites like flux pro art, almost real time and no stupid excuses like all other fraud AI websites. They will only keep saying garbage trash for all of time and space just so people keep subscribing again and again. Maximum AI tools have become like that.
Some days on the free plan you'll get 3 or 4 that go through, and then I have had some prompts that go to 99% and just never produce anything. They are still sitting in my queue waiting. I am going to start paying for Kling because I really like it alot.
I am searching for a ai website that let for a example a woman in a picture walk in a video, with prompt ai. Which one is the best for this ? Because many prompts dont work
@@taoprompts to me the move up is like a crane. From that little animation, it looks like what they want to do is go up and into a overhead shot. It would be great if it worked. I’m having a heckuva time trying to get time lapses using an existing image. It’s really hit or miss any advice.
@@mdominic For the timelapse they showed of the aging person they used keyframes for that, which would probably also work best for other scenarios like landscapes.
@@taoprompts when I provide just an image and give it commands for timelapse, it will work sometimes and other times it won’t. It’s early days still. In six months everything will be different. I’m curious to see what Sora is like.
The best way would be go create reference images first with an Image Generator and then animate then with Ai Video. It will struggle with multiple people interacting though, Ai video isn't quite good enough to do that effectively right now
You're correct. But it should move closer to the subject like it is on a dolly instead of like a zoom lens. He was getting the opposite. The point is, he didn't not get the desired result.
Just realized that at this point all these AI companies are basically inventing dumb useless features to take everyone's money as soon as possible until the hype about these AI's fades out.
Amazing - thank you Tao! I was on Luma from 1 to 3:15 this morning trying to get some videos generated, but the workflow was long. But I did notice this camera motion control...I had the same results and feeling that it still needs work. Thank you so much again!!
Sounds like you're working hard! I saw some other people get good results with the orbit motions, but I wasn't able to replicate them myself yet.
@@taoprompts Okay, please let us know when you do? Thank you Tao!
this is the usual inconsistencies we are al still facing among all these tools.. soon som day someone will make it through the next level. You have done a great honest job!
Thanks! I'm glad to see them trying to stay competitive with the other Ai video tools, hopefully this makes the other video generators also add in some sort of camera motion control
Thank you for all the effort required to test this app. Great video…you really help people interested in making cinematic quality videos.
Thanks! The camera motions really important for using their app, I hope they spend some more time refining it
it took me to the waitlist , but great video thank u
Thank you for the Great updates !!
I got you covered man, Luma has consistently released pretty innovative features even though they have a lot of competition right now
For motion scene, I use Kling only. I think it’s the best now for subject motion 😅
Kling is one of the best for image-to-video right now 👍
awesome updates -- I'm on Luma and started playing with Kling. Kling gets the motion nearly spot on -- e.g. on some vid generators I got a race car moving backwards (image to vid) while Kling got it right & even the steering wheel direction/motion matches
I've found that Kling can animate the largest variety of motions with the highest quality right now, other Ai generators can do decent motions but with a more narrow variety.
Thank you for updating. I think this new update from Luma is great, although there're still some limitations. However, based on the price and the quality of the videos created, I think Kling is still a better choice.
It's good that Luma is making an effort to try and add useful features, even if it's not super good. I hope this will motivate the other platforms like Kling to also update their camera motion controls
Zoom is where the image gets bigger/smaller because the field of view is changing. Push and pull is when the image gets bigger/smaller because the camera is moving. When the camera is moving objects will shift based on paralax effects, but during zoom everything is the same relative to everyhing else.
Thanks for the explanation, I wonder why they went with push and pull instead of zoom effects.
@@taopromptsI think they should have both! Maybe there is less training data for zoom since it's more rare in movies.
Thanks for these experiments! My hunch / question is whether these are possibly just text suggestions-basically autocompletions-more than actual new capabilities. Have you tried comparing the effectiveness of these commands vs. plain language? E.g. "The camera zooms in on a statue" vs "camera push in, statue," or "The camera rotates around a statue" vs "camera orbit right, statue"? That would be informative…
I think you're right that these are essentially text prompt suggestions. It's possible they slightly updated their model with refined data with camera motions labels attached to them.
Plain language also works for panning and push/pull, also rotation for bird's eye views.
Thanks man this really saved me a lot of time from using it or not and I think it’s still not ready so I’ll wait a bit longer before giving it a try
They do have a free version with 30 generations per month, although it will be very slow compared to the paid version.
I'm struggling with keeping images still let alone adding motion! I watched your video about the loop feature (which was great thank you) but using terms like fixed static shot, fixed camera, tripod shot don't seem to work... do you have any other tips that might work?
Did you try turning off the "enhance prompt" checkbox? That can help reduce the camera motions to get a static shot
Very helpful. Thank you! 💫
The good part they added this feature and now the others will do so ... In time will work. Anyone know is opener will have bugs, issues.
For sure, I'm glad they added this. Even though it's a bit inconsistent, it's a good start.
nice and amazing
Which prompt for a fix camera with no mouvement ?
😢
Try turning off the "enhance prompt" box and add words like "static camera" to the prompt
Tao, do you have any issues with Kling? After I stopped using the Pro account, the process stopped at 99%. I started it 24 hours ago, and it’s still pending. I’m wondering if this is a bug or if they are trying to push people to purchase the paid account.
Hey I think that's probably just because paid accounts have priority. A free account is going to be at the end of the queue unfortunately.
Same here, many processes stopped at 99% for days. They always give excuses about overuse and huge traffic, while there are free unlimited AI generation websites like flux pro art, almost real time and no stupid excuses like all other fraud AI websites. They will only keep saying garbage trash for all of time and space just so people keep subscribing again and again. Maximum AI tools have become like that.
Some days on the free plan you'll get 3 or 4 that go through, and then I have had some prompts that go to 99% and just never produce anything. They are still sitting in my queue waiting. I am going to start paying for Kling because I really like it alot.
🔥🔥🔥
I am searching for a ai website that let for a example a woman in a picture walk in a video, with prompt ai. Which one is the best for this ? Because many prompts dont work
Kling or Runway are pretty at animating people walking
I had the same experience with orbit and crane.
Crane is an interesting one, I wonder what they had in mind when they added that in
@@taoprompts to me the move up is like a crane. From that little animation, it looks like what they want to do is go up and into a overhead shot. It would be great if it worked. I’m having a heckuva time trying to get time lapses using an existing image. It’s really hit or miss any advice.
@@mdominic For the timelapse they showed of the aging person they used keyframes for that, which would probably also work best for other scenarios like landscapes.
@@taoprompts when I provide just an image and give it commands for timelapse, it will work sometimes and other times it won’t. It’s early days still. In six months everything will be different. I’m curious to see what Sora is like.
How to make still/static camera for rain video?
a prompt like this works for me "rain at a creek, static camera"
is videos created by this website is non copyrighted for youtube ?
commercial use is okay with a paid plan
Luma or kling Ai ??
I personally prefer Kling
One day there will be video inpaint. Or select the area you want to move 😀
You have brush in runway gen, can select an area to make it move
@@djmsr yea payed 🥲
*paid
@@djmsr my credits for runway are done. Dont want to make a new account to get free credits again
I've seen the sneak peek for that inpainting feature! I'm looking forward to it 👍
Kling still the King 👑
Kling is my favorite right now 👍
andddd thats why im not renewing my luma subscription just yet😂
It's good to see them adding in updates even though there are still some issues right now, let's see if they can improve on this
Why i didn't receive un notification ?
They typically post any updates on twitter/X
Can teach how to create fighting scene using AI? 😅
The best way would be go create reference images first with an Image Generator and then animate then with Ai Video. It will struggle with multiple people interacting though, Ai video isn't quite good enough to do that effectively right now
Push is not zoom.
You're correct. But it should move closer to the subject like it is on a dolly instead of like a zoom lens. He was getting the opposite. The point is, he didn't not get the desired result.
Infinite wait list period, weeks of queue waiting time, extremely expensive and inefficient. 💀
It is pretty slow for free users, they will prioritize paid users in the queues
Just realized that at this point all these AI companies are basically inventing dumb useless features to take everyone's money as soon as possible until the hype about these AI's fades out.
Yeah useless
So watch other channel 😊