AI Video’s Morphing Problem Is Over!

I really like your videos, and I have new inspiration every time I watch them. Recently, I have been exploring the field of AI and found that PSYCHE AI is very interesting and I think you will be interested.

Lore maschine

That's hilarious. No matter what kind of noises he makes during those exercises, they might still kick him out of the gym.

😂

Lore Maschine🤩

We'll all end up with that huge database of imagination with; The Batman of each major cities, the Wonder Woman for each major cities...the Doomsday for each major cities and so on and so forth. It could say something about each citys quality of life when it comes to aging with senile dementia...but maybee not...🤨...

Great video!

Why has your voice changed?

as a vfx artist and ai enthusiast, i was hoping to not see this technology.

should we already preparing to fully dive in the vr simulation world? i think it would be possible much more faster then everyone thinks

VideoJAM looks amazing. Can't wait to see this technique applied more widely,!

Oh dear, I'm a bit muddled :D

Right but will we ever be able to use any of this?

this channel has a discord server?

the wipe on the windshield might be there only to showcase the before and after (as in any demo reel of vfx studio) ;)

What would be killer is if Hostinger’s AI could build you a site using a specific CMS (Wordpress, Drupal, Joomla, etc) and a third-party off-the-shelf template.

Imagine training a number of realistic humanoid robots on the best sporting events. Inputting as much video as possible for say, the 2004 World Series. The robots interpret every move, every throw, catch, play, etc. Then you sell tickets to reenactments of the games.

I would always call upscaling "creative," by definition. It's creating something from nothing. That requires making 'creative' choices on what to fill in the blanks with. 

It's been an interesting ride so far with things like Adobe Firefly's inpainting, for instance. You can delete a section and it fills it in with something that's consistent, plausible, and virtually undetectable. Then click the forward arrow and you'll get SOMETHING ELSE that is consistent, plausible, and virtually undetectable.

That hostinger section in the video feels too large compared to the other chapters…

video jam to comfyui ASAP, make it happen people! lol

You sound different in this video? Did you use AI to make it?

"Videojam" sounds like a cool and edgy animated movie from the 90s

My suggestion is to take the comic book output and be able to port each of the story cells into a motion creation AI.

So, the scene would come to life when the reader clicks on the story cell.

This way they don’t have to build out the video, just app it into the workflow. Cha Ching

Interesting and well presented as always with good news that I hadn't seen before the video, best of all is the thumbnail mastery :D

Anyone know of tools trained on 360° video? I would love to turn some of my nature and beach reels into sci-fi scenes

At 6min, it's clearly a vfx wipe to show that nothing changes except for the water.

keep the good work my guy love yo ;)

Kling 1.6 so far has been quite fantastic, especially when it comes to img2video .. also least censored one so far (besides open source models).

Hi Tim, really quick one, following all your vids but all is going so fast with platforms competing like seagulls over a mouldy hot dog : how did you do your little detective sequence at 0.13 ?

Loremachine is a good concept however they own any content you produce with their services and you can't take it to another publisher. Plus they don't protect your content and can publish it for all to see so your story could be replicated by others. I would prefer a private service where I can own or have full licence to use the output. It's likely there terms may not be enforceable however is someone else takes the content because AI can't own copyright. All a bit sketchy to me.

this is what I always tell to people who are impressed because "chinese are advancing so fast".. look at the ASAP paper names, they're all chinese, except for a couple og guys. chinese are not advancing fast recently, they have been in the game long time already.

these tests mean nothing if they only test it with scenarios they would have in the training data.. like a dude jogging.... come on.. I want to see an octopus swinging through a tropical jungle trees.

Hey Tim, have you covered Qwen already? they do text, image and video. Pretty impressive!

james brown on the one

the video inpainting is great, can speed up workflow. a good vfx artist will take the ok output of it and enhance and change it anyway

❤❤❤❤

The Twilight Zone like video you showed of U as the host or in the stories

Great Album in the background - The inner mounting flame ;-)

와. 이번에도 뉴스가 가득하네요. AI발전은 빠르고 다양한 기능들은 쏟아지네요. :)😊

TOPAZ - Wow they are finally going to make the Moon landing look real. At least later generations will believe like they are supposed to.

It's going to be a good year.

Until it can output 1 minute, allow character & scene consistency not there yet & so far looks like China will pull the plug first

Great video again, you're my 'go to' on AI news these day, thank you, love your videos, concise, informative and with a great sense of homour (from a fellow guitarist that also doesn't need to dance! 😄). I have gone with Runway recently as my AI tool, intrigued to hear about their slow movement, glad it's not just me seeing that, wish there was more detailed prompt advice to get more movement in subjects. Loved learning about the story telling software at the end and loved the train the robot to a prime Mike Tyson comment 😂

DynVFX seems similar to what PikAdditions is on the Pika site. Similar theory at least. Not sure if it could do a tsunami.

I am so glad someone thought of a repair toolkit for all the garbage outputs I’ve collected from Kling 1.5.

Just tonight, Kling 1.6 produced a gorgeous video of me walking sensually down a Victorian hallway with stone flooring (camera follows subject). I wore a red leather mini and red leather boots with heels like you wouldn’t believe. However at the very end, the last 4 seconds of a 10-second pro video, Kling made one of my hands hold on to a chair. The chair rest must have had super-glue on it, because my hand wouldn’t get unstuck and then I developed“elastic arms” afterwards. 70 credits wasted!! 😒😭😭

Maybe I should add as Negative Prompt: Anatomical elongation?

VEO2 is becoming SORA. By the time VEO2 comes out publicly, it will be outdated. By the time SORA came out, VEO2 and KLING v6 were the kings!