SORA Uses 10k GPUs This AI Uses Just 32 (Fast, Cheap, AI Video For ALL)

Рет қаралды 13,403

Ай бұрын

Check it out here:
higgsfield.ai/Blog
📧Join my newsletter
delightfuldesign.eo.page/w7tf5
---
👨‍🏫Check out my AI courses:
www.udemy.com/user/samson-vow...
---
💼Business inquiries
samsonvowles@gmail.com
---
--- My top resources:
📖 Grab My AI Secrets! Dive into my handpicked prompts e-book.
delightfuldesign.eo.page/6sw6r
🎨 Unleash Your Creativity! Experience Adobe’s cutting-edge AI tools.
tinyurl.com/2sj2kwk6
🤖 Boost Your Efficiency! Automate with ClickUp.
tinyurl.com/4p9yxew5
🎤 Find Your AI Voice! Clone your voice with precision.
elevenlabs.io/?via=b38559
🎥 Revamp Your Content! Edit shorts effortlessly with AI.
www.opus.pro/?via=7fcbf8
✍️ Supercharge Your Writing! Generate content the smart way.
www.jasper.ai/?fpr=samson
📹 AI-Powered Videos! Elevate your content with Kaiber.
kaiber.ai/?via=samson
📢 Ads Made Easy! Create compelling ads in a flash.
free-trial.adcreative.ai/2yui...
👤 Stand Out Online! Craft your unique AI avatar.
www.synthesia.io/?via=delight...
🎬 Level Up Your Channel! Get coached by AI on KZfaq growth.
vidiq.com/samsonvowles
🖥️ Present Like a Pro! AI-driven slides for impactful presentations.
slidesai.io/?via=samuel-vowles

Пікірлер: 48

@outled Ай бұрын

Clearly it is nowhere near as realistic as Sora.

@TheBann90 Ай бұрын

Sure it's a step down from Sora, but at just 1% of the compute, the implications of that makes this model at least as interesting as Sora. Imagine one or two more iterations of this model, and I bet it will beat Sora while still using way less compute.

@outled Ай бұрын

@@TheBann90 Absolutely, I felt the video was mistitled. Good call. :)

@ezeppelin1 Ай бұрын

Honestly even other service like Haiper and pixverse can do better, i'm not too sure about their computing request but they are definitely less demanding of Sora.

@outled Ай бұрын

The video has been appropriately titled now, cheers Samson :)

@lifemarketing9876 Ай бұрын

Yeah this not even close. Not impressed at all.

@Glowbox3D Ай бұрын

This is pretty cool, but it doesn't compare to Sora. There is far too much instant morphing happening with so many elements of an animation. It also feels much nosier. Still cool.

@newfineart Ай бұрын

Still enjoying Haiper which you recommended about a month ago. Higgsfield looks impressive, especially considering the size of it's training data and compute power

@dougdoesstuff_uk Ай бұрын

You can almost see the frame blending going on, reminds me of 'Optical Flow effect' in Film Editing Software.

@thispc2487 Ай бұрын

Venturing into storytelling and creative video making recently. VideoGPT quietly became my secret weapon, imparting a sense of professionalism to my content effortlessly.

@egofooart Ай бұрын

I’ve also been using VideoGPT. It can be buggy at times but I’m starting to find work arounds for it.

@dogecoinx3093 Ай бұрын

sora is HD video that shit is like 480p

@mwj5368 Ай бұрын

Thanks for another great video! Could I with one or a few clicks make a movie from my soft sci-fi trilogy with this? Also if you do this kind of thing about how much would it cost to have you make an animated movie based on my trilogy I wrote? ChaptGPT said it's very translatable to movie format because I write in a cinematic prose style, plus other people have said the same. I slowly chapter by chapter had ChatGPT analyze each chapter and the whole of each of the three books and the whole of the trilogy that took about six hours to complete and all very positive.

@bigglyguy8429 Ай бұрын

According to their blog "Safety is a critical consideration for us and we have built in safeguards to limit inappropriate content" So not much use then really.

@NLPprompter Ай бұрын

it say can analyze video? I'd love to know more about that.

@I-Dophler Ай бұрын

This software's advancements in AI video technology mark a significant leap forward. Its efficiency and realism, achieved with minimal resources, set new standards. The Diffuse app, in particular, showcases their dedication to making advanced video creation tools widely available, democratizing the process for everyone. This development is a monumental shift for content creators worldwide, heralding a fresh chapter in digital expression and social media content creation. Eager to witness the impact it will have!

@egofooart Ай бұрын

App Store says the diffuse app isn’t available in my country (USA) or region?

@aisamsonreal Ай бұрын

What do you think of this AI?

@KiteChaser Ай бұрын

Wouldn't it make sense for the video generating models to train on specific training videos and not edited internet content? OpenAI etc should create in-house videos to train their models for them to understand the physics of the real world, movement, light, etc. A video library of everything. If they train them on internet content and movies it wouldn't be able to create real videos. Imagine having your whole experience of the world based on children's picture books. You wouldn't be able to create real world content.

@Jeffrie_Baer 25 күн бұрын

Ok, here's a bar that I am raising for AI. Let's say I have a camera and capture many videos. Then I feed them into the prompt and say "Make the subject of the video eat, drink, smoke, etc." THEN I will be impressed. There is a reason that I say this.

@aisamsonreal 25 күн бұрын

This would be epic and I think it will be coming soon

@XD-cr3du Ай бұрын

It looks good, but I would not say it's the most realistic yet, Sora clearly has a leg up in terms of realism.

@TheFeedRocket Ай бұрын

I like it, this is just the start of competitors entering video, my bet is on MidJourney to really knock it out of the park before the end of the year. Sora is not going to be out until next year, anything is better than nothing. I would rather have a tool to make video at a lower cost even if it's lower resolution right now, if it's cheap enough I can run it multiple times without paying tons of money. Also I see a point when we will "upscale" these videos, clean them up, etc.. The cool thing is by next year even the current Sora will look like crap, remember where images where this time last year?

@arpvisuals Ай бұрын

I agree with you. Midjourney's realism is so good that realistically animating the images can be done at a lower cost. Runway ML can still help with small animations

@babbagebrassworks4278 Ай бұрын

These smaller may not be as good but this is home local level AI. Who needs an expensive server farm? Will it run on my Pi5?

@dwainmorris7854 Ай бұрын

How come AI developers are not working on a way to merge 3d game design with AI generators . You would finally have the stability of 3d character design that look the same from every camera angle that are easy to pose like a 3d model , mixed the the speed of AI generated realistic scenes and lighting

@evacuatedspace6946 Ай бұрын

Are they all working making these videos as near to real life instead of making them run for longer with consistency a bit too much though? 🙊😿

@LaBonitaGraphicsAnimated-ft2nu Ай бұрын

How come AI apps creaters don't merge with 3d game design with AI generators . You would finally have the stability of 3d character design that looks the same from every camera angle that are easy to pose like a 3d model , mixed the the speed of AI generated scenes and lighting .

@darkman237 Ай бұрын

Looks like the Dreamshaper model to my eye.

@budekins542 Ай бұрын

Quite good.

@georgwalt7978 Ай бұрын

why dont they train AI on the same anatomy models real painters use to train? this way no unproportionate modeling is possible

@thebudman1980 Ай бұрын

Seems to do better still images than some Ai models. Even with Sora they haven't figured out how to do hands, fingers toes and feet yet. Other ai's can't even do proper hands and fingers and feet and toes on still images. So after looking at a lot if ai beauty art is like don't look at their hands or feet. Lol

@rayfellers Ай бұрын

The makeup model's teeth look straighter and better than mine 😂 and those carping about this not being as good as Sora then please make a better one if you can which I sincerely doubt.

@avikdsaha 6 сағат бұрын

excellent clarity

@aisamsonreal Сағат бұрын

Glad to hear that

@soundsoflife8885 Ай бұрын

Granted this is early days, I think it's a start, but no where near Sora,

@Octoberisms Ай бұрын

More AI video options the better!

@PREMCHAND-ei2sx 13 күн бұрын

May i learn to make a vdo from sora. I have no knowledge to make a vdo. I need a mentor.

@aisamsonreal 13 күн бұрын

How can I help you?

@gumvue Ай бұрын

I just upload a movie trailer, its close to be real.. its text to video not SORA

@DarthTekin Ай бұрын

👍

@omegalord7139 Ай бұрын

Sora is better I think but this one also is good

@realmarsattacksagain Ай бұрын

17:25 Yeahhh ... Anatomically it's the lips... Even Sora thinks real anglo women have full botox upper lips. Actually looks like Ai found the worst lip botch job on the internet to copy. 💋💋💋🤯 Maybe its inspired by celebrity photos??? Oh, and I can't help but mention the freckles Ai is putting on women these days. Like freckles have a genetic bias for women??? Seriously less than .000001% of women have freckles. Other than that I think this is a great video pointing out Sora's near future competition. Just give it 30 days..