Meta is DOMINATING AI! We haven’t seen ANYTHING Like this!

Meta AI, has recently revealed its new generative model for text and images called CM3leon. CM3leon is a multimodal foundation model that can generate images based on text input, as well as captions based on image input. CM3leon is the first multimodal model trained with a recipe adapted from text-only language models, which makes it more efficient and versatile than previous methods. CM3leon also achieves state-of-the-art performance for text-to-image generation, outperforming Google’s Parti model. CM3leon is currently a research effort and it is not clear when it will be available to the public.

▼ Link(s) From Today’s Video:

-------------------------------------------------

▼ Extra Links of Interest:

-------------------------------------------------

Thanks for watching Matt Video Productions! I make all sorts of videos here on Youtube! Technology, Tutorials, and Reviews! Enjoy Your stay here, and subscribe!

All Suggestions, Thoughts And Comments Are Greatly Appreciated… Because I Actually Read Them.

-------------------------------------------------

[MattVidPro]

IMPORTANT NOTE: much like SDXL and other Open Source Models they are ANNOUNCED before the weights and such are released. The code and weights should be coming in the future!

[Pawel_Mrozek]

The fact that this model can identify objects and modify them precisely is a functional game changer. Even if it is slightly inferior in quality to midjurney, people will still love it in 90% of applications that midjurney cannot handle.

[Yipper]

3:00 ok now that is what ive been waiting for. An image generator you can iterate on rather than needing it to be exactly generated in one go. This should also be good for some small text-to video esk things I imagine, if you tell it to make a bunch of small tweaks.

[yorth]

You and AI Explained are truly my premiere sources for AI news. Keep it up, man. You're great!

[nahiddotai]

It's super awesome that Meta out of all companies are investing and expanding their AI models. With their models being open source, it's a huge win for developers! I'm particularly looking forward to how AI image generators improve over time with all these new AI models coming out.. I'm bored with simple text to image now. Thanks for keeping us in the loop, Matt

[IPutFishInAWashingMachine]

Never expected META of all companies to be winning the AI race.

[MohamedAnsary]

Wow, Meta's new CM3leon model is truly groundbreaking! Its ability to generate images based on text input and captions based on images is remarkable.

[Vpg]

Amazing!! Very excited what people will produce with this!

[CreationTVstation]

nice review! i'm stoked to try it out!

[CrystalBreakfast]

These control features are a lot less impressive when Stable Diffusion already has them, and SDXL will have them not long after the official public release of SDXL 1.0. Matt hasn't given Controlnet 1.1+ a try yet? It keeps getting better and better and that's still with SD1.5. Can only imagine what SDXL+Controlnet will be like.

[sinayagubi]

Did they reupload the Meta models that were removed from Hugging Face?

[richardglady]

While the video sailed over my head, I love the Master AI List. I have never seen anything like it. Thank you for sharing it and creating this video.

[chanpasadopolska]

Which licence does Meta use for their Open Source projects?

[cnu]

Bro, what is the progress on 3d model generation using you make a video on that

[littlegravitas]

Why are people surprised that Meta is putting this stuff out open source? They have the whole metaverse they need filled and all of these systems help developers create the type of things needed to populate their cyber-space

[rohvir]

hmmmm its been a few months but the model and weights still haven't been released... wonder if they are ever planning to do so.

[Dron]

That is very impressive. I wonder when real multi modal models will appear and become open source. These model should work with audio, video, detect poses which could be thought as another dimension. It should also be able to generate 3d model and internally represent images as 3d model when needed. Same model should be able to manipulate robots limbs etc.

[SpaceCattttt]

What's frustrating for me is that every time a new model is released, one of two things happen.
1. It costs money. I refuse to pay for A.I.
2. It's open source and free, but it hasn't been released yet. And by the time it's released, a new model has already replaced it in terms of relevancy.

And then it's back to beginning of point 1 or point 2. - It's a vicious cycle.

[FusionDeveloper]

It's seems weird to think of it like "Facebook" is doing all this.
I guess it's just that Mark Z. has a lot of money and believes in AI stuff.

[FusionDeveloper]

People using Stable Diffusion, Check out the SD 1.5 "Photon" safetensor model, it is amazing.
I can render 1280x1024 pretty consistent without issues, depending on the prompt.
Of course if you want to do specific things, it's going to have the same tiling/repeating/extra stuff that all models have when you try to use too high of a resolution.