The Craziest Faceswap I've Seen Yet / Midjourney's Future & Two New AI Video Platforms!

Theoretically Media
25 Apr 202410:38

TLDRThis video delves into the latest advancements in face-swapping technology, showcasing a remarkable example from AI Katana that delivers convincing tracking and realism. The host also discusses the future of Midjourney, a 12-month roadmap hinting at a shift towards 3D, real-time video, and interactive world simulation. Synthesia's new Express model for AI avatars is introduced, highlighting the avatars' enhanced emotive capabilities. Midjourney's recent updates, including the 'style random' feature, are explored, demonstrating its potential for creative and useful applications. Finally, two new AI video platforms, Morph Studios and Nim Video, are introduced, with Morph Studios offering a unique node-based UI and Nim Video providing features like image-to-video conversion and video restyling.

Takeaways

  • 🤖 AI face-swapping technology has made significant advancements, with AI Katana showcasing a highly realistic example that tracks convincingly even during complex facial movements like eating.
  • 🚀 Midjourney's 12-month roadmap hints at a shift towards 3D real-time video generation, offering full 360° control over generated scenes.
  • 🎭 Synthesia introduces Expressive AI, a new model for AI avatars that can convey emotions, with pre-trained avatars available for users to employ without self-recording.
  • 🧐 The new face-swapping model from AI Katana is speculated to not be running in real-time, as real-time face-swapping still has some inconsistencies.
  • 📈 Midjourney's new feature 'style random' randomizes the style of generated images, offering a fun and useful tool for creative exploration.
  • 🌐 Morph Studios, currently in beta, offers an animated look for AI video generation with a node-based UI structure for customizing styles and transitions.
  • 📹 Nim Video, another AI video generator in beta, provides features like style and character options, lip-sync, and the ability to work with layers and motion control.
  • 🔍 The lack of data has been a challenge for 3D advancements in Midjourney, but data collection efforts are increasing to overcome this hurdle.
  • 👓 The Orb, a device speculated to manage thousands of 3D rooms, is taken seriously by Midjourney, with Ahmad, a key figure behind the Apple M1 Pro, hired as head of Hardware.
  • 🔗 Midjourney's co-founder, Alex Evans, has joined the company, bringing experience from developing the 3D creation engine 'Dreams' for PlayStation.
  • 🎨 The 'style random' feature in Midjourney can be used to discover new styles and then apply them consistently to other prompts for a cohesive creative output.

Q & A

  • What is the main topic discussed in the video?

    -The main topic discussed in the video is the advancements in face swapping technology, the future of Midjourney's 12-month roadmap, and the introduction of two new AI video platforms.

  • Which AI company is mentioned as having made significant progress in face swapping technology?

    -AI Katana is mentioned as having made significant progress in face swapping technology.

  • What is the speculated feature that Midjourney might introduce in the future?

    -The speculated feature that Midjourney might introduce is the ability to generate 3D scenes with full 360° rotational camera placement control.

  • Who is the co-founder of media molecule that has joined Midjourney?

    -Alex Evans, one of the co-founders of media molecule, has joined Midjourney as a principal research engineer.

  • What is the name of the new feature released by Midjourney?

    -The new feature released by Midjourney is called 'style random'.

  • How does the 'style random' feature work in Midjourney?

    -The 'style random' feature in Midjourney randomizes the style of the generated image, allowing users to explore different stylistic outcomes.

  • What are the two new AI video platforms mentioned in the video?

    -The two new AI video platforms mentioned are Morph Studios and Nim Video.

  • What is the unique aspect of Morph Studios' user interface?

    -The unique aspect of Morph Studios' user interface is its node-based structure, which allows for a different workflow and style rerolls.

  • What is the name of the Synthesia's new model that can express emotions?

    -The new model from Synthesia that can express emotions is called Express One.

  • What does the Orb, as described by David Holtz, potentially do?

    -The Orb, as described by David Holtz, is a device that could generate and manage thousands of 3D rooms.

  • What is the significance of data collection efforts ramping up for Midjourney's 3D development?

    -The ramping up of data collection efforts is significant for Midjourney's 3D development as it addresses the previous limitation of lack of data, which had held back the advancement of 3D features.

  • How does the video script suggest the future of AI avatars?

    -The video script suggests that the future of AI avatars will include more emotive and expressive capabilities, as demonstrated by Synthesia's Express One model.

Outlines

00:00

😲 Advanced Face Swapping and AI Avatars

The video introduces a new level of face swapping technology from AI Katana, which is highly realistic and tracks convincingly even during complex actions like eating or touching the face. The presenter discusses the potential differences between real-time and captured footage processing. The video also touches on the future of Mid Journey, a 12-month roadmap hinting at a surprising direction. Additionally, two new AI video generators are mentioned, which are of interest to the audience.

05:01

🚀 Next-Gen AI Avatars and Mid Journey's 3D Vision

The video showcases AI avatars from Synthesia that can express a range of emotions. It discusses the new Express one model, which uses pre-trained avatars and aligns lip movements more precisely with speech. The presenter expresses a desire for more details and a proof of concept. Mid Journey's future plans are detailed, with a focus on video, 3D, and real-time elements, aiming to create a non-interactive world simulator with an added interaction layer. The orb, a device for managing 3D rooms, is mentioned as a serious project with a new head of hardware from Apple's Fin Pro team. The video also covers Mid Journey's new 'style random' feature, which randomizes styles and can be useful for discovering new creative directions.

10:02

🎬 Exploring New AI Video Generators

The video discusses two new AI video generators: Morph Studios and Nim Video. Morph Studios is in beta and offers an animated look with character image uploads for consistency, lip sync, and sound features. Its user interface is based on a node structure allowing for style rerolls and connections between shots. Nim Video also in beta, provides options for style, character, camera motion, and lip syncing. It includes features like image to video conversion, video restyling, upscaling, and layer-based editing. The video concludes with an invitation to sign up for the beta of Nim Video.

Mindmap

Keywords

💡Face Swapping

Face swapping is a technology that allows the digital replacement of a person's face in a video or image with another person's face. In the video, it is mentioned as a significant leap forward with AI Katana showcasing a highly realistic face swap that tracks the subject's movements convincingly, even while eating or touching their cheeks.

💡AI Avatars

AI avatars are digital representations of a person that can be controlled or directed by AI. The video discusses the next generation of AI avatars from Synthesia, which are capable of expressing emotions. These avatars are not recordings of real people but pre-trained models that can be used to generate emotionally expressive content.

💡Midjourney

Midjourney refers to a company or technology focused on creating AI-generated content. The video outlines a 12-month roadmap for Midjourney, hinting at a shift towards 3D scene generation with full camera control, indicating a significant development in the capabilities of AI content creation.

💡3D Real Time

This term refers to the generation of three-dimensional content in real time. The video suggests that Midjourney is working on integrating 3D, real-time capabilities into their platform, which would allow for more dynamic and interactive AI-generated scenes.

💡Deepfake

Deepfake technology involves creating hyper-realistic videos where a person's likeness and voice are replicated to appear as if they are saying or doing something they did not. The video discusses the quality of deepfakes in the context of face swapping, noting that while impressive, there are still some inconsistencies.

💡Synthesia Express One

Synthesia Express One is a new model of AI avatars from Synthesia that can express a range of emotions. The video highlights the advancements in emotive capabilities of these avatars, which align more precisely with the speaker's words and voice, offering a more natural and engaging experience.

💡Morph Studios

Morph Studios is an AI video generator mentioned in the video. It is in beta and offers a node-based structure for creating animated-style videos with lip sync and sound features. The platform allows for consistent character creation and a unique workflow for video generation.

💡Nim Video

Nim Video is another AI video generator in beta, offering features like style and character options, camera motion, sound and lip sync, and the ability to work in layers. It also includes image to video conversion, video restyling, upscaling, and motion control, indicating a comprehensive suite of tools for video creation.

💡Style Random

Style Random is a feature released by Midjourney that randomizes the style of AI-generated images. The video demonstrates how this feature can be both fun and useful, allowing users to discover new and unique styles, and then apply those styles to subsequent image generations.

💡Media Molecule

Media Molecule is a developer known for creating the 3D creation engine 'Dreams' for PlayStation. The video notes that Alex Evans, a co-founder of Media Molecule, has joined Midjourney as a principal research engineer, which signals a significant boost to Midjourney's 3D development capabilities.

💡Orb

The Orb is described as a device that could generate and manage thousands of 3D rooms. It is mentioned in the context of Midjourney's future plans, suggesting that it is a part of their vision for creating expansive and interactive 3D environments.

Highlights

AI face-swapping technology has made significant advancements, with a demonstration that is highly realistic and impressive.

The face-swapping technology is showcased via AI Katana, with a video that is convincing even during complex facial movements.

The video features a person speaking in either Mandarin or Cantonese, with a translation to follow.

The presenter speculates that the face-swapping is not real-time, but rather a pre-recorded video processed through software.

Synthesia introduces a new Express one model that allows AI avatars to display emotions, enhancing their expressiveness.

The new AI avatars from Synthesia are pre-trained and do not require users to record themselves.

Midjourney's 12-month roadmap hints at a shift towards video, 3D, and real-time capabilities.

Speculation suggests that Midjourney will enable 360° camera control for generated scenes, offering a new dimension to content creation.

Alex Evans, co-founder of media molecule, has joined Midjourney as a principal research engineer, indicating a significant push towards 3D.

Midjourney's 'orb' device is a serious project aimed at managing thousands of 3D rooms, with a new head of Hardware hired from Apple.

Midjourney has released a new feature called 'style random' which randomizes the style of generated images, offering both fun and utility.

The 'style random' feature allows users to discover new styles and apply them to future images for a consistent aesthetic.

Morph Studios, currently in beta, is an AI video generator that leans towards an animated look and offers a node-based UI for creative control.

Nim Video is another AI video platform in beta that provides features like image to video conversion, video restyling, and motion control.

Nvidia's platform will utilize open-source models, allowing for community contributions and rapid innovation.

The presenter offers a free course on getting started with Midjourney for beginners, available through a link provided.

The video concludes with a teaser for future exploration of Morph Studios and Nim Video once the presenter gains access to them.