AI Art Weekly #110

Hello, my fellow dreamers, and welcome to issue #110 of AI Art Weekly! 👋

It has been unusually quiet this week, almost like we’re about to be hit by a big storm. Nonetheless, there were some cool updates, like Luma launching a new video model with incredible motion capabilities, as well as some other interesting papers that I cross my fingers will get a code release (I’ve already bookmarked them with Premium to get notified when they do).

Enjoy your weekend and until next week!

---

News & Papers

Highlights

Luma Ray2

Luma has launched Ray2, a powerful AI video generation model that represents a 10x computational increase over its predecessor. It currently supports text-to-video with advanced motion handling, ultra-realistic details and physics-simulation capabilities. Image-to-video and video-to-video transformation features are on their way as well.

3D

BioPose can predict accurate 3D human poses from single videos.

Image

SynthLight can relight human portraits using environment map lighting to create realistic effects like highlights and shadows.

FramePainter can edit images using sketches, allowing for highly coherent modifications. It excels in generalizing to out-of-domain scenarios, such as transforming a clownfish into a shark-like shape, while maintaining temporal consistency.

1-2-1 can achieve high-quality virtual try-on for images and videos. It uses a Modality-specific Normalization strategy to process different inputs and reduces the need for complex systems, making it easier to use for high-resolution applications.

Video

Go-with-the-Flow can control motion patterns in video diffusion models using real-time warped noise from optical flow fields. It allows users to manipulate object movements and camera motions while keeping high image quality and not needing changes to existing models.

FlexiClip can generate smooth animations from clipart images while keeping key points in the right place.

RepVideo can improve video generation by making visuals look better and ensuring smooth transitions.

LayerAnimate can animate single anime frames from text prompts or interpolate between two frames with or without sketch-guidance. It allows users to adjust foreground and background elements separately.

Video Alchemist can generate personalized videos using text prompts and reference images. It supports multiple subjects and backgrounds without long setup times, achieving high-quality results with better subject fidelity and text alignment.

Perception-as-Control can achieve fine-grained motion control for image animation by creating a 3D motion representation from a reference image.

Audio

AnCoGen can analyze and generate speech by estimating key attributes like speaker identity, pitch, and loudness. It can also perform tasks such as speech denoising, pitch shifting, and voice conversion using a unified masked autoencoder model.

And that my fellow dreamers, concludes yet another AI Art weekly issue. If you like what I do, you can support me by:

• Sharing it 🙏❤️
• Following me on Twitter: @dreamingtulpa
• Buying me a coffee (I could seriously use it, putting these issues together takes me 8-12 hours every Friday 😅)
• Buying my Midjourney prompt collection on PROMPTCACHE 🚀
• Buying access to AI Art Weekly Premium 👑

Reply to this email if you have any feedback or ideas for this newsletter.

Thanks for reading and talk to you next week!

– dreamingtulpa