What You May Have Missed this week:

1. Revolutionizing Video Generation: Google's VideoPoet

2. Groundbreaking Depth Perception: DMD's Zero-Shot Metric Depth Estimation

3. Upscale-A-Video: A Leap in Video Super-Resolution

4. DreaMoving: Dance Videos Reimagined

5. THE MOST MINDBLOWING DEMO I SEE THE LAST 2 WEEKS!

6. DREAM-Talk: A New Era of Talking Face Generation

7. Tutorial: create your own Pokemon

8. A funny moment between Tony & Leonardo

9. The best tool for this holidays

10. Our amazing Dress Like You Want workflow

1. Revolutionizing Video Generation: Google's VideoPoet

Google's VideoPoet transforms language models into high-quality video generators. It excels in creating diverse, high-fidelity videos with impressive temporal consistency, supporting audio generation from video input. A true step forward in multimedia storytelling.

2. Groundbreaking Depth Perception: DMD's Zero-Shot Metric Depth Estimation

DMD introduces a novel approach for monocular depth estimation, adeptly handling the diversity of indoor and outdoor scenes. It dramatically reduces error rates and promises significant advancements in depth perception technologies.

3. Upscale-A-Video: A Leap in Video Super-Resolution

Upscale-A-Video, a diffusion-based model, excels in upscaling videos while maintaining temporal coherence. It creatively uses text prompts for texture creation, balancing restoration and generative quality, and outperforms its peers in visual realism.

4. DreaMoving: Dance Videos Reimagined

DreaMoving offers a novel framework for generating custom human dance videos. Its unique Video ControlNet and Content Guider ensure motion control and identity preservation, making it a versatile tool in video generation.

5. THE MOST MINDBLOWING DEMO I SEE THE LAST 2 WEEKS!

VidToMe innovates in zero-shot video editing, enhancing temporal consistency by merging self-attention tokens across frames. It efficiently manages memory and processing complexity, marking a significant step in video editing technology.

6. DREAM-Talk: A New Era of Talking Face Generation

DREAM-Talk introduces a diffusion-based method for generating emotional talking faces from a single image. It uniquely combines expressive emotions with accurate lip-sync, outperforming existing methods in expressiveness and quality.

7. Tutorial: create your own Pokemon

The French version is available on our YT channel: https://www.youtube.com/@Fictions-ai

8. A funny moment between Tony & Leonardo

9. The best tool for this holidays

Create your song with only a prompt with suno.ai

9. Our amazing Dress Like You Want workflow

During the last weeks, Thibaud has been working on amazing workflows to create images with custom faces, poses, and faces.
We’re happy to share this workflow right now.

The results are amazing!

We’re recording a tutorial (it will be shared when it’s ready) to explain how it works, but for all advanced ComfyUI users, you can download it here and start playing with it! Download it here (just a short piece of advice, try to activate some combination between blip, llava, and reference)

AI moves fast – don't get left behind! Let us be your guide.

Follow us on social networks: