Kling AI's video character consistency feature
Your face, any setting, consistent quality
Total reading time around 3 minutes.
Welcome to Visually AI!
🔮AI News this Week
Kling AI Video Character Consistency
I got early access to "Custom Models" and the results are mind-blowing. Train a model on your face to generate videos in any setting.
Training videos:
You'll need to upload 10-30 videos between 10-15 seconds each. There should only be 1 clearly visible face per video, with different poses, angles, and backgrounds.
Generate with your custom model:
My model finished after about 90 minutes, and I started generating. You enter a prompt and select your custom model as a face reference.
I don't know when the feature will be available to everyone, but I'm sure it will be soon, based on previous launches.
New FLUX1.1 [pro] Ultra & Raw Modes
Black Forest Labs introduced the new high resolution FLUX1.1 [pro] Ultra and Raw Modes.
Ultra produces 4x higher resolution in the same amount of time (10 sec), without losing prompt adherence.
Raw mode is more realistic and has a less ‘processed’ appearance.
Here are some of the platforms with the new Ultra model:
📸AI Snapshots
OpenAI acquired chat.com (automatically redirects to ChatGPT) for $10M+ from tech billionaire, Dharmesh Shah. Seems like that would make it one of the largest domain purchases in history, and we could just call it “Chat,” instead of ChatGPT…
Runway launched Advanced Camera Controls for Gen-3 Alpha Turbo for image to video generation. You can choose the direction and intensity of the scene. There is no motion brush, but there are preset examples you can click on and automatically adjust the correct settings.
Here’s a few of my first results with camera controls:
Finally, we can search old chats in ChatGPT by content and title. Click the 🔍 in the top sidebar and type a word or phrase you need to find. Available now for Plus and Team users, rolling out to Enterprise and Edu this week, and Free users throughout November.
⚙️ Workflow
Inpainting with Freepik Retouch
💫 Another great option for inpainting images: Freepik Retouch
I've been using Freepik's Flux 1.1 and Mystic image models and Retouch is impressive for quickly adding, removing, or changing details.
Here's a quick demo of a Mystic image I edited with Retouch:
🛠️ This Week’s AI Tools
Map This: Upload docs or text to generate a mind map of the information. (link)
Clipify: AI video generator, editor, file convertor, and more. (link)
Meshcapade: Introduced a new Motion Blending feature lets you blend any video or text motion from the platform into seamless transitions. (link)
Text Behind Image: Place text behind objects in your images easily, for free. (link)
🖼️ Image Prompts
Prompt: A figure in a black floral kimono sits on rocks by stylized waves, with birds in the sky, in a Japanese ink painting style with minimalistic composition.
Prompt: Luxury sedan parked in a clean suburban garage with the door open, organized tools and bikes visible in the background, soft afternoon light, cozy and practical atmosphere, high-resolution details
🎬Video Prompt
Prompt: a forest at sunrise with sunlight filtering through the trees
Generated on Luma Dream Machine:
Thank you for reading, and have a creative week!
Great post! Blendic AI seems cool. Thanks for sharing this tool, gonna try it!