Welcome back to Visually AI!
Today’s reading time is 5 minutes.
In today’s edition:
Kling 3.0: Everyone a Director
Character consistency, native audio, 15-second generations & first results
Image & Video Prompts
Imagine Art 1.5 Pro, Genspark AI Workspace 2.0 & PixVerse Real-Time Video Workflows
Kling just dropped version 3.0, and it’s a legitimate leap forward for AI video production (Kling is the GOAT). After spending early access time testing the new capabilities, I can confirm this is the most significant update to video generation tools I’ve seen in months.
Key highlights:
Character & Element Consistency: Characters and elements stay locked across shots - no more regenerating until you get matching outputs
Flexible Video Production: 3-15 second generations with precise control and customizable multi-shot sequences
Native Audio with Dialogue & Singing: Multi-character voice support with multiple dialects, accents, and languages
Enhanced Image Generation: New image series mode for consistent visual storytelling
Professional Output: 1080p video with start and end frame control for precise editing workflows
Ultra subscribers have early access now, with general rollout coming soon.
I ran ~100 prompts to test the Elements system and new capabilities:
Camera movements: Handled multi-angle sequences automatically. Natural motion, clean transitions, no manual editing needed.
Character consistency: Elements reference from video and images with audio worked reliably. Same character across shots with different lighting and angles. Voice tone held in multi-character scenes.
Complex environments: Deep-Stack rendering reduced the “plastic AI look.” Texture mapping and lighting physics felt convincing in busy café, nightclub, and market scenes.
Photorealistic detail: Visual Chain-of-Thought improved spatial logic. Proper cinematic framing, realistic materials, accurate perspective.
Text and atmosphere: Native text output actually legible. Lighting effects (noir, golden hour, fog) rendered with convincing physics.
And much more (this is by far the best model on the market)…
Below are some examples from my testing (find more on 𝕏 in the coming week!)
Prompt: Cinematic portrait of woman in vintage film noir style, venetian blind shadows across face, high contrast black and white, dramatic side lighting, 1940s aesthetic, film grain texture
Prompt: Landscape Pop Art Deco Painting, Maximalistic, Coastal Town at golden hour, Watercolor ink paint, warm color grading, highly detailed, volumetric, hyper realistic, 3D, award-winning
Prompt (Kling 3.0):
Shot 1 (4s): Close-up of architect @Confident Female Voice reviewing blueprints at standing desk in modern studio, natural daylight from floor-to-ceiling windows, “These proportions need adjustment,” finger tracing lines on paper, glasses catching light
Shot 2 (3s): Medium shot as same architect walks to scale model across studio, carrying blueprints, sleek minimalist workspace, afternoon sun creating long shadows, plants in background, confident stride
Shot 3 (5s): Wide shot of architect presenting model to two clients seated across table, gesturing to specific details, “This is where the light enters,” natural conversation body language, studio visible through glass partition behind, professional atmosphere
Prompt (Kling 3.0): Vertical interior shot of vintage submarine observation deck, single figure pressed against thick circular porthole, bioluminescent creatures drifting past in the abyss outside, flickering interior red emergency lighting, condensation on glass, crushing depth and isolation, subtle groaning metal hull sounds
I’ve had some amazing opportunities to work on a variety of projects recently, and I don’t share everything in this newsletter or online. And you know I love testing tools to share.
Take a look at my portfolio to get a quick glimpse of my work:
Kling 3.0: Native multimodal video creation with character consistency, multi-shot sequences, and native audio. Ultra subscriber early access now live. (link)
Google Project Genie: Google’s world model creates interactive, explorable 3D environments from text prompts. Instead of static video, you navigate generated worlds in real-time at 20-24 fps with up to 60 seconds of visual memory. Best for stylized content (claymation, watercolor, anime aesthetics) rather than photorealism. Early access available to Google AI Ultra subscribers ($249.99/month, US only). Perfect for pre-visualization, concept development, and interactive mood boards. (link)
ImagineArt 1.5 Pro: Major update delivers native 4K photorealism with enhanced texture fidelity and cinematic workflows. The image-to-video pipeline produces genuinely usable outputs - realistic skin detail, natural motion, and proper lighting physics. I tested this extensively for a recent collaboration and the quality leap from previous versions is significant. Best for creating high-resolution assets that need to hold up in professional contexts or large format displays. (link)
Heather Cooper@HBCoop_
ImagineArt just launched 1.5 Pro. Native 4K. Photorealistic detail. Here's what it can do and how you can implement it into Cinematic Workflows 🧵
4:18 PM · Jan 21, 2026 · 132K Views
13 Replies · 37 Reposts · 114 Likes
Genspark AI Workspace 2.0: All-in-one workspace that just hit $100M ARR in 9 months. The update adds AI Slides Creative Mode, custom workflows in AI Inbox, and Speakly voice dictation (4× faster than typing). What stands out is the integration - you're not jumping between tools. AI agents handle presentations, visual generation, and design work in one environment. Best for creators managing multiple content types who want to consolidate their workflow stack. (link)
Heather Cooper@HBCoop_
Still juggling ChatGPT, Midjourney, Canva, and 15 browser tabs? @genspark_ai just hit $100M ARR (Annual Run Rate) in 9 months. I had to see why. Meet Genspark: the all-in-one AI Workspace 2.0 🧵
4:10 PM · Jan 28, 2026 · 4.3K Views
2 Replies · 9 Reposts · 27 Likes
PixVerse Real-Time Video: Alibaba-backed PixVerse launched real-time interactive video generation - direct characters to cry, dance, or pose with instant responses during generation. Ideal for rapid iteration and social content creation. (link)
Heather Cooper@HBCoop_
✨ PixVerse R1: Real-time World Model I got early access to test PixVerse's new real-time generative engine. This is interactive, adaptive content that responds as you go. Still in early testing, but here's a preview 🧵
8:00 PM · Jan 21, 2026 · 3.88K Views
5 Replies · 14 Reposts · 41 Likes
Thank you for reading.
I hope you have a creative week!
Heather Cooper
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.