This week’s new tools, tutorials, and resources 👇
Google Pomelli: A free tool for generating on-brand social media campaigns & assets with AI.
Very Good Components: A free collection of easily-editable UI components (no coding required).
Tab Surfer: A free Chrome extension for saving, managing, and restoring browsing sessions.
Figma: Examples of iconic web interactions from the past two decades.
In partnership with
FROM OUR FRIENDS @ STAN
Your Next Dream’s On Us
Share your dream business with the world and enter for a chance to win $100,000. We’re Creators, too. We know with the right support anything is possible.
Join the challenge today and tell us your story.
NO PURCHASE NECESSARY. VOID WHERE PROHIBITED. For full Official Rules, visit daretodream.stan.store/officialrules.
Interested in sponsoring our newsletter? Book an ad here.
ARTIFICIAL INTELLIGENCE
📺 Midjourney releases “Niji V7”
This week, Midjourney released "Niji V7," a major update to its anime-focused image model, the first new Niji version since V6 launched in January 2024.
The update focuses on coherency, prompt accuracy, and refined aesthetics, making it significantly easier to generate clean, professional anime artwork.
Midjourney is a community-funded research lab known for its AI image generation tools. Niji is its specialized model built for anime and Asian art styles, developed in collaboration with Spellbrush.
Prompt: “a girl with short green hair, hair in a single bun, blue eyes, fang, standing under the sakura trees”
With Niji V7, you can now:
Generate crystal-clear details: Eyes, backgrounds, and fine elements like floating flower petals now render with dramatically improved coherency and clarity.
Get exactly what you prompt: The model follows complex prompts more literally — if you ask for a character with four arms holding ice cream, you'll actually get it.
Use style references (sref): Apply visual styles from reference images, now with improved performance and backwards compatibility.
Create authentic anime linework: Use the prompt "anime screenshot" to access sophisticated line rendering that mimics traditional anime production styles.
One notable omission: character reference (cref) isn't supported in V7. The team says they're working on a "super special secret surprise" to replace it.
To use Niji V7, type --niji 7 after your prompt in Discord, or select "Niji 7" from the Version dropdown on the web. It's available to all Midjourney subscribers.
Watch a music video generated with Niji 7 👇
FROM OUR FRIENDS @ WISPR FLOW
Vibe code with your voice
Vibe code by voice. Wispr Flow lets you dictate prompts, PRDs, bug reproductions, and code review notes directly in Cursor, Warp, or your editor of choice. Speak instructions and Flow will auto-tag file names, preserve variable names and inline identifiers, and format lists and steps for immediate pasting into GitHub, Jira, or Docs. That means less retyping, fewer copy and paste errors, and faster triage. Use voice to dictate prompts and directions inside Cursor or Warp and get developer-ready text with file name recognition and variable recognition built in. For deeper context and examples, see our Vibe Coding article on wisprflow.ai. Try Wispr Flow for engineers.
Interested in sponsoring our newsletter? Book an ad here.
VIDEO EDITING
🤖 Descript releases “Underlord v2”
This week, Descript released "Underlord v2," a major upgrade to its AI co-editor that makes video editing feel less like surgery and more like a conversation.
The v2 update delivers faster performance, smarter edits, and uses about 20% fewer AI credits to fully edit a video before publishing.
Underlord is Descript's built-in AI assistant that can edit your videos, generate visuals, add effects, and handle tedious tasks through simple text prompts (like having a video editor on call).
With Underlord v2, you'll notice more thoughtful edits and stronger responses when using:
Slide-to-video workflows: "Turn these slides into a narrated video I can post on YouTube."
Time and duration-based editing: "Add captions for the first 30 seconds of this video, and make the video fade out for the last 5 seconds."
Image and video generation: "Create a visual for the highlighted section— take what he's saying and make a graph for the audience."
Underlord v2 is also better at explaining its reasoning and self-correcting mistakes. You'll see it present its initial plan, update you as it thinks through edits, and occasionally pause to check if it's on the right track.
You can also choose which AI model powers Underlord depending on your needs: Claude Haiku 4.5 (for quick, economical edits), Claude Opus 4.5 (for complex, multi-step projects), or Gemini 3.0 Pro and GPT 5.2 (for different creative personalities).
Underlord v2 is rolling out now to all Descript users with AI features enabled.
FROM OUR FRIENDS @ REMARKABLE
Start your year with clarity
Written by Shane Parrish and reMarkable, this workbook helps you pause and reflect. Rather than add pressure, it relieves it, bringing clarity to what truly matters.
A simple reset for January. A thoughtful way to review your year.
Interested in sponsoring our newsletter? Book an ad here.
WEB DESIGN
✏ Framer releases “On-Page Editing 2.0”
This week, Framer released "On-Page Editing 2.0," a major upgrade that gives editors more flexibility to make changes directly on live sites, while designers stay in control of what can (and can't) be touched.
On-Page Editing lets non-designers make quick content updates to a published Framer site without opening the full editor.
Think of it as a "safe mode" for marketers and content teams to tweak copy, swap images, or rearrange sections without accidentally breaking your layout.
With On-Page Editing 2.0, editors can now:
Rearrange content directly on page: Drag and drop text, frames, and components to reorder cards, sublines, or sections (no need to open Framer's design editor).
Swap icons and variants with one click: Click any icon to browse and replace it instantly, or switch component variants to better match page content.
Edit localized content in real time: Visit a localized version of your site and update translated text or images directly, with full localization support now built in.
Move the edit button anywhere: The circular edit button can now be dragged to any edge of the screen, so it no longer overlaps your site content.
Designers retain full control over what editors can modify. Lock any layer or section in Framer using ⌘ + L (Mac) or Ctrl + L (Windows) to prevent accidental changes (locked elements can't be rearranged or edited on page).
Teams using On-Page Editing extensively can now also unsubscribe from project update emails directly within the email itself.
OTHER STORIES
Everything else in creative news 🗞
Figma’s Jira app now support webhooks, so teams can get instant design status updates like “Ready for Dev” directly in Jira tickets.
Figma introduced a new left navigation bar in Figma Design, Draw, and Dev Mode, making it easier to move between assets, variables, and search.
Framer is now offering a free custom domain for the first year when users upgrade their sites to a yearly plan.
Gamma released “Remix,” a new feature allowing you to: -adapt any presentation for a new audience or purpose, replace any placeholders for different customers, update reports with new data/charts you upload, and more.
Runway released a video showcasing the first demonstration of video generation on the NVIDIA Rubin platform.
Runway released “Seamless Transitions,” allowing you to create multiple generations with a single consistent subject interacting with different objects across various scenes and environments.
Rive for Android (Compose) is now in beta.
Freepik released “Variations,” allowing you to turn a single image into multiple frames, build a storyboard, and animate it into a full video.
Bezi released “Actions” in Agent Mode — a simple prompt to Bezi, and it can directly create GameObjects, edit prefabs, set up components, and generate materials.
Tella now allows you to embed your Tella videos inside Linear issues and comments.
Cap released version 0.4.0, which includes Screenshot Mode, recoverable recordings, captions, ew tracks, a new recording picker, and better performance.
MagicPath added completion sounds to MagicPath which notify you when a design finishes.
Higgsfield AI released “Relight,” allowing you to pick a direction, adjust intensity, set color temperature, and relight any scene.
Higgsfield AI released “Cinema Studio v1.5,“ adding new aperture control bokeh, new aspect ratios with DP-curated presets, and more.
Higgsfield AI released “AI Stylist,” allowing you to upload an image, pick from 7 categories, customize backgrounds & poses, and generate product-ready assets in seconds.
Gamma added support for Flux 2 Max.
KREA AI released trainers for Qwen-2512 and Z-Image, allowng you to train LoRAs for these two models and use them in Krea Image.
*Some links in this newsletter may be from sponsors or affiliates, which means we might earn a commission if you make a purchase.
