Magnific adds Recraft V4, Ideogram 4, Kling Omni 3, Astra 2, and Auto Layers

• Magnific added magnific's Auto Layers demo, which splits a flat image into editable text, subject, and background layers inside Designer, Spaces, and Magnific MCP.
• magnific's Designer node walkthrough positions Spaces as a full layout editor with multi-page canvases, fonts, effects, and in-place image and text generation, while the Designer node docs show those pages can feed downstream workflow nodes.
• The new model drop in magnific's five-model roundup pulls Recraft V4, Ideogram 4, Kling Omni 3, and three new Topaz upscalers into one workspace, with magnific's Recraft V4 note saying V4 is unlimited for Pro users while V4 Pro stays credit-based.
• For brand and UI work, magnific's Ideogram 4 post highlights bounding-box layout control, hex color palette conditioning, and native 2K output, while the older Recraft V4 guide explains why Magnific is leaning into layout-friendly image models.
• Magnific's MCP story comes with a catch: magnific's MCP launch says it is live on all paid plans, but magnific's credit reply and the Magnific MCP page both say every MCP action still burns credits, even on plans with unlimited web generations.

You can browse the Magnific MCP page, the Designer node docs, and the image model catalog. The strange little detail is that Magnific is shipping two different kinds of control at once: Auto Layers turns static images back into editable parts, while Kling Omni 3's update pushes video in the other direction, toward shot-to-shot character and product consistency.

Auto Layers is the cleanest new trick in this batch. According to magnific's launch post, you can drop in an existing image and get separate editable layers for text, subject, and background, instead of rebuilding the file because one word changed.

That lines up with what the Magnific MCP page already promises on the editing side: layer edits, resizing, relighting, and reuse of past creations from chat clients and IDEs. Magnific also says Auto Layers works in Designer, Spaces, and MCP, so the same decomposition step can sit inside the app or inside an assistant-driven workflow.

Magnific is also trying to keep layout work inside the same canvas. In magnific's walkthrough, the Designer node is a full editor inside Spaces with multi-page designs, custom fonts, effects, and direct placement of generated images and text.

The Designer node docs add the concrete mechanics that matter:

• multi-page outputs, with each page exported as its own image
• placeholders that pull in AI-generated text or images from connected nodes
• a layers panel, inspector, templates, and export controls inside the same editor
• standalone Designer mode outside Spaces, with the same core feature set

Magnific paired that rollout with a Magnific Academy course for Spaces, which suggests the product is drifting from one-off generation toward reusable production flows.

The new inventory breaks down into image generation, video generation, and video upscaling:

1. Recraft V4 and V4 Pro: magnific's Recraft V4 note says V4 is unlimited for Pro users, V4 Pro is credit-based, and both support style references.
2. Ideogram 4: magnific's Ideogram 4 post calls out text rendering, bounding boxes for layout control, hex color palette conditioning, and native 2K output.
3. Kling Omni 3: magnific's Kling Omni 3 post adds up to three character references and three product references per generation, with multishot support for image and video references.
4. Topaz Starlight Fast 2: magnific's Topaz models post frames it as the speed option in the Video Upscaler.
5. Topaz Starlight Precise 2.5 and Astra 2: the same Topaz models post positions Precise 2.5 as the maximum-detail mode and Astra 2 as the balance between quality and performance.

The roster matters less as a shopping list than as a workflow signal. The image model docs say Magnific already exposes 30-plus image models and an Auto selector, so this week's change is really about putting more specialized choices next to the editor, not asking creators to leave the tool every time a model trend shifts.

Two of the new additions are aimed at the same old production headache: keeping assets stable after the first nice frame.

For stills, magnific's Ideogram 4 post focuses on text, bounding boxes, and brand color control, which is much closer to ad and UI work than to freeform prompting. For motion, magnific's Kling Omni 3 post adds multi-reference support for characters and products across shots.

Magnific's earlier Kling 3.0 guide describes Omni as the production version of Kling, built around reference control and cross-scene consistency. That makes this week's Omni 3 update feel less like a random model refresh and more like Magnific tightening the same pitch across image, video, and layout: get the asset into the system once, then keep it stable while the format changes.

Magnific is happy to plug into Claude, ChatGPT, Cursor, Codex, VS Code, Gemini, and Hermes on the front end, according to the Magnific MCP page. The billing model is stricter than the marketing copy.

Across one reply about balance checks, a reply about unlimited plans, and a follow-up on web versus MCP usage, Magnific repeats the same rule: MCP generations always consume Magnific credits. The plans and pricing docs still reserve unlimited generations for Premium+ and Pro on most web models, but that exemption does not carry over to MCP.

That makes the product split pretty clear. The web app is where Magnific can sell abundance. MCP is where it sells access to the whole stack from the tools people already live in.