## July 6, 2026

### Image Generator & Video Generator — new standalone tabs

The Canvas **Image Generator** and **Video Generator** nodes now have their own dedicated pages, available from the left sidebar on the [Beeble web app](https://app.beeble.ai/image-generator).

- **Same models as Canvas** — every model available in the Canvas Image / Video Generator nodes is available here, with identical options and [credit pricing](https://docs.beeble.ai/help/credit-policy)
- **History feed** — all your generations in one scrollable feed, with batch generation support

### Video Generator — Gemini Omni Flash & new Seedance tiers

**Gemini Omni Flash**, Google’s natively multimodal video model, is now available in the Video Generator — both the standalone tab and the Canvas node.

- **Four tasks** — text-to-video, image-to-video (starting frame), reference-to-video (up to 3 reference images as subject / style guides), and prompt-based video editing
- **Audio included** — every generation comes with generated audio
- **Output** — 720p, 16:9 or 9:16, 3–10 second clips (video editing accepts source videos up to 10 seconds)

Seedance 2.0 also gains two new tiers:

- **Seedance 2.0 — 4K output**: Standard mode now supports **4K** resolution, alongside 480p / 720p / 1080p
- **Seedance 2.0 Mini**: a new cheapest tier — about half the credit cost of Standard (480p / 720p only)

See the [Credit Policy](https://docs.beeble.ai/help/credit-policy) for pricing.

## July 3, 2026

### Introducing SwitchHDR — True 16-bit HDR from any video

**SDR to HDR**, powered by our new model, [**SwitchHDR**](/content/research/switchhdr/index.html) is now available on [Beeble web app](https://app.beeble.ai/sdr2hdr). Convert any 8-bit clips into true 16-bit HDR EXR sequences that work in your professional production pipelines.

- **True 16-bit EXR & ACES Workflow**: Generate production-ready EXRs with native ACES support for high-end VFX and color grading.
- **Expanded Depth and Range**: Restore blown highlights and push shadow depth smoothly in the grade without banding artifacts.
- **Precise Control**: Mask target areas, use text prompts, and direct exactly what needs to come back for your final look.
- **Powered by [SwitchHDR](/content/research/switchhdr/index.html)**: Our dedicated model trained specifically for accurate SDR to HDR reconstruction.

## July 2, 2026

### Canvas — Share your canvas with anyone

You can now **share a Canvas project with a link or by email** — from the **Share** button in the canvas toolbar.

- **Share with a link** — switch General access to **Anyone with the link** and copy the link. Anyone can view; switch back to **Restricted** to stop sharing.
- **Share by email** — invite specific people as viewers while keeping the canvas Restricted.
- **Duplicate & Edit** — viewers can duplicate a shared canvas into their own workspace and build on it.

## June 17, 2026

### Video Generator — 20+ new models

We now support **20+ new video generation models in Beeble Canvas!**

- **Grok Imagine (incl. 1.5)** — text / image / reference-to-video, plus video edit & extend
- **Wan 2.7 / 2.6** — text / image / reference-to-video, plus video edit
- **LTX 2.3 (Pro / Fast)** — text / image-to-video, plus extend & retake
- **Luma Ray 3.2** — text / image-to-video, plus video edit & reframe

### Image Generator — 10+ new models

We now support **10+ new image generation models!**

- **Grok Imagine (incl. Quality)**
- **Seedream (V5 Lite / V4.5)**
- **FLUX.2 (Pro / Max / Flex / Klein)**
- **Recraft V4.1 (Pro / Utility / Utility Pro)**
- **Z-Image (Turbo / Base)**

### Image / Video Describer

- Generate descriptions from up to 4 media inputs
- Choose your preferred LLM and add optional prompts

### Quality of Life Updates

- **Sticky Note — Rich text & markdown supported**  
  - Write headings, **bold** / _italic_ / ~~strikethrough~~, lists, blockquotes, inline code, and links  
  - Paste markdown to auto-format, or copy your content out as clean markdown

- **Topaz Upscaler — built-in comparison added**  
  - Drag the split slider to instantly compare original and upscaled images  
  - No extra Compare node needed

- **Compositor — magnetic snap added**  
  - Clips now snap to clip edges, the playhead, or frame 0 during edits  
  - Use the magnet button or press `S` to toggle

## June 11, 2026

### Canvas — New Nodes and Node Updates

New nodes and a wave of node updates for Canvas, plus expanded documentation.

#### New Nodes

- **Topaz Video Upscaler**— Upscale videos up to 4K
  - Upscale videos up to 4K with Topaz (more providers coming later)
  - Choose an upscaling method (e.g., diffusion-based or classic models)
  - Set the scale factor from 1× to 4× (output is capped at 4K)
  - Frame interpolation is supported (for smoother motion / higher FPS)

- **Topaz Image Upscaler**— Upscale images up to 8K

- **Video Iterator**— Batch-process many videos at once
  - Upload multiple videos at once, then connect the iterator to supported nodes to batch-process a full set of inputs
  - Useful when you want to apply the same workflow to many clips without uploading one-by-one

- **Image Iterator**— Batch-process many images at once

- **Text Concatenator** — Concatenate two prompts; use one as a system prompt and the other as a user prompt

#### Node Updates

- **VFX Pass Generator**— PNG Sequence export and “Export All” added

- **Depth Generator**— 16-bit EXR export option added

- **Video Matte**— RGBA export options (MOV + PNG) and alpha PNG export added

- **Export**— choose a download path for your files

- **LLM**— Gemini 3.5 Flash model added + video input now supported

- **Rotate / Flip**(formerly Flip) — Rotate 90° now supported

- **SwitchX**— set a random seed

- **Erode / Dilate**— range is now 2× larger (−200 to +200)

### Documentation

- **[Credit Policy](https://docs.beeble.ai/help/credit-policy)** — We documented how many credits each node uses, so you can plan workflows with costs in mind—especially as more third-party models are added.
- **[Model List](https://beeble-ai-docs.notion.site/Beeble-Trust-Center-AI-Model-List-3768ee80e61380eb8d32e682e8c00b65)** — We list all available models (third-party and internal) with clear sourcing and policy details. “Verified” means we have confirmed commercial licensing and non-training terms directly with the provider.

## May 19, 2026

### Introducing Canvas: Your Ultimate AI Compositor

Canvas is a node-graph compositing tool built for filmmakers, VFX artists, and post-production professionals. It bridges generative AI and traditional compositing, letting you create high-end video content at the speed of AI with total node-based control.

### AI Tools

- **SwitchX** and **VFX Pass Generator** (SwitchLight) available directly in your node workflow
- **Video Matte** — AI rotoscoping with instant subject isolation (MatAnyone & CorridorKey)
- **LLM** and **Image Generator** — text and image generation inside the graph
- **Video Generator** — native integration with Seedance 2.0, Kling 3.0, and more
- **Camera Angle (Qwen)** — re-render stills from a different camera angle
- **Depth Generator** — depth map extraction using Video Depth Anything

### Edit Nodes

- **Trim**, **Crop**, **Resize**, **Extract Frame** — essential video/image manipulation
- **Curves**, **Invert**, **Flip**, **Blur**, **Channels** — color and pixel-level control
- **Erode / Dilate** — matte cleanup for alpha passes
- **Merge** — foreground/background compositing with 16 blend modes

### Utility Nodes

- **Viewer**, **Compare**, **Export** — preview, A/B comparison, and download
- **Router**, **Sticky Note**, **Text Iterator** — graph organization and multi-prompt batches

Available now on the [Beeble Web app](https://app.beeble.ai/canvas).

## April 16, 2026

### Background Remover — CorridorKey Model

Added **CorridorKey**, a new AI model optimized for **green screen footage**. You can now choose between two models when using Background Remover:

- **MatAnyone** (default) — Best for general background removal
- **CorridorKey** — Best for green screen footage, delivering cleaner edges and more accurate keying on chroma-key shots

## March 24, 2026

Resolved flickering and exposure shift issues in SwitchX that affected videos longer than 80 frames.

## February 20, 2026

Introducing **SwitchX** — our first video-to-video model. It lets you modify lighting, background, props, and more, while keeping your original subject intact.

## January 27, 2026

We’ve rebuilt the Edit Alpha experience from the ground up with full control over your masks and new tools to speed up user’s workflow.

### Full Control Over Your Masks

- **Mask Inversion**: Flip any mask instantly
- **Mask Opacity Control**: Adjust mask visibility
- **Custom Alpha Import**: Bring user’s own masks

### Tools & Shortcuts

- Pen tool with Bezier curves
- `Shift` + Click/Drag: Create perfectly straight lines
- `Alt`: Switch brush/erase mode on the fly

## January 17, 2026

New rotoscoping tools, batch downloads, and improved file management for VFX passes and projects.

### Rotoscoping Update

We upgraded the refine alpha model to **SAM3**, enabling precise object selection with **text prompts**. We also added **Brush** and **Pen** tools for detailed masking control.

### Batch-download VFX Passes

You can now select multiple VFX passes at once and download everything together in a single action.

### List View & Filters

Added List View and new filters (VFX pass, status, type) for better file management.

## January 9, 2026

### Alpha Viewing Modes

- **Black / White**: Check edge transparency against solid backgrounds
- **Checkerboard**: Standard transparency grid for compositing checks
- **Raw Alpha Mask**: View the grayscale mask directly to identify noise or edge issues

## December 19, 2025

This week brings expanded platform support for Linux users, new Video to VFX controls, and quality-of-life improvements across Beeble Studio and 3D Relight.

## December 12, 2025

Practical updates across the platform focused on improving Video to VFX workflows, lighting control, pipeline stability, and on-set capture flexibility.

## December 1, 2025

Beeble Studio now supports EXR and refined alpha generation for professional workflows.