Product Updates

🚀 New Model Support

- Hailuo 2 & 2.3 (MiniMax): Pixmax now integrates the Hailuo 2 and Hailuo 2.3 models from MiniMax, offering new creative possibilities for high-quality video generation.

✨ Canvas Improvements

- Upscale with Frame Rate Support: Video Upscale now supports adjusting the Frame rate (interpolation mode), resulting in smoother motion and higher visual quality.

🚀 New Features

- Wan 2.7 Image/Pro Model: Now available for users, fully supported in Image-to-Image generation.

- Camera Motion: Added to video generation. Insert specific camera movement prompts directly for dynamic cinematic effects.

- Project Duplication: One-click copy of entire projects (including all canvas data) via the top-right menu. *Batch duplication coming soon.*

✨ Improvements

- Generation Time Display: Full-screen view now shows start and end timestamps for images and videos. - History Download: Directly download assets from the history panel. - Inpainting & Marking Tools:

- Brush defaults to maximum size for faster coverage. - Added keyboard shortcuts for Undo, Redo, and Delete.

🚀 New Models

Seedream 5.0 Pro (Image Model): The latest high-performance image generation model is now live, offering enhanced detail and realism.

Wan 2.7 (Video Model): Users on Pixmax.ai can now access the Wan 2.7 model for superior video generation quality.

✨ Canvas Improvements

High Bitrate Video Options: Added high bitrate parameters for Volcano Engine models (Seedance 2, Fast, and Mini) in video generation, ensuring sharper and clearer output.

Invite friends — 30 credits on sign-up

Share your personal link. For each friend who completes sign-up, the inviter and the invitee each receive 30 credits immediately.

Invite limit & instant credit

You can invite up to 10 friends. Rewards are credited to your balance as soon as your friend registers. Credits never expire and can be used anytime.

Upgrade bonuses — bigger rewards when they subscribe

When an invited friend becomes a paying subscriber for the first time, you’ll receive bonus credits based on their plan:

Starter → +200 credits

Pro → +500 credits

Ultra → +1,000 credits

Max → +2,000 credits

Note: Bonus credits are granted only the first time an invited friend becomes a paying subscriber.

🚀 New Model Support

- Text Models: Added Doubao Seed 2.1 Pro and Doubao Seed 2.1 Turbo from Volcano Engine. - Image Models: Integrated Midjourney v8.1 and Midjourney Niji v7 for superior image generation quality.

✨ Canvas Experience Optimizations

- Enhanced Annotation Tools: Image nodes now support inserting text and arrows. You can freely scale and move annotation symbols for precise feedback. - Smart Project Resume: When you re-enter a project, Pixmax automatically opens the last edited canvas, saving you time finding your place. - Zoom Shortcuts: Quickly zoom in/out the canvas using keyboard shortcuts: Cmd/Ctrl + "+" to zoom in and Cmd/Ctrl + "-" to zoom out. - Connection Line Cutting: Hover over any connection line between nodes to reveal a scissors icon, which lets you instantly cut and disconnect links.

🎬 New Feature: Stage Editor (Beta)

- Precise Scene Control: Add a Stage Editor node to your canvas to precisely adjust character positioning and camera angles for generating accurate reference images. - Intuitive 3D Logic: Operates with standard 3D software controls: Left-click to rotate, Right-click to pan, and Scroll to zoom. You can move, rotate, and scale multiple puppets freely, adjusting their position, depth, orientation, and size to visualize complex spatial relationships that are hard to describe with text.

- Versatile Material Support: Supports panoramic images (automatically set as background) and generates corresponding Stage Editor nodes from 3D models within the canvas. Take snapshots directly into the canvas to serve as references for subsequent image generation.

1. HappyHorse Is Now Available

HappyHorse now supports text-to-video, image-to-video, reference-based generation, and video editing workflows.

2. Team Asset Library

Teams can now create shared asset libraries, making project-specific assets visible and reusable across members.

3. Grid Storyboard Generator

Create grid-style storyboards faster by connecting reference assets with prompt descriptions.

AI models can also help refine prompts for better storyboard consistency.

4. Frame & Node Grouping

5. Keyboard Shortcuts

1. Real-Person Asset Libraries for Seedance 2.0

Verified users can now build real-person asset libraries and use them directly in Seedance 2.0 workflows for more consistent character generation.

2. Redemption Code Support

3. HD Upscaling and Frame Interpolation

4. AI Subtitle Remover

5. Canvas Grid Split

1. Seedance 2.0 Native 1080p Is Live

2. PixVerse V6 model

1. Individual users can access Seedance 2.0

Individual users can now access Seedance 2.0 without additional verification.

2. Enterprise owner sub-accounts are now available

Enterprise owners can create multiple sub-accounts with independent credit limits, while sharing the main account's credit pool.

3. Vidu Q3 Pro & Mix-series

Q3Mix supports reference image-to-video generation and requires a connected image asset.

1. Use multi-angle images for better video reference

2. Batch Asset Uploads

Users can upload multiple assets at once by clicking Upload or dragging local files into the workspace.

3. Selective Downloads for Multi-Image and Video Results

1. New project tree view for managing complex projects

Click the lower-left corner to expand the project tree view.

2. Node generation prompt input supports expansion

3. New nodes reuse the previous model and parameters by default

When creating a new node, the selected model and parameters are selected from the last-used settings by default.

4. Copying nodes can retain input-node connections

Ctrl+V: paste the node only.

Ctrl+Shift+V: paste the node together with the original input-node connection.

5. Optimized real-time credit balance feedback

Credit balance feedback has been optimized so creators can understand usage in time.

1. Seedance 2.0 model support

This model is currently available only to enterprise users.

Seedance multimodal reference-to-video supports reference images (0-9), reference videos (0-3), reference audio clips (0-3), and optional text prompts.

2. Prompt specifications

Chinese and English prompts are supported. We recommend keeping Chinese prompts under 500 characters and English prompts under 1,000 words. When prompts are too long, information can become scattered and the model may ignore details or focus only on key points, causing missing elements in the video.

3. Single-image input requirements

Format: jpeg, png, webp, bmp, tiff, gif.

Aspect ratio (width / height): (0.4, 2.5).

Width and height length (px): (300, 6000).

Size: each image must be under 30 MB. Request body size must not exceed 64 MB. Do not use Base64 encoding for large files.

Image count: 1 image for image-to-video first frame; 2 images for image-to-video first and last frames; 1-9 images for Seedance 2.0 and 2.0 fast multimodal reference-to-video.

4. Single-video input requirements

Video format: mp4, mov.

Resolution: 480p, 720p.

Duration: each video must be 2-15 seconds. Up to 3 reference videos can be provided, and total duration across all videos must not exceed 15 seconds.

Aspect ratio (width / height): [0.4, 2.5]. Width and height length (px): [300, 6000].

Frame pixels (width x height): [409600, 927408]. For example, 640 x 640 = 409600 meets the minimum value, and 834 x 1112 = 927408 meets the maximum value.

Size: each video must be under 50 MB. Frame rate (FPS): [24, 60].

5. Single-audio input requirements

Format: wav, mp3.

Duration: each audio clip must be 2-15 seconds. Up to 3 reference audio clips can be provided, and total duration across all audio clips must not exceed 15 seconds.

Size: each audio clip must be under 15 MB. Request body size must not exceed 64 MB.

6. Run compliance detection first

If the input contains people in images, videos, or audio, click the Compliance Check button above the node before generation.

7. Generate after detection succeeds

A green mark means detection succeeded. You can then proceed to generation.

8. Batch asset detection workflow

If you have many people-related images, videos, or audio clips, run batch asset detection in advance to avoid copyright issues.

9. Enter the Asset Library from the homepage

10. Create an asset group

11. Add assets and wait for them to be stored

The "Stored" status means the asset passed detection. Note that uploaded assets can include images, videos, and audio.

Image requirements: jpeg, png, webp, bmp, tiff, gif, heic/heif; aspect ratio (width / height): (0.4, 2.5); width and height length (px): (300, 6000); each image must be under 30 MB.

Video requirements: mp4, mov; resolution: 480p, 720p; each video must be 2-15 seconds; aspect ratio (width / height): [0.4, 2.5]; width and height length (px): [300, 6000]; total pixels must fit [409600, 927408]; each video must be under 50 MB; frame rate (FPS): [24, 60].

Audio requirements: wav, mp3; each audio clip must be 2-15 seconds; each audio clip must be under 15 MB.

12. Click Assets in the video node

13. Select an asset group to view

14. Choose the virtual portrait you want to use

15. Use the compliant image in the video node

The video node will show the compliant image that can be used.