Skip to content

Changelog

What's new on Layer

Product updates, new models, improvements, and fixes — shipped continuously.

AI image editing, batch selection, and new video models

NewModelsFixed

AI image editing, batch selection, and new video models

New Features

  • Prompt Editing: You can now refine images by editing their prompts. Use Refine → Prompt Edit to tweak image details or provide reference images - perfect for tasks like changing outfits on a character.
  • Batch Selection in Activity, Drive, and Session: Select multiple items to refine them in bulk, saving time and effort.
  • Seed Support for Video Models: For models that support it, you can now specify a seed to control video generation results.
  • Reference Type Memory: The system now remembers your previously selected reference type to streamline repeated tasks.
  • Manual Date Entry in Activity: Directly type in dates for faster and more precise control.
Fixes
  • Fixed an issue where CMD + Enter didn't submit forms in Refine.
  • Fixed a bug where zoom controls appeared on small devices, despite being unusable.
  • Corrected video duration calculation for more accurate playback timing.
  • Fixed an issue where Minimax wasn't forging at the correct ratio.
  • Fixed a bug where Forge errors weren't being displayed when failures occurred immediately.
  • Resolved a rounding issue with decimal values in the density slider.
  • Fixed a layout bug where the uploaded files bar in Drive was clipped by the sidebar info bar.
New models2
GPT Image 1Image
OpenAI

Powerful, versatile OpenAI image model for creative and professional apps.

Magi DistilledVideo
Sand AI

Fast, efficient open-source I2V model. Animates still images using an autoregressive approach.

New Canvas editor, Activity feed, and batch Refine

NewImprovedFixed

New Canvas editor, Activity feed, and batch Refine

New Features

  • Canvas in Session: You can now open images directly in Session without disrupting your workflow. Your earlier assets have been migrated into new Sessions. Starting next week, you'll be able to filter your Canvas creations via the new Activity feature.
  • Activity: A new way to browse your past generations across Sessions. Art directors and admins can filter by user to keep tabs on creative output.
  • Batch Operations for Refine: You can now upscale, vectorize, or remove backgrounds from multiple assets at once.
  • Reference Copying: Copy assets directly from the Session feed to reuse as references.
  • Improved Filepicker: You can now Double-click items to quickly select them.
  • Updated Session Feed Tiles: New visual tiles make browsing more intuitive.
  • Consistent UX: Unified look and feel across 2D, 3D, and Video experiences.
Fixes
  • Fixed pose references not applying correctly.
  • Improved sharpness of poser tool results.
  • Fixed video image references not displaying.
  • Resolved sporadic crash when switching modes in Session.
  • Fixed SCIM integration defaulting incorrectly for new enterprise workspaces.
New models1
FramepackVideo
Layer

Highly efficient, open-source I2V model that generates video by predicting the next frame.

New File Picker, Kling 2.0 for video, and Prompt Genie

NewModelsFixed

New File Picker, Kling 2.0 for video, and Prompt Genie

New Features

  • Released a new consolidated File Picker system and Reference selection flow. These will serve as the foundation for more advanced features in the future.
  • Launched Prompt Genie, a tool for enhancing prompts with guided improvements. Check out our docs for more details.
  • Improved reference types support for the FLUX 2D model.
  • Added workspace controls to disable automatic addition of new styles.
Fixes
  • Improved visibility of selected modalities in workspace settings.
  • Fixed an issue where disabled actions appeared in Forge and Drive views.
  • Fixed CMD + Enter not submitting forms.
  • Fixed discrepancies between user management and SCIM integration.
  • Fixed an issue where prompt enhancement altered the prefix & suffix.
  • Fixed suspend user option incorrectly appearing for SCIM-enabled workspaces.
  • Fixed an issue preventing posts to the Style Gallery.
Internal Improvements
  • Replaced moment.js with date-fns.
  • Improved database monitoring.
New models3
Kling v2.0 MasterVideo
Kling

Master-grade video model. Enhanced realism and physics simulation for cinematic, high-impact clips.

Tripo v2.53D
Tripo

Incremental 3D update. Refined performance, improved mesh topology, and texture fidelity.

Hunyuan 3D v2 Mini3D
Tencent

Lightweight, efficient 3D version optimized for less powerful hardware. Delivers good quality assets.

New branding, video upscaling, and bug fixes

NewImprovedFixed

New branding, video upscaling, and bug fixes

New Features

  • Revamped official Layer branding and refreshed visual design for the app homepage.
  • Added support for video upscaling (more on this soon!).
  • Added the ability for users to update profile pictures.
  • Improved notifications on the Style page to better indicate the maximum number of styles allowed.
Fixes
  • Fixed an issue where users could be manually added to a workspace even when SCIM integration was enabled.
  • Fixed a bug where saving a JPEG image incorrectly defaulted to PNG file extension.
  • Resolved an issue where Forges occasionally failed without clear errors, leading to a broken UI state.
  • Refinement icons now display correctly.
  • Fixed an issue where closing the Drive dialog also closed the breadcrumbs unexpectedly.
  • Flattening the Drive directory now works with a single button click.
  • Resolved transparency issues when generating images with SDXL.
  • Fixed a bug where assets could be exported before saving was complete, causing errors.
  • Fixed a problem where the 3D viewer sometimes failed to open.
  • Corrected the label for Tripo 2.5 Style, which was still showing as 2.0 in the app.

Custom style controls and prompt management

NewImprovedFixed

Custom style controls and prompt management

New Features

  • Introduced prefix and suffix support for styles, enabling style admins to "define default values to improve output quality."
  • Added the ability to view the full prompt in the session feed through an expandable menu.
Fixes
  • Fixed an issue where Video length settings would revert to their previous value.
  • Resolved a bug where unfinished Forges were not marked as "Failed," causing infinite timers.
  • Fixed a problem with deleting items from the session feed.
  • Addressed a display issue where non-functional refinement options appeared for SVGs.
  • Fixed an error that occurred when exporting Canvas assets with invalid layer names.
  • Resolved a bug where users would see errors when browsing featured styles.
  • Fixed an issue where the Drive view didn't refresh to show newly uploaded files.
  • Addressed an overflow issue with user emails in the user menu.

New product tour and Layer Library

NewImprovedFixed

New product tour and Layer Library

New Features

  • Added a new product tour for new customers.
  • Released Layer Library, a new improved experience to improve the discoverability of assets.
Fixes
  • Fixed an issue where renaming a layer and hitting enter resulted in reforge on canvas.
  • Fixed an issue with OIDC users could not login.
  • Improved filenames for assets generated via forging.

UX updates and the Hunyuan 3D 2.0 model

NewModelsFixed

UX updates and the Hunyuan 3D 2.0 model

New Features

  • Added a "My Favorites" tab in Style Studio for easy access to favorite styles.
  • Included a link to the documentation Members page on Roles.
  • Launched a new and improved Session feed with enhanced performance and user experience.
  • Updated the User Settings page with a new design and features.
Fixes
  • Intercom pop-up now includes an "X" button for easier closure.
  • Resolved an issue where preview images were not displaying correctly on Canvas due to a restrictive security policy.
  • Fixed a bug where liking an image with multiple selections only liked the image being clicked.
  • Addressed an issue where the styles area remained sticky at the top of the Forge form.
  • Fixed thumbnail generation for 3D assets.
  • Corrected an issue where Refine options were unclickable in the context menu.
  • Resolved a tilability issue when Fooocus styles were selected.
  • Fixed an issue where the video forge didn't correctly apply aspect ratio values.
  • Addressed a bug where the batch size was incorrectly set to 1 for 2D styles.
New models1
Hunyuan 3D v23D
Tencent

Powerful, open-source 3D model producing high-res, textured 3D objects from text or image inputs.

New onboarding, workspace management, and video options

NewImprovedFixed

New onboarding, workspace management, and video options

New Features

  • Introduced a new onboarding experience for users.
  • Improved the invitation flow for workspace members, now indicating whether they already have an account or need to create a new one.
  • Added a link to the Video Prompting Guide in the Video Generation section.
  • Added portrait and square aspect ratios for Veo 2.
Fixes
  • Fixed an issue where assets were automatically created in the drive, preventing unnecessary clutter.
  • Users can still choose to save Canvas assets explicitly into the drive.
  • Fixed source of some crashes in Canvas tool.
  • Resolved an issue where LoRA was not being applied correctly for BRIA.
  • Fixed an issue with the Contact Form on the website.
  • Addressed an issue where the batch size was incorrectly resetting to 1.
  • Fixed a bug preventing users from proceeding in the style stepper under certain conditions.

Vectorization, Refine Tool, and auto-enhanced prompts

NewImprovedFixed

Vectorization, Refine Tool, and auto-enhanced prompts

New Features

  • Added an option to control opt-in into marketing emails in user settings.
  • Added improved rate-limiting to the application to ensure that bad actors cannot introduce regressions to the wider system.
  • Added refine tool that now supports vectorization, remove background, and upscaling for images.
  • Added ability to filter out suspended users on Members page and added new sorting options. The page also now shows the total number of users in the workspace.
  • Added ability to change the duration for video generations where applicable.
  • Added an alert banner that appears in the application when there is an active incident to notify users.
  • Added support for "auto-translate" and "auto-enhance" to the prompt form.
  • Improved performance on Canvas by moving the storage of temporary assets to not be part of the primary canvas object.
Fixes
  • Fixed an issue where filenames with spaces resulted in broken behavior due to URL encoding.
  • Fixed an issue when opening up the Plans page.
  • Fixed source of some crashes in Canvas tool.
  • Fixed an issue where if the user was suspended, they would get "locked" when returning back to the app.
  • Introduced a post mortem process and template that now gets associated with incidents on our public status page.
New models2
Wan 2.1Video
Alibaba

State-of-the-art video model. Generates detailed, stylistically diverse clips with fluid motion.

Kling v1.6 ProVideo
Kling

Powerful video model for high-fidelity, imaginative content with complex character motion.

Video arrives on Layer, batch generations, and prompting

NewModelsImproved

Video arrives on Layer, batch generations, and prompting

New Features

  • Added Video generation option for Pro, Studio, and Enterprise customers.
  • Added ethically trained, copyright compliant Shutterstock/Nvidia Edify 3D model for Enterprise customers.
  • Added description mode for Realtime canvas, and improved generation quality of Draw references.
  • Added support for batch size, allowing users to Forge up to 16 images at a time.
  • Added support for users to generate prompts based on the uploaded reference image.
  • Improved prompt suggestions by making it more of an autocomplete.
  • Improved speed of generations via parallelization, achieving way faster Forge speeds.
  • Improved file options across Forge Sessions, Drive, and History to offer a more uniform and consistent experience for File post-processing.
Fixes
  • Fixed small bugs on realtime canvas.
  • Fixed a couple of long outstanding performance issues with Layer, and now the application should work way smoother than before.
New models6
Hunyuan Video 1.5Video
Tencent

High quality open-source video model from Tencent Hunyuan.

Minimax Video 01Video
MiniMax

Accessible video model. Generates engaging 720p clips from text with good visual consistency.

Minimax Video 01 LiveVideo
MiniMax

Specialized video model. Optimized for a dynamic, live-action feel with naturalistic camera work.

Veo 2Video
Google

Legacy video model generating high-definition, long-form cinematic video content.

Rodin Gen-1.53D
Hyper3D

Advanced 3D generation model creating high-quality, textured T-pose avatars from a single image.

Trellis3D
Microsoft

Open-source 3D model creating high-quality objects with realistic materials and geometry from text.

Ready to scale your creative production?

Start for freeContact sales