v5.4.1
This release includes support for SD 3.5, plus a number of fixes and improvements.
We pulled v5.4.0 due to issues with model installation and loading. This release (v5.4.1) resolves those issues and includes all changes from v5.4.0.
Enhancements
- Support for SD 3.5 Medium and Large. You can download them in the starter models.
- Moved
Denoising Strength
slider to top of Layers panel with updated UI and info popover. - Update viewer styling to have a menubar-ish header. Also used for image comparison.
- Layer preview tooltip with larger sized preview.
- Empty state for Control Layers, guiding user to upload an image, drag an image from gallery or start drawing.
- "Simple" mode for control layers filtering, triggered when you select a control model. This automatically selects and processes the default filter for that model, if a default exists. If the user clicks
Advanced
, they get the full filter settings UI. Control Layers now start with no model selected, implicitly directing users into this flow. - Updated default control weight to 0.75 and end step % to 0.75. Updated label for the
Balanced
control mode, indicating it is the recommended setting. - The FLUX Denoise node now has a flag to skip adding noise to input latents. This is useful for switching models mid-generation and other fancy denoising techniques. Thanks @JPPhoto!
- Migrated the UI's drag-and-drop functionality to a new library, improving the overall performance of dnd. It also allows us to support external dnd. For example, dragging images from the OS directly into the canvas is now possible (not implemented in this release).
- Canvas layers may be sorted via drag-and-drop.
- When graph building fails, you will see an error toast. Previously, it only logged a message to the browser's JS console.
- Updated model warnings on Upscaling tab to not be misleading.
- More FLUX LoRA format support.
- Show alert over canvas/viewer with invocation progress event messages.
- Tweak gallery image selection styles to better differentiate between hovered and selected.
Output Only Masked Regions
renamed toOutput only Generated Regions
and now enabled by default.
Fixes
- Canvas progress images do not clear when canceling generation after at least one image has been staged.
- Tooltips on boards list stay open when scrolling, potentially causing the whole app to scroll.
- Saving canvas to gallery does not create a new image in gallery.
- Applying a filter could erase or otherwise change a layer's data unexpectedly, causing a range of user-facing generation issues.
- Unable to queue graphs with the Segment Anything node when its inputs were provided by connection.
- Unable to load a workflow from file when using the three-dots menu.
pip
downloadstorch
twice. This didn't cause any application issues - just a waste of time and bandwidth. We pinnedtorch
to<2.5.0
to preventpip
's dependency resolver from getting confused.mediapipe
install issue on Windows, related to its latest release. We pinnedmediapipe
to a known working version.- CLIP Vision error when using FLUX IP Adapter.
- Fit bbox to layers math could result in a slightly-too-large bbox.
- Outdated link to FLUX IP Adapter. Thanks @RadTechDad!
- Force a compatible precision for FLUX VAEs to prevent black outputs.
Translations
We have had some issues communicating with "walk-in" translators on Weblate, resulting in translations being changed when they are already correct. To mitigate this, we are trying a more restricted Weblate translation setup. Access to contribute translations must be granted by @Harvester62. Please @ them in the #translators channel on discord to get access.
We apologize for any inconvenience this change may cause. Thanks to all our translators for their continued efforts!
- Updated German. Thanks @rikublock!
- Updated Chinese (Simplified). Thanks @youo0o0!
- Updated Italian. Thanks @Harvester62 & @dakota2472!
- Updated Spanish. Thanks gallegonovato (weblate user)!
- Updated Vietnamese. Thanks Linos (weblate user)!
- Updated Japanese. Thanks @GGSSKK!
Internal
- Simplified parameter schema declarations. Thanks @rikublock!
- Simplified dnd image and gallery rendering, resulting in improved performance.
Installation and Updating
To install or update, download the latest installer and follow the installation instructions
To update, select the same installation location. Your user data (images, models, etc) will be retained.
What's Changed
- Pin torch to <2.5.0 to prevent unnecessary downloads by @ebr in #7247
- fix(ui): misc fixes by @psychedelicious in #7252
- fix(nodes): relaxed validation for segment anything by @psychedelicious in #7253
- fix(ui): load workflow from file by @psychedelicious in #7254
- fix(deps): pin mediapipe strictly to a known working version by @ebr in #7258
- Ryan/sd3 diffusers by @maryhipp in #7222
- chore: bump version to v5.4.0 by @psychedelicious in #7259
- fix: Never throw an exception when finding the clip variant type by @brandonrising in #7263
- fix(ui): add some feedback while layers are merging by @maryhipp in #7265
- feat(api,ui): allow Whats New module to get content from back-end by @maryhipp in #7268
- fix(ui): fix link for infill method popover by @maryhipp in #7269
- update Whats New for 5.3.1 by @maryhipp in #7272
- fix(ui): bad merge conflict by @maryhipp in #7273
- fix: get_clip_variant_type should never return None by @brandonrising in #7275
- feat(ui): denoising strength UI by @maryhipp in #7262
- chore(gh): update pr template w/ reminder for what's new copy by @psychedelicious in #7274
- feat(ui): simplify parameter schema declaration by @rikublock in #7281
- feat(ui): new user ux improvements by @psychedelicious in #7279
- fix: Avoid downloading unsafe .bin files if a safetensors file is ava… by @brandonrising in #7282
- fix: Look in known subfolders for configs for clip variants by @brandonrising in #7283
- chore: release v5.4.1rc1 by @psychedelicious in #7285
- Update flux_denoise.py by @JPPhoto in #7270
- fix(app): silence pydantic protected namespace warning by @psychedelicious in #7291
- chore(ui): migrate dnd library by @psychedelicious in #7288
- chore: bump version to v5.4.1rc2 by @psychedelicious in #7293
- fix(ui): FLUX IP adapter clip vision model error by @psychedelicious in #7292
- fix(ui): excessive toasts when generating on canvas by @psychedelicious in #7294
- fix(ui): fit bbox to layers math by @psychedelicious in #7297
- fix(ui): restore missing image paste handler by @psychedelicious in #7298
- feat(ui): update warnings on upscaling tab based on model arch by @psychedelicious in #7299
- feat(ui): clarify denoising strength badge text by @psychedelicious in #7304
- Updated link to flux ip adapter model by @RadTechDad in #7309
- Flux Vae broke for float16, force bfloat16 or float32 were compatible by @Vargol in #7213
- Fix FLUX diffusers LoRA models with no
.proj_mlp
layers by @RyanJDick in #7313 - SD3 Image-to-Image and Inpainting by @RyanJDick in #7295
- (ui): update en string for full IP adapter by @maryhipp in #7312
- feat(ui, worker): add invocation progress events to model loading by @maryhipp in #7286
- feat(ui): tweak image selection/hover styling by @psychedelicious in #7302
- feat(ui): update output only masked regions by @psychedelicious in #7303
- fix(ui): image uploading handling by @psychedelicious in #7315
- fix(ui): unable to use text inputs within draggable by @psychedelicious in #7316
- fix(ui): multiple selection dnd sometimes doesn't get full selection by @psychedelicious in #7317
- fix(ui): remove progress message condition for canvas destinations by @psychedelicious in #7319
- ui: translations update from weblate by @weblate in #7242
New Contributors
- @RadTechDad made their first contribution in #7309
Full Changelog: v5.3.1...v5.4.1