Automatic 1111, the popular stable diffusion neural network interface, has been updated. Version 1.7 what’s new, how to update?

Automatic 1111, the popular stable diffusion neural network interface, has been updated. Version 1.7 what’s new, how to update?

Hello everyone, while there is only talk about Fooocus’e in the community, the automation has been updated. There are a lot of changes, but nothing that is straight wow or cool, a lot of necessary little things, carefully translated changelog below. And at the end, how will the instructions be updated, if you suddenly did not know or forgot.

Features (Features):

  1. Reworked settings tab: added search field, added categories, UI settings page divided into many sections.

  2. Support altdiffusion-m18.

  3. Support for inferences of LyCORIS GLora networks.

  4. Added lora-embedding package system.

  5. Option to move the request (prompt) from the top row of generation parameters.

  6. Added support for SSD-1B.

  7. Support for inferences of OFT networks.

  8. Script metadata and DAG sorting mechanism.

  9. HyperTile optimization support.

  10. Added support for SD 2.1 Turbo.

  11. The Train->Preprocessing tab has been removed and all its functions have been moved to the Extras tab.

  12. Initial IPEX support for Intel Arc GPUs.

Minor changes:

  1. Ability to read model hash from images in img2img batch processing mode.

  2. Added option to align with sgm repository sampling implementation.

  3. Additional field to view lora metadata: ss_output_name.

  4. Added a button in settings to calculate all SD checkpoint hashes.

  5. Added a button to copy the request to the style editor.

  6. Added –skip-load-model-at-start option.

  7. Infotext is now written to a GIF image.

  8. Reading infotext from GIF images.

  9. Possibility to configure the initial state of InputAccordion in ui-config.json.

  10. Editing skip sections for editing ctrl+up/ctrl+down requests.

  11. Preventing accidental closing of pop-up dialogs.

  12. Added an option to play a sound message.

  13. Show a preview of the image in the full-screen image viewer, if available.

  14. Support for webi.settings.bat.

  15. Option to disable stack trace printing when ctrl+c is pressed.

  16. Start/restart generation with Ctrl (Alt) + Enter.

  17. Updated the prompts_from_file script to combine records with a common prompt.

  18. Added a visible checkmark to InputAccordion.

  19. The option hides all txt2img/img2img options in accordion.

  20. Added ‘Path’ sorting option for Extra network cards.

  21. Enable query hotkeys in the style editor.

  22. Option to display the results of img2img batch processing in the user interface.

  23. Update infotext: option to ignore certain infotext fields, option not to include VAE in infotext, explanation on infotext settings page, moved some options to infotext settings page.

  24. Added FP32 fallback support on sd_vae_approx.

  25. Support for XYZ scripts / splitting the hires path from unet.

  26. Ability to use multiple CSV style files.

Extensions and API (Extensions and API):

  1. Update to version 3.41.2.

  2. API support for the list of installed extensions.

  3. Update pnginfo API to return a dictionary with parsed values.

  4. Adding ExtraNoiseParams latent noise for the callback.

  5. Display date and time extension in UTC.

  6. Option to choose the method of combining hires fix and refiner.

  7. Enabling the program version in response info.

  8. sd_unet support for SDXL.

  9. Patch DDPM.register_betas to use given_betas in yaml model.

  10. xyz_grid: add prepare function.

  11. Ability to use multiple localization files with one language in extensions.

  12. Adding onEdit function for js and refactoring token-counter.js to use it.

  13. Fix key bug when handling override_settings keys.

  14. Ability for extensions to return user data via api in response.images.

  15. *Call state.jobnext() before postproces()**.

  16. Option to set the sound message volume.

  17. Ruff update to 0.1.6.

  18. Adding a callback to create a Block.

  19. Catch uncaught exception of UI build scripts.

  20. Using the extension name to identify the installed index extension.

  21. Update is_installed() from launch_utils.py to fix reinstallation of already installed packages.

Bug Fixes:

  1. Fix for pix2pix giving bad results.

  2. Fixed bug on the default settings page when the main UI tabs are hidden.

  3. Fixed a bug where some additional networks would be disabled if both lora: and lyco: were in the query.

  4. Fixed UI reload feature: If you reload the UI in one tab, the other open tabs will no longer stop working.

  5. Avoid duplicating the resize handler.

  6. Minor bug: vae resolve bug fix.

  7. Hides the broken image cropping tool.

  8. No hidden selections are displayed in the list for XYZ script.

  9. Fix edit styles dialog if open in img2img and txt2img tabs.

  10. Hiding –gradio-auth and –api-auth values ​​from /internal/sysinfo report.

  11. Adding insufficient infotext for RNG in options.

  12. Fix message not playing when embedded webi tab is inactive.

  13. Compliance with the –skip-install option for extension installers.

  14. Don’t print empty stdout in extension installers.

  15. Getting the progress bar to display properly in the extensions tab.

  16. Preserving the order in the checkpoint list when loading a model that does not have a checksum.

  17. Fixing inpainting of models in txt2img, creating black images.

  18. Fix regular expression of generation parameters.

  19. Fix img2img batch output directory with script.

  20. Bug fix with sigma min/max overrides.

  21. More accurate testing to enable cuDNN benchmark on 16XX cards.

  22. Do not use multicond parser for negative request counter.

  23. Fix data-sort-name containing spaces.

  24. Updating information on the correct tab when editing metadata.

  25. Fixes for viewing/editing metadata.

This is a full translation of the changelog version 1.7.0 for Automatic1111. The English version can be found on the project’s git.

How to update?

Add a launch argument git pull in your webui-user.batfor this, just open the file in notepad, if you don’t want to update every time you start (I recommend leaving it), you can delete it from the file after updating.

\

If you have the portable version, it might have it update.batthen activate it to update.

And that’s all for me, I don’t seem to hear in the chats that something breaks or doesn’t work after the update, I’m fine too, but if you don’t want to risk it, wait a week until the hotfix is ​​released, if you need one.

I talk more about neural networks on YouTube, Telegram and Busta. I will be glad for your subscription and support. He hugged everyone.

Related posts