Skip to content

Releases: parlance-zz/g-diffuser-bot

G-Diffuser AIO (Dev) Installer (Windows + Linux)

12 Dec 00:08
Compare
Choose a tag to compare

This (all-in-one) installation / will setup a new G-Diffuser installation that will update from the aio-dev branch of the g-diffuser project. Use this installer if you want to test new features, but there may still be bugs and other problems to run into.

After downloading, run 'install_or_update.cmd' first, then update the 'config' file with your hugging face token, then finally use 'run.cmd' to start your G-Diffuser system. There will now be a "system tray" icon in the bottom right corner of your screen when the server is running, and you can use it to show the server console as well as launch G-Diffuser programs. Please see the full installation instructions. Note - There are now bash scripts for Linux users in the 'linux' folder.

Please be aware that 40GB of free space is needed to download and install the required files after installation, and default settings require at least 8GB of VRAM. Please see the full system requirements.

Partial list of changes:

  • Xformers now included again There seems to be a problem with older versions of Xformers and SD2, removed for now...
  • Bash scripts to start the server and g-diffuser components for Linux users (and a helper script for both platforms if you need to enter the "env" in command prompt)
  • Lots of changes to the CLI, see the updated CLI help
  • zoom_maker script now auto-resumes, doesn't need a starting image, and has an interactive / "cherry-picker" mode.
  • zoom_composite can now write raw video frames as png instead of encoding a compressed video.
  • scripts like zoom_maker and zoom_composite can now properly take keyword args, and ALL parameters used will be saved in the corresponding args file to make sure your settings are never lost. Saved settings can be also be re-used by specifying a saved settings file.
  • Discord bot split into 2 commands, one for txt2img and one for img2img / in-painting and out-painting
  • "Expand" no longer destroys the image's original mask / alpha channel, if it had one
  • "Hiresfix", similar to what users may be familiar with from other popular frontends is now available, and enabled by default
  • SD2.0/SD2.1 support (enabled by default)
  • SD2 "V-pred" support

Download G-Diffuser AIO Installer (Windows + Linux)

G-Diffuser AIO Installer (Windows 10+ 64-bit)

24 Nov 00:32
Compare
Choose a tag to compare

This is the first version of G-Diffuser that uses the AIO (all-in-one) installation / update framework used by other Stable Cabal projects.

After downloading, run 'install_or_update.cmd' first, then update the 'config' file with your hugging face token, then finally use 'run.cmd' to start your G-Diffuser system. There will now be a "system tray" icon in the bottom right corner of your screen when the server is running, and you can use it to show the server console as well as launch G-Diffuser programs. Please see the full installation instructions.

Please be aware that 40GB of free space is needed to download and install the required files after installation, and default settings require at least 8GB of VRAM. Please see the full system requirements.

Comprehensive documentation and big changes for the CLI / scripting interface is the next thing I have on my radar at the moment, but I am also planning on creating a local (non web-app) GUI to create interesting animations and pseudo-3D outpainting sequences with an interactive interface as well.

There are probably still bugs and other things that will need to be changed. If you find a problem please report it in the project issues page. Thank you for your patience.

Download G-Diffuser AIO Installer (Windows 10+ 64-bit)

G-Diffuser Systray

GUI is coming soon(tm)