1.5.0-RC

[AUTOMATIC1111]

To use this pre-release, you must use the release_candidate branch.

To get the release_candidate branch in a new webui installation, run those commands (this will create a directory called stable-diffusion-webui):

git clone https://github.com/AUTOMATIC1111/stable-diffusion-webui.git
cd stable-diffusion-webui
git switch release_candidate
webui-user.bat

Alternatively, to switch your existing installation to release_candidate branch:

git switch release_candidate
git pull

To switch back to master branch:

git switch master

Features:

• SD XL support
• user metadata system for custom networks
• extended Lora metadata editor: set activation text, default weight, view tags, training info
• Lora extension rework to include other types of networks (all that were previously handled by LyCORIS extension)
  • Lora page now lists all models from both Lora and LyCORIS directories
  • You can still use the LyCORIS extension if you want to
• show github stars for extenstions
• img2img batch mode can read extra stuff from png info
• img2img batch works with subdirectories
• hotkeys to move prompt elements: alt+left/right
• restyle time taken/VRAM display
• add textual inversion hashes to infotext
• optimization: cache git extension repo information
• move generate button next to the generated picture for mobile clients
• hide cards for networks of incompatible Stable Diffusion version in Lora extra networks interface
• skip installing packages with pip if they all are already installed - startup speedup of about 2 seconds

Minor:

• checkbox to check/uncheck all extensions in the Installed tab
• add gradio user to infotext and to filename patterns
• allow gif for extra network previews
• add options to change colors in grid
• use natural sort for items in extra networks
• Mac: use empty_cache() from torch 2 to clear VRAM
• added automatic support for installing the right libraries for Navi3 (AMD)
• add option SWIN_torch_compile to accelerate SwinIR upscale
• suppress printing TI embedding info at start to console by default
• speedup extra networks listing
• added [none] filename token.
• removed thumbs extra networks view mode (use settings tab to change width/height/scale to get thumbs)
• add always_discard_next_to_last_sigma option to XYZ plot

Extensions and API:

• api endpoints: /sdapi/v1/server-kill, /sdapi/v1/server-restart, /sdapi/v1/server-stop
• allow Script to have custom metaclass
• add model exists status check /sdapi/v1/options
• rename --add-stop-route to --api-server-stop
• add before_hr script callback
• add callback after_extra_networks_activate
• disable rich exception output in console for API by default, use WEBUI_RICH_EXCEPTIONS env var to enable
• return http 404 when thumb file not found
• allow replacing extensions index with environment variable

Bug Fixes:

• fix for catch errors when retrieving extension index catch errors when retrieving extension index #11290
• fix very slow loading speed of .safetensors files when reading from network drives
• API cache cleanup
• fix UnicodeEncodeError when writing to file CLIP Interrogator batch mode
• fix warning of 'has_mps' deprecated from PyTorch
• fix problem with extra network saving images as previews losing generation info
• fix throwing exception when trying to resize image with I;16 mode
• fix for [Bug]: Canvas Zoom/Pan built-in extension hijacks keyboard #11534: canvas zoom and pan extension hijacking shortcut keys
• fixed launch script to be runnable from any directory
• don't add "Seed Resize: -1x-1" to API image metadata
• correctly remove end parenthesis with ctrl+up/down
• fixing --subpath on newer gradio version
• fix: check fill size none zero when resize (fixes [Bug]: {'error': 'ValueError', 'detail': '', 'body': '', 'errors': 'height and width must be > 0'} #11425)
• use submit and blur for quick settings textbox
• save img2img batch with images.save_image()

---

This discussion was created from the release 1.5.0-RC.

[RYG81]

How to update to this. I have previous version installed. When I run update.batch it says already up to date. On running, it shows version 1.4.
Please help as I am new to this.

[eniora]

Open CMD on web-ui root folder and do git switch release_candidate
You may need to run git pull afterwards.

[SunGreen777]

Pre not for AMD?
And yes, thank you for supporting AMD :)

[igor469]

I conducted a group of tests to measure the generation speed
on different models and resolutions.
Computer CPU i9-10900, RAM 64 Gb, GTX 1070 (8 Gb)
Torch 2.0.1+cu118, xformers 0.0.20, cudnn 8700
Prompt 'Cat', Sampling method 'Euler a', Sampling steps 20,
Conducted several generations in a row and averaged over time
Models___Resolution____SD_v1.5RC____ConfyUI
SD_1.5____512x512___________13s________12s
SD_2.1____768x768___________26s________25s
SDXL0.9_1024x1024_________900s________74s
How can one explain such a huge difference for the model SD XL 0.9 ?

[igor469]

Thanks, now 900s changed to 120s
This is fine?

[dhwz]

Yeah probably won't get much faster your 1070 is a bit slow. I've ~38secs on a 2070Super.
I'm using pruned fp16 base model and also fp16 fixed VAE (https://huggingface.co/madebyollin/sdxl-vae-fp16-fix/tree/main), that lowers VRAM usage a bit more.

[igor469]

O! 38s is very good.
Which file should I put in which folder SDv1.5RC?
And what time did ConfyUI?

[alfonslm]

I got an RTX since the 1080 I used to run it on had similar numbers and now I can do like 9 images during the time it took to get 1 with no issues.

[igor469]

New RTX 4080?

[Milor123]

❌ RuntimeError: mixed dtype (CPU): expect parameter to have scalar type of Float in my AMD Rx 6750 XT with ROCm 5.5 on ubuntu studio 22.04 LTS
what should i do? I do it:

git switch release_candidate
git pull

then download refiner, model base and VAE all for XL and select it. But not working.
i always get RuntimeError: mixed dtype (CPU): expect parameter to have scalar type of Float ❌

My settings are

export PYTORCH_CUDA_ALLOC_CONF="garbage_collection_threshold:0.7,max_split_size_mb:128"
export COMMANDLINE_ARGS=" --opt-split-attention --upcast-sampling --no-half-vae  --medvram --always-batch-cond-uncond --api  --theme dark  --disable-nan-check --precision full"
export ATTN_PRECISION="fp16"
export SAFETENSORS_FAST_GPU="1"
export HSA_OVERRIDE_GFX_VERSION="10.3.0"
export XFORMERS_PACKAGE=none

XL only works nice if i use the extension

[dhwz]

Which files exactly did you download?

[Milor123]

I try downloding the next files:

• sdxl_vae.safetensors (normal version) (from official repo)
• sdxl_vae.safetensors (FP16 version)
• sd_xl_refiner_0.9.safetensors (from official repo)
• sd_xl_base_0.9.safetensors (from official repo)

[dhwz]

I don't know AMD very well are you required to use these precision args?

Can you please try without --no-half-vae, --precision full and ATTN_PRECISION="fp16"
FYI
SAFETENSORS_FAST_GPU is obsolete and can safely be removed

[Milor123]

Ohh thanks bro, it works for me !!!

[munkeyflying]

Looks awesome! What's the ETA on this update?

[CaptivatingTime]

Why all of the samplers are not supported for SDXL?
Getting error when using DDIM or UniPC: "Exception: Sampler UniPC is not supported for SDXL"

[dhwz]

Code for these samplers is not yet compatible with SDXL that's why @AUTOMATIC1111 has disabled them, else you would get just some errors thrown out.

[CaptivatingTime]

Understood, thank you.

[zbulrush]

The images generated by the Loha model trained with sdxl have no effect. I've already upgraded to the latest lycoris_lora version. Is it still not supported

[dhwz]

you've to uninstall the extension, lycoris is now natively supported

[eniora]

@dhwz Is there an effective way that we can uninstall the extension other than deleting the lycoris add-on folder? would that be enough? because I heard some leftovers will be in the venv folder. Thanks.

[dhwz]

yeah just delete the extensions folder, there is no way to remove unused packages from venv, but they don't do any harm

[zbulrush]

thanks, it works for me