1.10.0

[AUTOMATIC1111]

Features:

• A lot of performance improvements (see below in Performance section)
• Stable Diffusion 3 support (#16030, #16164, #16212)
  • Recommended Euler sampler; DDIM and other timestamp samplers currently not supported
  • T5 text model is disabled by default, enable it in settings
• New schedulers:
  • Align Your Steps (#15751)
  • KL Optimal (#15608)
  • Normal (#16149)
  • DDIM (#16149)
  • Simple (#16142)
  • Beta (#16235)
• New sampler: DDIM CFG++ (#16035)

Minor:

• Option to skip CFG on early steps (#15607)
• Add --models-dir option (#15742)
• Allow mobile users to open context menu by using two fingers press (#15682)
• Infotext: add Lora name as TI hashes for bundled Textual Inversion (#15679)
• Check model's hash after downloading it to prevent corruped downloads (#15602)
• More extension tag filtering options (#15627)
• When saving AVIF, use JPEG's quality setting (#15610)
• Add filename pattern: [basename] (#15978)
• Add option to enable clip skip for clip L on SDXL (#15992)
• Option to prevent screen sleep during generation (#16001)
• ToggleLivePriview button in image viewer (#16065)
• Remove ui flashing on reloading and fast scrollong (#16153)
• option to disable save button log.csv (#16242)

Extensions and API:

• Add process_before_every_sampling hook (#15984)
• Return HTTP 400 instead of 404 on invalid sampler error (#16140)

Performance:

• [Performance 1/6] use_checkpoint = False (#15803)
• [Performance 2/6] Replace einops.rearrange with torch native ops (#15804)
• [Performance 4/6] Precompute is_sdxl_inpaint flag (#15806)
• [Performance 5/6] Prevent unnecessary extra networks bias backup (#15816)
• [Performance 6/6] Add --precision half option to avoid casting during inference (#15820)
• [Performance] LDM optimization patches (#15824)
• [Performance] Keep sigmas on CPU (#15823)
• Check for nans in unet only once, after all steps have been completed
• Added pption to run torch profiler for image generation

Bug Fixes:

• Fix for grids without comprehensive infotexts (#15958)
• feat: lora partial update precede full update (#15943)
• Fix bug where file extension had an extra '.' under some circumstances (#15893)
• Fix corrupt model initial load loop (#15600)
• Allow old sampler names in API (#15656)
• more old sampler scheduler compatibility (#15681)
• Fix Hypertile xyz (#15831)
• XYZ CSV skipinitialspace (#15832)
• fix soft inpainting on mps and xpu, torch_utils.float64 (#15815)
• fix extention update when not on main branch (#15797)
• update pickle safe filenames
• use relative path for webui-assets css (#15757)
• When creating a virtual environment, upgrade pip in webui.bat/webui.sh (#15750)
• Fix AttributeError (#15738)
• use script_path for webui root in launch_utils (#15705)
• fix extra batch mode P Transparency (#15664)
• use gradio theme colors in css (#15680)
• Fix dragging text within prompt input (#15657)
• Add correct mimetype for .mjs files (#15654)
• QOL Items - handle metadata issues more cleanly for SD models, Loras and embeddings (#15632)
• replace wsl-open with wslpath and explorer.exe (#15968)
• Fix SDXL Inpaint (#15976)
• multi size grid (#15988)
• fix Replace preview (#16118)
• Possible fix of wrong scale in weight decomposition (#16151)
• Ensure use of python from venv on Mac and Linux (#16116)
• Prioritize python3.10 over python3 if both are available on Linux and Mac (with fallback) (#16092)
• stoping generation extras (#16085)
• Fix SD2 loading (#16078, #16079)
• fix infotext Lora hashes for hires fix different lora (#16062)
• Fix sampler scheduler autocorrection warning (#16054)
• fix ui flashing on reloading and fast scrollong (#16153)
• fix upscale logic (#16239)
• [bug] do not break progressbar on non-job actions (add wrap_gradio_call_no_job) (#16202)
• fix OSError: cannot write mode P as JPEG (#16194)

Other:

• fix changelog Fix method 2 : patch__pkg_resources__packaging #15883 -> Fix method 1 : pin Setuptools==69.5.1 #15882 (#15907)
• ReloadUI backgroundColor --background-fill-primary (#15864)
• Use different torch versions for Intel and ARM Macs (#15851)
• XYZ override rework (#15836)
• scroll extensions table on overflow (#15830)
• img2img batch upload method (#15817)
• chore: sync v1.8.0 packages according to changelog (#15783)
• Add AVIF MIME type support to mimetype definitions (#15739)
• Update imageviewer.js (#15730)
• no-referrer (#15641)
• .gitignore trace.json (#15980)
• Bump spandrel to 0.3.4 (#16144)
• Defunct --max-batch-count (#16119)
• docs: update bug_report.yml (#16102)
• Maintaining Project Compatibility for Python 3.9 Users Without Upgrade Requirements. (#16088, #16169, #16192)
• Update torch for ARM Macs to 2.3.1 (#16059)
• remove deprecated setting dont_fix_second_order_samplers_schedule (#16061)
• chore: fix typos (#16060)
• shlex.join launch args in console log (#16170)
• activate venv .bat (#16231)
• add ids to the resize tabs in img2img (#16218)
• update installation guide linux (#16178)
• Robust sysinfo (#16173)
• do not send image size on paste inpaint (#16180)
• Fix noisy DS_Store files for MacOS (#16166)

---

This discussion was created from the release 1.10.0.

[guythis31]

Thanks!

[tangze-Asuka]

good

[bobba84]

Yeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeahhg!

[CalculonPrime]

Aside from performance, is there any significant advantage of SD3 over SDXL? Has the "understanding" of prompts been improved so that finally when you specify two people they don't get mixed together? Because I see the downsides - only Euler sampler is supported, and also I see that if you choose 512x512 resolution, the output content seems to get cropped, as if it didn't like the output being that small. Going to 512x768 seems to avoid this.

Also the official SD3 checkpoint can't draw (i.e. "doesn't know") Britney Spears or Madonna. How do you release an official checkpoint that can't draw those iconic figures? The official SDXL checkpoint can do it. SD3 can draw Taylor Swift, but it seems it wasn't trained with the same depth as SDXL.

[bobba84]

I've tried a few merges from civit (now that they're allowed) that seem miles better than the standard SD3 checkpoint.

[CalculonPrime]

Which models are you referring to? I tried "Realistic Freedom" for SD3 and it seemed to suffer from the same issues affecting the base SD3 model - limitations on the sampling method, lack of knowledge of common celebrities I want to draw, clipping when the canvas is set to 512x512,...

[WhiteZeroX]

I can't seem to render anything with SD3. All I get is garbage. When I first loaded SD3 medium, it downloaded the 2 needed clip models.
EDIT: Fixed by disabling "SGM noise multiplier" in settings.

[bobba84]

Are you using the euler sampler?

[WhiteZeroX]

Are you using the euler sampler?

I've tried euler and just about every other samplers and scheduler but get the same result.

[WhiteZeroX]

I found the issue by clearing my config file, generations were then fine with SD3. I then compared two image PNG info, and traced the issue back to the fact that I had "SGM noise multiplier" enabled in settings. Turning that off ultimately fixes this problem.

[NulliferBones]

Is there anything changed from the RC?

[a3nima]

Can it be that in img2img inpainting the image size is not detected and set from the original image (when sent from text2img to inpainting) ? I have to do this by hand now? Why is that change?

[txiaou]

FLUX!PLEASE!

[txiaou]

flux 谢谢啦 加油

[Godblesskhuy]

raise RuntimeError(

RuntimeError: Torch is not able to use GPU; add --skip-torch-cuda-test to COMMANDLINE_ARGS variable to disable this check
Press any key to continue . . .
this problem how to fix

[Godblesskhuy]

Plz help me