0.9.4-Beta

Reddit thread about this https://new.reddit.com/r/StableDiffusion/comments/1h81y4c/swarmui_094_release/

Major Updates

• added support for alimama flux inpaint controlnet - see notes at #1 (comment)
• added support for BFL's Flux Tools (redux, canny, depth, fill) - see notes at https://github.com/mcmonkeyprojects/SwarmUI/blob/master/docs/Model%20Support.md#flux1-tools
• added support for Genmo Mochi 1 Text-To-Video - see notes at #1 (comment)
• SD 3.5 Large & Large Turbo, and Medium are supported, including the Large controlnets - see notes at https://github.com/mcmonkeyprojects/SwarmUI/blob/master/docs/Model%20Support.md#stable-diffusion-35-large
• Lightricks LTX-Video text2video & image2video support - see notes at https://github.com/mcmonkeyprojects/SwarmUI/blob/master/docs/Model%20Support.md#lightricks-ltx-video
• UI feature: 'prompt add button' to make it easy to use advanced prompt syntax like Segment or Region - #1 (comment)

Minor Updates

• there's now a "popover on hover with delay" option in user settings if don't like either the ? popups or the popover on hover instantly
• fleshed out the troubleshooting doc https://github.com/mcmonkeyprojects/SwarmUI/blob/master/docs/Troubleshooting.md
• added better special error detection & handling (eg Swarm can now tell you clearly in logs if a crash happened due to RAM issues and how to fix it if so)
• automatic model downloading is now improved, and can even autodownload Flux/SD35 VAE if needed and all
• new parameter VAE Tile Overlap lets you control Tiled VAE quality
• improved image metadata format and documented it https://github.com/mcmonkeyprojects/SwarmUI/blob/master/docs/Image%20Metadata%20Format.md
• there's now a server option to enable model hashes in image metadata
• there's now a Copy Raw Metadata button on images
• the Prompt Syntax doc now has pictures throughout to demo the function https://github.com/mcmonkeyprojects/SwarmUI/blob/master/docs/Features/Prompt%20Syntax.md
• the Model Support doc now has pictures to compare the models with https://github.com/mcmonkeyprojects/SwarmUI/blob/master/docs/Model%20Support.md
• there's now a nicer batch view while generating, with model icon placeholders, blur effect, loading spinner, etc. (you can toggle the load spinner via the gear menu above the batch view)
• added advanced parameter Segment Mask Oversize to allow even finer control of segmentation behavior
• new MaxPromptLines user setting to configure the visual scroll limit of the prompt box
• preset select now persists across page reloads along with other parameters
• can now uninstall extensions
• can now use nonstandard IPAdapter models by saving them to the ipadapter models folder
• as usual, many small improvements, bug fixes, etc.

Security Notice: Ultralytics Breach 2024-12-05

On 2024-12-05, for a period of several hours, a version of the "Ultralytics" python package was marked latest on pip that contained crypto-miner malware. Swarm uses this package for "YOLO" model autosegmentation features. This was marked as version 8.3.41 and 42

If the following describes you:

• You are running Linux or Mac (not windows)
• you installed Swarm on 2024-12-05, or manually updated python packages that day (Swarm does not autoupdate these normally)
• (Possibly also only if) You used YOLO models on that day (Swarm installs the package, but doesn't load or run it until you actually try to use a yolo model)

Then you may be infected with a crypto miner. This will be visible as high CPU usage from a process named ultralytics visible in top. Killing the process and deleting the executable may suffice to end its control, though manage your personal system security to your own standards and at your own risk, it's hard to offer any guarantees about whether or not they slipped anything else in.
You can read more about this attack in this thread: ultralytics/ultralytics#18027

Swarm will not autoinstall those versions anymore, and also will remove existing installs of it and give a warning in your console.

0.9.3-Beta

Been a couple months since the last version release, was busy with work on ComfyUI and a surprise trip to Tokyo, but I'm back in full force now.

Reddit announcement of this release: https://new.reddit.com/r/StableDiffusion/comments/1g2n9ya/swarmui_093_beta_release/

Major Updates

• Now supports and uses the new modern Comfy Frontend!
• Now have tooltips on all Swarm-Comfy Nodes
• added native support for GGUF models, see https://github.com/mcmonkeyprojects/SwarmUI/blob/master/docs/Model%20Support.md#gguf-quantized-models for info
• added a bulk civitai metadata lookup utility #1 (comment)
• comfy backend settings now have an option for which version of the comfy frontend should be used in the comfy workflow tab
• autocomplete user settings have been upgraded with a bunch of handy new configuration options requested by a bunch of weebs
• added an Extension Manager tab under Server #1 (comment)
• Swarm now builds as an executable, rather than using dotnet to launch the process. Everything behaves the same, and the old .dll launch works if you have custom scripts, just Swarm is uniquely identified in task manager and has an exe icon and stuff like that now. This might make Windows do that Do you want to let this app access the network? popup thing.

Minor Updates

• GenerateForeverQueueSize for multi-GPU users to abuse generate forever - added in a PR from @wottso
• improved auto webui image metadata import
• comfy backend gpu_id can now use CUDA syntax to work with the comfy multigpu node pack that someone made (I don't recommend it, but you can do it if you want)
• added native support for NF4 models
• added a button on Server->Logs to quickly submit your recent debug logs to a pastebin service, so if you have issues and are asking for help you can easily share clear full logs
• model downloader now has a folder selector
• new theme Solarized Light for light with a bit less eyesear - thanks to @yoinked-h
• new prompt syntax setvar and var, to allow reusing dynamic variables on the fly inside a prompt - thanks to @Targren
• new SwarmMaskGrow comfy node
• new Mask Grow advanced option under Init Image, to grow your masks automatically, can be handy in some cases
• better advanced settings for remote backends (headers/auth stuff)
• unet folder renamed to diffusion_models since nobody actually puts unets in there
• starting to build up a list of extensions https://github.com/mcmonkeyprojects/SwarmUI/blob/master/docs/Extensions.md see also the discord extensions channel
• swarm internal option for 16bit png file output (high precision data unlikely to be useful)
• relatively robust and documented webhook firing engine https://github.com/mcmonkeyprojects/SwarmUI/blob/master/docs/Features/Webhooks.md
• clicking wildcards while you have the prompt or negative prompt input selected now inserts the wildcard where you had selected
• you can now list multiple model paths with ; separators in server config
• edit model metadata now lets you search civitai using model hash
• new 'refine image' button on image menus, to quickly refine an existing image
• improved utilities around extension development
• as usual, many small improvements, bug fixes, etc. - over 200 commits since the last release!

0.9.2-Beta

Reddit post for this release https://new.reddit.com/r/StableDiffusion/comments/1eosnug/swarmui_092_beta_release/

Major Updates

• Dropdown parameters now have human-friendly names rather than just raw IDs, more info here: #1 (comment)
• Native support for Flux! #1 (comment)
• SAM2 autosegmenter for the Image Editor #1 (comment)

Minor Updates

• new docs for Extension usage https://github.com/mcmonkeyprojects/SwarmUI/blob/master/docs/Extensions.md featuring a ReActor extension published by Quaggles
• new param Mask Composite Unthresholded under advanced -> regional prompting, may yield smoother refinements for <segment:> usage (but worse content replacement)
• new server setting XLDefaultAsXL1 as a lazy hack for if you're tired of model authors breaking the metadata of XL models
• new server setting EditMetadataWriteJSON, write json instead of editing model metadata, useful for slow harddrives
• there's now an advanced parameter for ip adapter start/stop
• added prompt syntax shorthands, eg <wc:my/wildcard/name>
• added support for AuraFlow v0.1 (must select architecture manuallay atm, AuraFlow team intend to add proper model metadata in the near future)
• near parameter [FreeU] Version for FreeU - lets you select FreeU Version 2
• you can now use a negative threshold value in <segment: to invert the mask
• new setting Logs -> RepeatTimestampAfterMinutes, handy for swarm servers that are left running for very long times that want clearer timestamps when there's big gaps in logs
• added advanced param ControlNet Union Type for if you're using controlnet-union and want to manually specify the subtype
• added Perturbed Attention Guidance under Advanced Sampling
• added IP-Adapter Weight Type for if you want "prompt is more important" or "style transfer" ip-adapter settings
• tab completion of booru tags now displays the tag count
• improvements to installation, esp. AMD (amd is still cursed but it's a little better now)
• as usual, many small improvements, bug fixes, etc.

0.9.1-Beta

Reddit post for this release https://new.reddit.com/r/StableDiffusion/comments/1dy9rzr/swarmui_091beta_release/

Major Updates

• added support for PixArt Sigma MS XL 2 https://github.com/mcmonkeyprojects/SwarmUI/blob/master/docs/Model%20Support.md#pixart-sigma
• Moved to a fully independent repo (hi, this one)
• now have full support for SD3 controlnets (InstantX) #1 (comment)
• Started a doc for troubleshooting issues https://github.com/mcmonkeyprojects/SwarmUI/blob/master/docs/Troubleshooting.md and opened a help-forum channel on the Discord
• added support for graphics tablets with dynamic pen pressure values in the Edit Image interface, currently just for brush size dynamics. If you're an artist that uses a graphics tablet pen and wants to see expansion on this feature, please @ me on the Swarm discord

Minor Updates

• removed the legacy stability api backend
• civitai now has an API key option under user settings for civit autodownloads of authentication-only files, thanks to @maedtb
• added initial CFG++ support via sampler euler_cfg_pp_regular comfy native now
• Remote swarm backends can now have an "Authorization" header applied (as some remote servers require this)
• the github issues page here now has issues templates
• made the "refiner" group more clear in its purpose (it's also upscaler and more) and added some more param-group-docs
• tab title now updates to show the number of current pending generations, so you can see progress even when tabbed out
• new user setting AutoCompleteSuffix to have autocompletions autoappend commas or things like that
• segment yolo now sorts left to right to give better control over selecting characters individually
• controlnets now autodetect canny,depth,pose,scribble from just filename cues if there's no metadata for it (ie if you haven't specified a preprocessor manually)
• added user setting ButtonsUnderMainImages to customize which buttons appear when you open an image vs. which get moved to 'More'
• majorly upgraded the Model Downloader utility, see #44 for more
• added server setting RecycleDeletedImages which you can use to send images to the recycle bin instead of full file deletion
• added Zero Negative parameter under Sampling, which may sometimes improve quality on SD3
• as usual, many small improvements, bug fixes, etc.

0.9.0-Beta

SwarmUI Independent Edition, version 0.9.0 Beta.

This is the first version ID as an independent repository.

Swarm Legacy Beta Pre-Release

This is the final Swarm files from the Stability version of Swarm before SwarmUI became independent.