v0.3.2

0.3.2

Fixed

• Fixed Registry and Manager dependency metadata for Google prompt translation by declaring googletrans-py==4.0.0 in both pyproject.toml and requirements.txt.

Manual Install

Download ComfyUI-EasyUseAnima-0.3.2-manual-install.zip and extract it so the folder is installed as ComfyUI/custom_nodes/ComfyUI-EasyUseAnima/.

SHA256: 868EBDA7DD045F86B4067E9A9F1029C4478F6D9A3F5E9D56B53C217CF3C6F95D

v0.3.1

EasyUse Anima 0.3.1

This release focuses on final-stage AiO upscaling/postprocess controls and a wider bundled autocomplete dataset.

Highlights

• Added the AiO Generator final Upscale stage after Detailer and before Save, with mutually exclusive USDU and ResShift backends.
• Added USDU helper controls for automatic tile sizing, sampler settings, no-general prompt rebuilding, and per-USDU Spectrum/DCW settings.
• Added a Postprocess stage for final size fitting by max long edge or megapixels.
• Added Anima Safe PAG model patch controls.
• Added an Image Saver prompt metadata toggle.
• Added Danbooru 2025-09-01, e621 2025-09-01, and Danbooru+e621 merged autocomplete CSV sources from DraconicDragon/dbr-e621-lists-archive under Unlicense.
• Changed the default autocomplete CSV to danbooru_2025-09-01.csv; danbooru_tags_classified.csv remains available for Korean search.
• Fixed wildcard syntax reference table rendering for literal pipe syntax.

Notes

• The Danbooru+e621 merged CSV is opt-in. The upstream merge format can have future category-collision risk, so the default source remains Danbooru-only.
• Comfy Registry publication is not included here; this release prepares the checked-in Registry changelog path.

Manual Install

1. Download ComfyUI-EasyUseAnima-0.3.1-manual-install.zip.
2. Extract it into ComfyUI/custom_nodes/.
3. The extracted folder should be ComfyUI/custom_nodes/ComfyUI-EasyUseAnima/.
4. Restart ComfyUI.
5. If the web UI was already open, hard refresh the browser.

Asset SHA256

ComfyUI-EasyUseAnima-0.3.1-manual-install.zip

B77926D867826185E62E437D4A57D93902CBB8887BC06E92A03AA7212F72AD9D

Validation

• git diff --check
• Get-ChildItem -File web\js\*.js | ForEach-Object { node --check $_.FullName }
• python -m json.tool locales\ko\nodeDefs.json
• TOML/Registry metadata sanity check
• Registry scanner grep
• autocomplete source/license smoke
• python -m unittest discover -s tests passed, 296 tests
• python -m compileall -q .
• tools\check_custom_node.ps1 -Project ComfyUI-EasyUseAnima -Profile quick

EasyUse Anima 0.3.0

Changed

• Prepared the package for Comfy Registry publication by tightening the shipped archive surface with .comfyignore, documenting Registry scanner safety rules, and keeping Registry metadata aligned in pyproject.toml.
• Removed the standalone public Anima SAM3 Context and Anima SAM3 Detailer node registrations and their user-facing node docs. AiO Generator's internal SAM3 detailer path remains available through AiO Detailer settings.
• Kept optional external-provider paths safer by default: prompt translation is off unless selected, NAIA remote hosts require an explicit allow setting, and API-key environment auto-detection was removed.
• Set AiO release workflow optimization defaults to conservative values: FP16 accumulation off, SageAttention disabled, Sage compile disabled, and TorchCompile off. The v6 workflow guide markdown now explains when to enable those options manually.

Fixed

• Fixed AiO Generator Image Saver metadata so saved positive/negative prompts use Prompt Studio metadata prompt outputs when available, while generation and Mod Guidance still use the generation prompt outputs.

Validation

• Full unit tests: 274 tests OK.
• compileall, full web/js/**/*.js syntax checks, workflow JSON tests, git diff --check, Registry scanner grep, and comfy node validate passed.
• Remote v0.3.0 tag and release did not exist before publication.

Manual Install Asset

• Asset: ComfyUI-EasyUseAnima-v0.3.0-manual-install.zip
• SHA256: 2F3C1B286F910E5D45DA3F4986532BE5A8D56C7016525C51BE9FA7AB3B7E04AD

Notes

• Comfy Registry publish was not run by this GitHub release step. Use the manual Registry workflow or comfy node publish after confirming the Registry token setup.
• fix/node2-legacy-layout-compat PR #19 remains separate and is not included in this release.

EasyUse Anima 0.2.7

Fixed

• Fixed global autocomplete attachment in ComfyUI Nodes 2.0 so prompt text fields that are created or re-rendered after extension startup still receive EasyUse Anima autocomplete behavior.
• Fixed autocomplete search for escaped prompt parentheses so inputs such as \(blue archive\) search the inner literal tag text, matching tags like asuna \(blue archive\).
• Fixed autocomplete popup scroll reset timing after popup hide/show cycles so new suggestion lists reopen at the top instead of preserving a previous scrolled position.

Changed

• Autocomplete popup reset now runs after the popup becomes visible and once more on the next animation frame, avoiding browser scroll restoration while the menu is display: none.
• Disabled scroll anchoring for the autocomplete popup to keep result-list replacement from preserving stale scroll offsets.

Validation

• Added frontend source guards for Nodes 2.0 autocomplete attachment, prompt syntax stripping, popup hide/show scroll reset order, and outside-click state cleanup.
• Added regression coverage for escaped literal parenthesis search.
• Validated with pyproject TOML sanity, full unit tests, compileall, frontend JS syntax checks, and git diff whitespace checks.

Notes

• No ComfyUI restart is required for the frontend JavaScript changes.
• Browser hard refresh may be required to clear cached extension JavaScript.

EasyUse Anima 0.2.6

Added

• Added hover tooltips for trained Prompt Studio highlight tags, using
  autocomplete metadata when it is available.
• Added the prompt_studio.trained_tag_tooltip setting and matching Comfy
  setting key so tag hover tooltips can be enabled or disabled from Prompt
  Studio settings.
• Added a dedicated tabbed Prompt Studio highlight color editor so the growing
  highlight color controls are managed outside the main Prompt Studio settings
  list.
• Added Community Standards documentation for GitHub, including Code of
  Conduct, English/Korean contributing guides, Security Policy, issue forms,
  and a pull request template.

Changed

• Reused autocomplete tooltip formatting for Prompt Studio tag highlights so
  hover text stays consistent between autocomplete suggestions and highlighted
  prompt tokens.
• Registered the tag tooltip and highlight color settings in backend defaults,
  frontend settings UI, and settings key mapping.
• Shared Prompt Studio highlight color handling across the main Prompt Studio
  editor and the common highlighter path.
• Documented that issues, discussions, and pull requests may be written in
  English or Korean.

Validation Notes

• Added regression coverage for Prompt Studio tag tooltip settings.
• Validated the changed Prompt Studio, common highlighter, autocomplete, and
  settings JavaScript with syntax checks during 0.2.6 preparation.
• Validated the GitHub issue template YAML files with PyYAML and checked the
  Community Standards documentation for trailing whitespace.

EasyUse Anima 0.2.5

0.2.5

Added

• Added prompt translation marker support for Prompt Studio and autocomplete flows, backed by the new prompt translation runtime helper and settings.
• Added English ANIMA Easy Use workflow documentation alongside the existing Korean guide.

Changed

• Prompt Studio text areas now use the system UI font and root font size by default instead of forcing the Advanced editor to a fixed 12px monospace font.
• Prompt Studio highlight overlays re-copy input text metrics after settings changes so input text and overlay rendering stay aligned.
• Added manual Prompt Studio font correction settings for display mismatch cases:
  • prompt_studio.font_override
  • prompt_studio.font_family
  • prompt_studio.font_size
• AiO Generator preview handling was refined so Comfy native denoising previews are rendered in the EasyUse Anima preview panel while duplicate native overlay previews are suppressed.

Fixed

• Fixed AiO native preview overlay cleanup so stale overlay state is cleared when preview ownership changes.
• Updated workflow validation coverage for the native preview suppression path.

Validation

• python -m unittest discover -s tests: 244 tests OK
• python -m compileall -q .: passed
• node --check for changed frontend files: passed
• git diff --check: passed
• pyproject TOML metadata sanity check: 0.2.5 passed

EasyUse Anima 0.2.4

Fixed

• Fixed Anima AiO Generator Image Saver integration so Civitai Hash Fetcher API failures, empty results, or temporary upstream exceptions skip only the optional metadata hash instead of failing a completed image save.

Changed

• Includes the post-0.2.3 documentation refresh commits for guides, demo videos, and the ANIMA Easy Use workflow guide.
• Updated bundled example workflow package metadata to 0.2.4.

Validation

• python -m unittest discover -s tests: 235 tests passed, 2 skipped.
• python -m compileall -q .: passed.
• node --check web/js/*.js: passed.
• python -m json.tool locales/ko/nodeDefs.json: passed.

EasyUse Anima 0.2.3

Fixed

• Fixed Anima AiO Generator Detailer Settings so Face/Eye detailer blocks
  render again after opening the popup.
• Fixed Detailer Settings so custom detailer blocks can be added, saved,
  reordered, and executed after the built-in Face/Eye blocks.
• Fixed Highres and Detailer stage optimization so Spectrum/DCW settings are
  saved and applied per stage/block even when CFG, sampler, and scheduler follow
  the main sampler.
• Restored broader autocomplete suggestions when inline autocomplete preview is
  disabled, while keeping strict token-range behavior for preview mode.
• Fixed inline autocomplete preview filtering so Korean description/keyword
  searches such as 하츠네 미쿠 keep the matched tag suggestion.
• Fixed IME composition refresh so Korean input can update autocomplete
  suggestions while composing instead of waiting for a space or refocus.
• Fixed keyboard navigation in the autocomplete menu so the active suggestion
  scrolls into view with at least one adjacent item visible when possible.
• Fixed wildcard autocomplete for empty __ prompts, Korean wildcard keys, and
  normalized space/underscore/path searches.
• Fixed wildcard and dynamic-prompt highlight classification, including Korean
  wildcard syntax and comma-containing dynamic prompts such as
  {1-3$$, $$red|blue}.

Changed

• Updated bundled Korean AiO workflow JSON files with refreshed embedded
  workflow defaults and 0.2.3 package metadata.

Validation Notes

• Added regression coverage for IME autocomplete refresh, arrow-key menu
  scrolling, wildcard autocomplete token detection, Korean wildcard keys, and
  wildcard/dynamic-prompt highlight classification.
• Added regression coverage for custom Detailer blocks and stage-specific
  Highres/Detailer Spectrum/DCW settings.

v0.2.2

0.2.2

Highlights

• Added the first release-ready Anima AiO Generator path: Easy Use Anima Input loads ANIMA diffusion model, VAE, and CLIP separately, then passes a
  dedicated context socket into the generator.
• The generator keeps the compact node UI focused on seed, steps, CFG, denoise,
  sampler/scheduler, Highres, Detailer, preview, and save controls. Sampler
  backend controls, model patch/optimization controls, save metadata, and
  detailer controls stay in popup settings.
• AiO generation settings are stored as keyed versioned JSON so changing
  parameter order does not shift saved workflow data.

Added

• Added Anima AiO Generator with three sampler execution modes:
  comfy_ksampler, spectrum_mod_guidance_advanced, and
  spectrum_spd_speed.
• Added Anima Prompt Corrector Simple, a compact one-input and one-output
  prompt correction node for regular multiline string workflows.
• Added optional Anima DAVE model patch controls in Advanced Options.
• Added Easy Use Anima Input, which consumes prompt data and stores split
  ANIMA resource selections for diffusion model, VAE, and CLIP.
• Added optional LORA_STACK support on the AiO generator, including Image
  Saver-compatible Civitai LoRA resource/weight metadata.
• Added Image Saver integration with default saving enabled, workflow embedding,
  manual additional_hashes, and Civitai Hash Fetcher bundle rows.
• Added AiO preview UI with WebP temp previews, current/previous comparison,
  image feed history, selected image metadata, and feed-count settings.
• Added Highres and SAM3/Impact Detailer settings to the AiO generator.
• Added Chrome-style Detailer tabs with user-editable block names and left/right
  order controls.
• Added popup-based Prompt Studio Advanced controls for Artist Mix,
  Mod Guidance, wildcard seed, and resolution bucket settings so the node body
  stays compact.
• Added Artist Mix group syntax such as [[artist_a, artist_b:0.7]] for
  keeping multiple artists in one conditioning branch. The final :0.7 is an
  Artist Mix conditioning weight, not a prompt-string weight.
• Added inline autocomplete ghost preview and optional editor-style closing
  bracket insertion settings.
• Added optional underline rendering for weighted prompt syntax such as
  (tag:1.2) and Artist Mix groups.
• Added a maintained AiO generator sample workflow:
  docs/example_workflows/EasyUse_Anima_AiO_generator_release_ko.json.
• Added the compact release workflow:
  docs/example_workflows/ANIMA_Easy_Use_workflow_v1_release_ko.json.
• Added required_node_packs metadata to release workflow templates so
  required custom nodes are visible from the workflow JSON itself.
• Added user-facing AiO node documentation in Korean and English.
• Added an ANIMA Easy Use workflow v1 usage draft and Simple prompt corrector
  node screenshot.

Changed

• Bumped package metadata to 0.2.2.
• Updated AiO defaults: first-pass steps 32, sampler er_sde, scheduler
  simple, AuraFlow shift 3.0 as the Anima model-recommended default,
  Highres scale 1.5, and Highres denoise 0.25.
• Moved KJNodes FP16 accumulation, SageAttention, Torch Compile, AuraFlow shift,
  and optional DAVE controls into Advanced Options. Sampler Details now
  keeps only sampler backend, Mod Guidance, Spectrum, and SPD/SPEED controls.
• Changed Highres and Detailer popup bodies to a single-column layout so long
  settings stay readable and scroll vertically instead of forming cramped grids.
• Changed Prompt Studio Advanced wildcard seed and resolution controls to show a
  single popup button plus a compact current-setting summary in the node body.
• Changed Prompt Studio autocomplete, correction, and highlight handling to
  share the same prompt text rules for spaces, escaped parentheses,
  Pony score tags, artist tags, and weighted syntax.
• Replaced the old KR_danbooru_tags_with_description v3_modified.csv source
  with the maintained danbooru_tags_classified.csv autocomplete/highlight
  source.
• Updated AiO frontend tooltips so popup settings describe the actual runtime
  effect instead of only saying that a value is saved. The new tooltip keys are
  available in English, Korean, Japanese, and Chinese.
• Split AiO dependencies into required and optional feature packs. Missing
  optional packs are locked in the settings UI and sanitized out of queued
  generation_settings before execution.

Fixed

• Fixed AiO sampler dispatch so the selected sampler mode is the only first-pass
  sampler path called.
• Fixed the spectrum_mod_guidance_advanced mode so it no longer creates an
  unused standalone Mod Guidance model clone for the first pass. Standalone Mod
  Guidance model patching is created once and reused by KSampler-based stages
  instead of being stacked repeatedly.
• Changed AiO Highres sampling to either reuse the first-pass sampler path or
  use the general KSampler path. SPD/SPEED first passes now use general KSampler
  for Highres, and Highres keeps its own Steps and Denoise.
• Hardened Spectrum sampler calls against node-pack API drift by filtering
  SpectrumKSamplerAdvanced and SpectrumSPDKSampler keyword arguments against
  the installed sample() signature. Sampler Details also reads /object_info
  to show detected extra inputs and node-pack tooltips when available.
• Fixed general KSampler + Spectrum Patch routing so installed node packs that
  expose DiTSpectrumPatch instead of DiTSpectrumPatchAdvanced are supported.
• Fixed AiO SPD/SPEED preview and Highres handoff sizing so the node output,
  preview feed, and Highres target resolution stay aligned with the requested
  workflow resolution.
• Fixed AiO preview result handling so ComfyUI's default images UI payload is
  suppressed and only the dedicated easyuse_anima_preview payload is used.
• Hardened intermediate preview feed updates so live preview events are tagged
  by run and displayed in the node feed immediately.
• Fixed Image Saver metadata routing so Steps, CFG, Sampler,
  Scheduler, Seed, and Denoise come from the first-pass sampler while
  Size uses the final Highres/Detailer output resolution.
• Fixed Prompt Studio highlighting and autocomplete replacement around weighted
  prompt syntax so applying a suggestion does not delete surrounding
  parentheses or weights.
• Fixed unweighted parenthesized tags such as (@artist name) or
  (highres, long hair) so their inner tags are classified and highlighted by
  category.
• Fixed autocomplete popup activation so syntax-only caret positions, including
  plain brackets, do not open irrelevant suggestions.
• Fixed middle-click behavior on Prompt Studio inputs so it is forwarded to
  canvas panning instead of selecting a text caret.

Required And Optional Node Packs

• Required for this package: ComfyUI-EasyUseAnima.
• Required by the included AiO sample defaults: ComfyUI-Spectrum-KSampler
  and ComfyUI-Image-Saver.
• Optional: ComfyUI-Anima-DAVE for the DAVE model patch in Advanced Options.
• Optional: ComfyUI-KJNodes for SageAttention and Torch Compile options.
• Optional: ComfyUI-Impact-Pack for AiO SAM3 Detailer.

Validation Notes

• Added regression coverage for all three AiO sampler modes so
  comfy_ksampler, spectrum_mod_guidance_advanced,
  and spectrum_spd_speed dispatch to their intended paths only.
• Added regression coverage for generator-level sampler/model-patch routing,
  including the integrated Spectrum Mod Guidance sampler and Highres stage
  model reuse.
• Added regression coverage for Artist Mix grouped branches, group-weight
  flattening, invalid weight syntax, weighted tag groups, and unweighted
  parenthesized tag classification across meta, character, general, and artist
  tags.
• Ran python -m unittest discover -s tests, python -m compileall -q .,
  frontend node --check, workflow/locale tests, Markdown link checks, and
  Registry metadata sanity checks during release validation.

v0.2.1

Highlights

• Prepare Anima Prompt Studio Advanced v2 with structured prompt data output.
• Downstream nodes should read Prompt Studio data by dict keys instead of
  positional output indexes so future output changes remain compatible.
• Add artist-field handling modes so text from Prompt Studio's artist-tag input
  fields can stay inline or be routed as separate structured data for
  artist-conditioning nodes, following the ComfyUI-AnimaPromptEditor
  artist-mix payload as the default compatibility target.
• Keep the existing Anima Prompt Studio Advanced node compatible while v2
  output contracts are introduced.

Added

• Added Anima Prompt Studio Advanced v2 with a single output socket of type
  EASYUSE_ANIMA_PROMPT_DATA.
• The v2 prompt data output is a Python dict with keyed compatibility outputs,
  node parameter state, resolution data, wildcard state, Mod Guidance flags,
  and artist-field data.
• Added EASYUSE_ANIMA_PROMPT_DATA helper node to pass prompt data through,
  optionally override compatibility fields, and unfold it into the existing
  Prompt Studio compatibility outputs.
• Added Anima Prompt Data Conditioning to consume EASYUSE_ANIMA_PROMPT_DATA,
  encode positive/negative CONDITIONING, create a batch-size-1 latent image
  from prompt-data width/height, and apply the comfyui-spectrum-ksampler
AnimaModGuidance MODEL patch when enabled.
• Anima Prompt Data Conditioning now supports artist mix modes from
  prompt-data or node controls. off/prompt preserve the inline positive
  prompt, while average and exact separate Advanced artist-field text from
  the base prompt and rebuild artist variants through EasyUseAnima's Anima
  prompt ordering rules.
• Added approximate artist mix modes for cheaper exact-like mixing:
  delta_rms, hybrid, and clustered. These reduce positive conditioning
  branch count by compressing artist deltas while preserving the existing
  average and exact behavior.
• Artist mix controls now expose tuning values for style gain, RMS scale cap,
  hybrid top-K, clustered branch count, and clustered dominant-artist
  isolation.
• Artist mix mode tooltips now describe the method and expected conditioning
  branch cost.
• Added Anima Artist Mix Conditioning as a standalone artist mix node. It
  accepts a regular prompt plus separate artist_tags, supports
  correct/front/back artist positioning, and outputs positive
  CONDITIONING without requiring Prompt Data.
• Added user-facing artist mix documentation covering Advanced v2 Prompt Data
  routing, standalone artist tag positioning, artist mix modes, and branch cost
  tradeoffs.
• Added an Artist Mix example workflow and preview under
  docs/example_workflows/.
• Prompt data now stores global_prompt /
  positive_without_artist_section, structured artist.tags, and
  artist_mix routing values for artist-conditioning nodes.
• Prompt data now stores a parameters dict generated from every required
  Anima Prompt Studio Advanced v2 input, so new v2 parameters are caught by
  tests instead of being silently omitted from EASYUSE_ANIMA_PROMPT_DATA.
• Anima Prompt Studio Advanced v2 now exposes inline foldout controls for
  Mod Guidance and Artist Mix, with Artist Mix mode written into prompt data.
• Prompt-data socket names are fixed to English identifiers across locales so
  displayed socket names match the Python node contract.
• Added NAIA resolution bucket fit mode for Anima Prompt Studio Advanced.
  NAIA width/height can now resolve to the nearest aspect ratio inside a
  configured saved resolution bucket.
• Added naia.resolution_mode and naia.resolution_bucket settings plus
  localized frontend controls for original-scale and bucket-fit modes.

Changed

• Bumped package metadata and maintained example workflow package_version
  metadata to 0.2.1.

Fixed

• Fixed Regional Prompt Studio dynamic field sockets so field input sockets stay
  synchronized with node fields and connection changes.
• Connected Regional Prompt Studio STRING inputs can now override masked
  prompt text at queue time without overwriting the saved field text.
• Foldout controls in Prompt Studio keep their open state across internal
  re-renders so clicks inside the control do not immediately collapse them.
• Advanced Prompt Studio native controls such as resolution bucket, wildcard,
  and Artist Mix dropdowns now consistently stop canvas event propagation so
  opened dropdowns stay open while selecting values.
• Anima Prompt Data Conditioning now calls Spectrum
  AnimaModGuidance.patch() with the installed node's supported signature, so
  current Spectrum builds no longer raise an argument-count TypeError.
• Older Spectrum AnimaModGuidance builds now emit a warning when their
  signature cannot receive separate negative quality tags. Execution continues,
  but negative Mod Guidance quality tags are ignored by that model patch.

Validation Notes

• Added regression coverage for the Advanced v2 prompt data socket, dict
  payload, EASYUSE_ANIMA_PROMPT_DATA helper outputs and overrides, and
  artist-field-only artist data extraction.
• Added regression coverage that every required Advanced v2 input is present
  in prompt data parameters.
• Added regression coverage for prompt-data conditioning and Spectrum
  AnimaModGuidance patcher invocation.
• Added regression coverage for current Spectrum AnimaModGuidance invocation
  without a separate negative quality-tag argument and for future-compatible
  invocation when that argument is supported.
• Added regression coverage for the old Spectrum warning path.
• Added regression coverage for prompt-data latent image creation with fixed
  batch size 1.
• Added regression coverage for artist-field-only prompt data, artist-free
  base prompts, artist mix prompt rebuilding, and exact-mode conditioning
  metadata.
• Added regression coverage for duplicate artist coalescing, hybrid top-K tail
  compression, approximate artist mix mode routing, and prompt-data storage of
  artist mix tuning values.
• Added regression coverage for the standalone artist mix conditioning node,
  including default corrected placement and fixed front/back placement.
• Added regression coverage for Regional Prompt Studio field socket overrides.
• Added regression coverage for NAIA resolution mode/bucket validation and
  bucket-fit output sizing.