Character Sheet Pack

Separate mannequin, wardrobe, character-sheet and dataset tasks. Packs at most three references and creates an explicit image-role map. Raw body references are used only in the mannequin stage. Connect positive/negative and all three images to the same Qwen encoder. Defaults: prompts/character_sheets.json.

By r-vage·Created 11 months ago·Updated about 15 hours ago· 38

layout
face
wardrobe
front_body
rear_body
front
rear

positive
negative
image_1
image_2
image_3
report

◄stagesheet►

◄generation_modePreset►

◄genderunspecified►

◄buildaverage►

◄muscularityunspecified►

◄breast_sizeunspecified►

◄chest_breadthunspecified►

◄hip_widthunspecified►

◄buttock_sizeunspecified►

◄width1536►

◄height864►

◄details►

◄use_front_bodytrue►

◄use_rear_bodytrue►

◄use_wardrobetrue►

◄use_reartrue►

◄positive_override—►

◄negative_override—►

Category🌒 Eclipse/ Image

Inputs (25)

Name	Type	Default	Description
stage	COMBO	sheet	mannequin: choose Preset controls or Reference images; wardrobe: outfit source; sheet: approved mannequin + face + optional wardrobe; dataset: individual front + portrait + optional rear panels. Changing stage also requires updating connected prompt overrides.
generation_mode	COMBO	Preset	Mannequin only. Preset skips every image input and uses a blank canvas plus selected attributes. Reference uses optional layout/front/rear images and ignores hidden body presets.
gender	COMBO	unspecified	Mannequin: Gender applies in both modes; body attributes apply only in Preset. All attributes are independent of gender. Unspecified adds no instruction. Explicit details refine only named attributes. Selection names and descriptions come from mannequin_attributes in prompts/character_sheets.json. After adding entries, restart ComfyUI and refresh the browser.
build	COMBO	average	Mannequin: Gender applies in both modes; body attributes apply only in Preset. All attributes are independent of gender. Unspecified adds no instruction. Explicit details refine only named attributes. Selection names and descriptions come from mannequin_attributes in prompts/character_sheets.json. After adding entries, restart ComfyUI and refresh the browser.
muscularity	COMBO	unspecified	Mannequin: Gender applies in both modes; body attributes apply only in Preset. All attributes are independent of gender. Unspecified adds no instruction. Explicit details refine only named attributes. Selection names and descriptions come from mannequin_attributes in prompts/character_sheets.json. After adding entries, restart ComfyUI and refresh the browser.
breast_size	COMBO	unspecified	Mannequin: Gender applies in both modes; body attributes apply only in Preset. All attributes are independent of gender. Unspecified adds no instruction. Explicit details refine only named attributes. Selection names and descriptions come from mannequin_attributes in prompts/character_sheets.json. After adding entries, restart ComfyUI and refresh the browser.
chest_breadth	COMBO	unspecified	Mannequin: Gender applies in both modes; body attributes apply only in Preset. All attributes are independent of gender. Unspecified adds no instruction. Explicit details refine only named attributes. Selection names and descriptions come from mannequin_attributes in prompts/character_sheets.json. After adding entries, restart ComfyUI and refresh the browser.
hip_width	COMBO	unspecified	Mannequin: Gender applies in both modes; body attributes apply only in Preset. All attributes are independent of gender. Unspecified adds no instruction. Explicit details refine only named attributes. Selection names and descriptions come from mannequin_attributes in prompts/character_sheets.json. After adding entries, restart ComfyUI and refresh the browser.
buttock_size	COMBO	unspecified	Mannequin: Gender applies in both modes; body attributes apply only in Preset. All attributes are independent of gender. Unspecified adds no instruction. Explicit details refine only named attributes. Selection names and descriptions come from mannequin_attributes in prompts/character_sheets.json. After adding entries, restart ComfyUI and refresh the browser.
width	INT	1536256–2048	Output/reference canvas width. Use a landscape canvas for sheets and a portrait canvas for individual dataset shots. Qwen encoder resolution must be 0.
height	INT	864256–2048	Output canvas height. Mannequin/sheet stages: only the first image sets the canvas; supporting body, face and wardrobe references keep their own aspect ratios, are never enlarged, and are capped at 1024 pixels on the longest side. Other stages pad references without stretching. Without a mannequin template, width/height set the blank generation canvas. When using a template, match its aspect ratio for accurate panel splits.
details	STRING		Additional task specification. Mannequin: physique; wardrobe: missing outfit choices; sheet: identity and outfit; dataset: identity/wardrobe continuity only, without fixed pose instructions. Required when sheet has no wardrobe image.
use_front_body	BOOLEAN	true	Mannequin Reference mode only: use the connected frontal body image. Off skips this input branch. Has no effect on other stages.
use_rear_body	BOOLEAN	true	Mannequin Reference mode only: use the connected rear body image. Off skips this input branch. No effect on the final sheet's rear panel, which is always generated.
use_wardrobe	BOOLEAN	true	Sheet stage only: use the wardrobe board. Off uses the written outfit specification and skips this input branch unless another output separately requests it.
use_rear	BOOLEAN	true	Dataset stage only: include the approved rear crop alongside front and face. Off can reduce rear-view influence while retaining front/portrait references. It does not change planner camera choices.
layoutopt	IMAGE		Mannequin Reference mode: leave disconnected to generate from the written three-panel layout; the node supplies a blank size-setting canvas. Optionally connect a pose template for framing only. Sheet stage: the approved generated mannequin guide is required and supplies poses and proportions.
faceopt	IMAGE		Required for sheet/dataset: one identity portrait. Dataset normally uses the extracted portrait panel. Remains a separate image even when a rear reference is enabled.
wardrobeopt	IMAGE		Wardrobe stage: source outfit photograph. Sheet stage: approved wardrobe board. Different board views describe one outfit. Ignored by mannequin/dataset stages.
front_bodyopt	IMAGE		Optional original frontal physique reference, used ONLY in mannequin Reference mode; Preset skips it. Never routed into dressed-character or dataset generation. No extraction or undressing required.
rear_bodyopt	IMAGE		MANNEQUIN Reference mode only (Preset skips it): original rear body photograph for visible physique/proportions. Its clothing and identity are ignored. Different from rear, which accepts the finished dressed character's rear panel for the dataset stage. Ignored by wardrobe/sheet/dataset stages.
frontopt	IMAGE		Dataset: approved front panel, one image. Supplies established outfit and proportions. A contact sheet or image batch is not appropriate here.
rearopt	IMAGE		DATASET stage only: approved rear panel cropped from the finished dressed character sheet, preserving the chosen outfit and rear appearance. Different from rear_body, the original physique photograph. Ignored by mannequin/wardrobe/sheet stages. Does not request every shot to face away.
positive_overrideopt	STRING		Replace the stage's file task text with an editable String Multiline input. The Pack adds the reference-role map, fixed mannequin layout, Gender, active Preset selections and details, then resolves uppercase role names such as FRONT BODY, FACE and WARDROBE to the actual <image1>, <image2>, etc. throughout the final task text. This keeps optional-input numbering correct. Connect positive to Show Any to inspect the resolved prompt. Disconnect this input to load current file text.
negative_overrideopt	STRING		Replace this stage's file negative independently. An empty connected string intentionally clears it. File edits do not change connected workflow text.

Outputs (6)

Name	Type	Description
positive	STRING	Stage prompt plus actual reference mapping. In dataset stage, connect to the shot planner's character_lock; other stages connect directly to their encoder.
negative	STRING	Negative for this stage's encoder. Sheet negatives allow exactly three panels; dataset negatives prohibit panels. At sampler CFG=1 the negative has no effect; higher CFG enables it.
image_1	IMAGE	Connect to the same encoder's image_1. Roles depend on stage and enabled inputs; check report. Unused slots return no image.
image_2	IMAGE	Connect to the same encoder's image_2. Roles depend on stage and enabled inputs; check report. Unused slots return no image.
image_3	IMAGE	Connect to the same encoder's image_3. Roles depend on stage and enabled inputs; check report. Unused slots return no image.
report	STRING	Lists actual image labels, role mapping, omitted references, target canvas and individual reference dimensions. No images are generated by this node.