Regional Composition — ComfyUI workflow

16Graph nodes

10Node classes

16Inputs

2Models

Tags: compositionregionalarea conditioningmulti-promptplacement

Inputs (16)#

The typed parameter surface callers bind when they request this workflow. Enum options and numeric bounds are the values the workflow document declares.

base_promptstringrequired

Overall scene prompt applied to the whole canvas

region_1_promptstringrequired

What appears inside region 1

region_1_areaenumdefault left_half

Where region 1 sits on the canvas

options: left_halfright_halftop_halfbottom_halfcentertop_lefttop_rightbottom_leftbottom_right

region_1_strengthfloatdefault 1.1min 0.1max 3.0

Conditioning strength of region 1 relative to the base prompt

region_2_enabledbooleandefault false

Add a second prompted region

region_2_promptstringdefault

What appears inside region 2

region_2_areaenumdefault right_half

Where region 2 sits on the canvas

options: left_halfright_halftop_halfbottom_halfcentertop_lefttop_rightbottom_leftbottom_right

region_2_strengthfloatdefault 1.1min 0.1max 3.0

Conditioning strength of region 2 relative to the base prompt

seedintegerdefault -1

Random seed (-1 for random)

widthintegerdefault 1024min 768max 1536

Output width

heightintegerdefault 1024min 768max 1536

Output height

stepsintegerdefault 34min 15max 60

Sampling steps

cfg_scalefloatdefault 7.5min 1.0max 15.0

Prompt adherence strength

hires_fixbooleandefault true

Two-pass hires refinement (1.5x latent upscale + low-denoise second pass)

freeubooleandefault true

FreeU v2 feature re-weighting

checkpointenumdefault sdxl_base

SDXL checkpoint

options: realvis_xl_v5sdxl_base

ComfyUI node graph (16)#

The executable ComfyUI prompt graph: 16 nodes across 10 distinct node classes, wired by 23 data dependencies. Nodes tinted green come from a custom node pack this workflow declares; the rest are ComfyUI core / baked-community classes.

flowchart TD n0["1: CheckpointLoaderSimple"] n1["2: CLIPTextEncode"] n2["3: CLIPTextEncode"] n3["4: CLIPTextEncode"] n4["5: ConditioningSetAreaPercentage"] n5["6: ConditioningCombine"] n6["7: EmptyLatentImage"] n7["8: KSampler"] n8["9: VAEDecode"] n9["10: SaveImage"] n10["r2_encode: CLIPTextEncode"] n11["r2_area: ConditioningSetAreaPercentage"] n12["r2_combine: ConditioningCombine"] n13["freeu: FreeU_V2"] n14["hires_upscale: LatentUpscaleBy"] n15["hires_sampler: KSampler"] n0 -->|clip| n1 n0 -->|clip| n2 n0 -->|clip| n3 n3 -->|conditioning| n4 n1 -->|conditioning_1| n5 n4 -->|conditioning_2| n5 n13 -->|model| n7 n12 -->|positive| n7 n2 -->|negative| n7 n6 -->|latent_image| n7 n15 -->|samples| n8 n0 -->|vae| n8 n8 -->|images| n9 n0 -->|clip| n10 n10 -->|conditioning| n11 n5 -->|conditioning_1| n12 n11 -->|conditioning_2| n12 n0 -->|model| n13 n7 -->|samples| n14 n13 -->|model| n15 n12 -->|positive| n15 n2 -->|negative| n15 n14 -->|latent_image| n15

Data-flow DAG — scroll to zoom, drag to pan.

Nodes (16)#

1CheckpointLoaderSimplecore

ckpt_name = {{checkpoint_files[checkpoint]}} tmpl

outputs: MODELCLIPVAE

2CLIPTextEncodecore

text = {{base_prompt}} tmplclip = ◂ node 1 · out[1]

outputs: CONDITIONING

3CLIPTextEncodecore

text = ugly, deformed, blurry, low quality, watermark, text, signature, duplicated subject, merged figuresclip = ◂ node 1 · out[1]

outputs: CONDITIONING

4CLIPTextEncodecore

text = {{region_1_prompt}} tmplclip = ◂ node 1 · out[1]

outputs: CONDITIONING

5ConditioningSetAreaPercentagecore

conditioning = ◂ node 4 · out[0]width = {{area_configs[region_1_area].width}} tmplheight = {{area_configs[region_1_area].height}} tmplx = {{area_configs[region_1_area].x}} tmply = {{area_configs[region_1_area].y}} tmplstrength = {{region_1_strength}} tmpl

outputs: CONDITIONING

6ConditioningCombinecore

conditioning_1 = ◂ node 2 · out[0]conditioning_2 = ◂ node 5 · out[0]

outputs: CONDITIONING

7EmptyLatentImagecore

width = {{width}} tmplheight = {{height}} tmplbatch_size = 1

outputs: LATENT

8KSamplercore

model = ◂ node freeu · out[0]positive = ◂ node r2_combine · out[0]negative = ◂ node 3 · out[0]latent_image = ◂ node 7 · out[0]seed = {{seed}} tmplsteps = {{steps}} tmplcfg = {{cfg_scale}} tmplsampler_name = dpmpp_2m_sdescheduler = karrasdenoise = 1.0

outputs: LATENT

9VAEDecodecore

samples = ◂ node hires_sampler · out[0]vae = ◂ node 1 · out[2]

outputs: IMAGE

10SaveImagecore

images = ◂ node 9 · out[0]filename_prefix = regional_composition

r2_encodeCLIPTextEncodecore

text = {{region_2_prompt}} tmplclip = ◂ node 1 · out[1]

outputs: CONDITIONING

r2_areaConditioningSetAreaPercentagecore

conditioning = ◂ node r2_encode · out[0]width = {{area_configs[region_2_area].width}} tmplheight = {{area_configs[region_2_area].height}} tmplx = {{area_configs[region_2_area].x}} tmply = {{area_configs[region_2_area].y}} tmplstrength = {{region_2_strength}} tmpl

outputs: CONDITIONING

r2_combineConditioningCombinecore

conditioning_1 = ◂ node 6 · out[0]conditioning_2 = ◂ node r2_area · out[0]

outputs: CONDITIONING

freeuFreeU_V2core

model = ◂ node 1 · out[0]b1 = 1.1b2 = 1.2s1 = 0.6s2 = 0.4

outputs: MODEL

hires_upscaleLatentUpscaleBycore

samples = ◂ node 8 · out[0]upscale_method = bislerpscale_by = 1.5

outputs: LATENT

hires_samplerKSamplercore

model = ◂ node freeu · out[0]positive = ◂ node r2_combine · out[0]negative = ◂ node 3 · out[0]latent_image = ◂ node hires_upscale · out[0]seed = {{seed}} tmplsteps = 14cfg = {{cfg_scale}} tmplsampler_name = dpmpp_2mscheduler = karrasdenoise = 0.45

outputs: LATENT

Parameter banks (1)#

The prompt / configuration lookup tables this workflow keys into from its inputs — the vocabulary that turns a style / palette / preset selection into graph parameters.

`area_configs` (9)#

left_half

{"x": 0.0, "y": 0.0, "width": 0.5, "height": 1.0}

right_half

{"x": 0.5, "y": 0.0, "width": 0.5, "height": 1.0}

top_half

{"x": 0.0, "y": 0.0, "width": 1.0, "height": 0.5}

bottom_half

{"x": 0.0, "y": 0.5, "width": 1.0, "height": 0.5}

center

{"x": 0.25, "y": 0.25, "width": 0.5, "height": 0.5}

top_left

{"x": 0.0, "y": 0.0, "width": 0.5, "height": 0.5}

top_right

{"x": 0.5, "y": 0.0, "width": 0.5, "height": 0.5}

bottom_left

{"x": 0.0, "y": 0.5, "width": 0.5, "height": 0.5}

bottom_right

{"x": 0.5, "y": 0.5, "width": 0.5, "height": 0.5}

Models & dependencies#

Checkpoint aliases (2)#

realvis_xl_v5

RealVisXL_V5.0_fp16.safetensors

sdxl_base

sd_xl_base_1.0.safetensors

Models required (2)#

RealVisXL_V5.0_fp16.safetensors

sd_xl_base_1.0.safetensors

Output contract#

What a successful run of this workflow returns.

type

image

format

png

channels

4

Taxonomy & routing#

How the control plane classifies this workflow — from the committed workflow-taxonomy-registry.json. It drives the consistency / control surface the agentic director can exercise over the workflow.

assetFamily

multi-region-composed-still

outputPackageProfile

image-single-profile

controlModalities

prompt-template-lockmodel-locksampler-scheduler-lockseed-lock

consistencyDimensions

environment

notes

Core percentage-area conditioning: base scene plus up to two placed regions with independent prompts and strengths.

Inputs (16)#

ComfyUI node graph (16)#

Nodes (16)#

Parameter banks (1)#

area_configs (9)#

Models & dependencies#

Checkpoint aliases (2)#

Models required (2)#

Output contract#

Taxonomy & routing#

`area_configs` (9)#