Stable Zero123 Multi-View Generator — ComfyUI workflow

10Graph nodes

9Node classes

10Inputs

1Models

Tags: 3dmultiviewzero123reconstructionnovel-view

Inputs (10)#

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

reference_imageimagerequired

Reference image to generate views from

view_configurationenumdefault orbital_8

Configuration of output views

options: cardinal_4orbital_8hemisphere_12full_sphere_24custom

custom_view_countintegerdefault 12min 2max 36

Number of views to generate when view_configuration is 'custom' (evenly spaced around the azimuth at elevation 0)

elevation_rangeenumdefault slight_variation

Range of vertical viewing angles

options: level_onlyslight_variationfull_hemispherefull_sphere

object_typeenumdefault general

Type of object for optimized processing

options: generalcharacterstatueartifactsacred_objectcrystalnatural_objectarchitectural

consistency_levelenumdefault high

Level of view consistency

options: standardhighmaximum

detail_preservationenumdefault high_detail

Detail preservation in generated views

options: balancedhigh_detailsmooth

background_handlingenumdefault remove

How to handle background in outputs

options: preserveremoveneutral_graytransparent

output_resolutionenumdefault 512x512

Resolution of output views

options: 256x256512x512768x768

seedintegerdefault -1

Random seed for reproducibility

ComfyUI node graph (10)#

The executable ComfyUI prompt graph: 10 nodes across 9 distinct node classes, wired by 14 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: LoadImage"] n1["2: ImageScale"] n2["3: BackgroundRemover"] n3["4: ImageOnlyCheckpointLoader"] n4["5: StableZero123_Conditioning_Batched"] n5["6: KSampler"] n6["7: VAEDecode"] n7["8: SaveImage"] n8["9: CreateViewGrid"] n9["10: SaveImage"] n0 -->|image| n1 n1 -->|image| n2 n3 -->|clip_vision| n4 n2 -->|init_image| n4 n3 -->|vae| n4 n3 -->|model| n5 n4 -->|positive| n5 n4 -->|negative| n5 n4 -->|latent_image| n5 n5 -->|samples| n6 n3 -->|vae| n6 n6 -->|images| n7 n6 -->|images| n8 n8 -->|images| n9 classDef custom fill:#12b886,stroke:#0b6b4f,color:#04110c; class n2,n8 custom;

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

Nodes (10)#

1LoadImagecore

image = {{reference_image}} tmpl

outputs: IMAGEMASK

2ImageScalecore

image = ◂ node 1 · out[0]upscale_method = lanczoswidth = 256height = 256crop = center

outputs: IMAGE

3BackgroundRemovercustom pack

image = ◂ node 2 · out[0]mode = {{background_handling}} tmpl

outputs: IMAGEMASK

4ImageOnlyCheckpointLoadercore

ckpt_name = stable_zero123.ckpt

outputs: MODELCLIP_VISIONVAE

5StableZero123_Conditioning_Batchedcore

clip_vision = ◂ node 4 · out[1]init_image = ◂ node 3 · out[0]vae = ◂ node 4 · out[2]width = 256height = 256batch_size = {{custom_view_count if view_configuration == 'custom' else view_configurations[view_configuration].count}} tmplelevation = {{0.0 if view_configuration == 'custom' else view_configurations[view_configuration].elevation}} tmplazimuth = 0.0elevation_batch_increment = 0.0azimuth_batch_increment =

{{round(360.0 / custom_view_count, 2) if view_configuration == 'custom' else view_configurations[view_configuration].azimuth_increment}}

tmpl

outputs: positivenegativelatent

6KSamplercore

model = ◂ node 4 · out[0]positive = ◂ node 5 · out[0]negative = ◂ node 5 · out[1]latent_image = ◂ node 5 · out[2]seed = {{seed}} tmplsteps = 20cfg = 4.0sampler_name = eulerscheduler = sgm_uniformdenoise = 1.0

outputs: LATENT

7VAEDecodecore

samples = ◂ node 6 · out[0]vae = ◂ node 4 · out[2]

outputs: IMAGE

8SaveImagecore

images = ◂ node 7 · out[0]filename_prefix = multiview

9CreateViewGridcustom pack

images = ◂ node 7 · out[0]columns = 4caption = {{view_configuration}} multiview tmpl

outputs: IMAGE

10SaveImagecore

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

Prompt construction#

template

Multi-view consistent 3D object rendering, {object_prompt}, same object from different angles, consistent lighting, consistent materials, {detail} detail, professional 3D asset views

Variables (2)#

object_prompt

{{object_type_prompts[object_type]}} tmpl

detail

{{detail_preservation}} tmpl

Parameter banks (4)#

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

`view_configurations` (4)#

cardinal_4

{"views": 4, "azimuths": [0, 90, 180, 270], "elevations": [0, 0, 0, 0], "description": "Front, right, back, left views", "count": 4, "azimuth_increment": 90.0, "elevation": 0.0}

orbital_8

{"views": 8, "azimuths": [0, 45, 90, 135, 180, 225, 270, 315], "elevations": [0, 0, 0, 0, 0, 0, 0, 0], "description": "8 views around the object", "count": 8, "azimuth_increment": 45.0, "elevation": 0.0}

hemisphere_12

{"views": 12, "azimuths": [0, 60, 120, 180, 240, 300, 0, 60, 120, 180, 240, 300], "elevations": [0, 0, 0, 0, 0, 0, 30, 30, 30, 30, 30, 30], "description": "12 views covering upper hemisphere", "count": 12, "azimuth_increment": 30.0, "elevation": 0.0}

full_sphere_24

{"views": 24, "description": "24 views covering full sphere", "count": 24, "azimuth_increment": 15.0, "elevation": 0.0}

`object_type_prompts` (8)#

general

3D object, consistent appearance from all angles

character

3D character, figure, consistent anatomy and clothing from all views

statue

3D statue, sculpture, consistent material and form from all angles

artifact

3D artifact, relic, consistent details and textures from all views

sacred_object

3D sacred object, spiritual item, consistent symbolic details

crystal

3D crystal, gem, consistent facets and reflections from all angles

natural_object

3D natural object, organic form, consistent texture from all views

architectural

3D architectural element, building detail, consistent structure

`resolution_map` (3)#

256x256

{"width": 256, "height": 256}

512x512

{"width": 512, "height": 512}

768x768

{"width": 768, "height": 768}

`usage_notes` (2)#

best_results

Use clean, well-lit reference imagesCenter object in frame with some paddingRemove or use neutral background for best consistencyHigher consistency may reduce view variety

3d_reconstruction

Output views can be used with NeRF, 3D Gaussian Splatting, or photogrammetry tools

Models & dependencies#

Custom node packs (2)#

The non-core ComfyUI node classes this graph requires; the RunPod worker image the workflow runs on must bake or install a pack that provides every one of them.

BackgroundRemoverCreateViewGrid

Models required (1)#

stable_zero123.ckpt

Output contract#

What a successful run of this workflow returns.

primary

{"type": "image_sequence", "format": "png", "description": "Individual view images with angle metadata"}

secondary

{"type": "image", "format": "png", "description": "Grid overview of all views"}

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

game-character-game-ready-bundle

outputPackageProfile

game-ready-3d-profile

controlModalities

camera-lockcontrolnet-depthcontrolnet-normaldepth-constraintidentity-adapterlatent-reuselens-profile-lockmodel-lockpalette-lockpose-constraintreference-ensemblesampler-scheduler-lockscene-lockseed-lockstyle-anchortemporal-lock +1 more

consistencyDimensions

propenvironmentlightinglens

notes

Multi-view generation aligned to game-ready character bundle.

Inputs (10)#

ComfyUI node graph (10)#

Nodes (10)#

Prompt construction#

Variables (2)#

Parameter banks (4)#

view_configurations (4)#

object_type_prompts (8)#

resolution_map (3)#

usage_notes (2)#

Models & dependencies#

Custom node packs (2)#

Models required (1)#

Output contract#

Taxonomy & routing#

`view_configurations` (4)#

`object_type_prompts` (8)#

`resolution_map` (3)#

`usage_notes` (2)#