Save 4 hours! We auto-setup your workflow! Free!

Drop your workflow.json — we handle every dependency, custom node, and model. Just open the link and run.

Auto-Setup Workflow Json (Free) Now!
ComfyUI > Nodes > OpenKatana > OpenKatana · Run model workflow

ComfyUI Node: OpenKatana · Run model workflow

Class Name

OpenKatana_run_model_workflow

Category
OpenKatana/Tools
Author
rk0neki (Account age: 235days)
Extension
OpenKatana
Latest Updated
2026-11-05
Github Stars
0.05K

How to Install OpenKatana

Install this extension via the ComfyUI Manager by searching for OpenKatana
  • 1. Click the Manager button in the main menu
  • 2. Select Custom Nodes Manager button
  • 3. Enter OpenKatana in the search bar
After installation, click the Restart button to restart ComfyUI. Then, manually refresh your browser to clear the cache and access the updated list of nodes.

Visit ComfyUI Online for ready-to-use ComfyUI environment

  • Free trial available
  • 16GB VRAM to 80GB VRAM GPU machines
  • 400+ preloaded models/nodes
  • Freedom to upload custom models/nodes
  • 200+ ready-to-run workflows
  • 100% private workspace with up to 200GB storage
  • Dedicated Support

Run ComfyUI Online

OpenKatana · Run model workflow Description

Runs OpenKatana model workflows for image editing, upscaling, video interpolation, and restoration.

OpenKatana · Run model workflow:

OpenKatana_run_model_workflow lets you run model-based creative workflows through OpenKatana rather than manually assembling each processing step in ComfyUI. The supplied workflow code includes reference-guided image editing, model-based image upscaling, video frame interpolation, and video restoration, along with supporting generation and segmentation workflows. These workflows organize tasks such as loading installed models, preparing source media, running processing, and saving the resulting images or videos. For example, frame interpolation creates intermediate frames for smoother motion, while video restoration combines resizing with model-based processing and color correction. The node acts as a tool-call interface to OpenKatana, so available capabilities depend on the selected workflow, installed models, required ComfyUI nodes, and accessible execution services.

OpenKatana · Run model workflow Input Parameters:

project_id

A required STRING identifying the OpenKatana project for the tool call. The node passes this value into the argument-building process so the operation can be associated with the intended project. Use the identifier for the project you intend to work with, rather than assuming a project’s display name or folder path is interchangeable with its identifier. The supplied node interface specifies no default value, allowed identifier format, minimum or maximum length, or list of available projects; it also does not show project creation or project validation behavior.

arguments

A required STRING carrying the tool’s arguments in serialized form. For this tool, the underlying API signature identifies workflow settings named kind, parameters, assets, and request_id, along with service addresses named endpoint and native_endpoint; these belong within the tool arguments rather than being separate node inputs. The API declares parameters as a dictionary of workflow-specific settings and assets as a dictionary mapping strings to strings, but the supplied code does not establish the accepted asset keys or whether their values must be paths, identifiers, or another reference format. Workflow settings directly affect processing: for example, qwen-edit validates its reference count and resolution, while model-interpolation validates its frame multiplier and frame rate. Supply a valid JSON object using the tool’s supported argument structure, and include only settings accepted by the selected workflow. The node interface shows no default string, length limits, or enumerated options for this input. Separately, the underlying API defaults its execution addresses to http://127.0.0.1:8189 and http://127.0.0.1:8000; these are API defaults, not additional node inputs.

after

An optional STRING accepted by the node’s execution method. The supplied implementation does not read this value, add it to the tool arguments, or pass it into the tool invocation, so no direct effect on model settings or generated results is demonstrated. It may be available for connection-based ordering in a larger graph, but the provided code does not define a dependency-token format or a completion-checking mechanism. Leave it unconnected unless your surrounding workflow requires that connection. The execution method defaults it to None when omitted; no interface-level string default, allowed options, or minimum and maximum lengths are shown.

OpenKatana · Run model workflow Output Parameters:

result_json

A STRING containing the tool response after it has been processed by the node’s response helper. This is the node’s only output: it is not a directly connectable ComfyUI image, mask, audio, or video output. The underlying API declares a dictionary return type, and the output name indicates that the response is exposed as JSON text, making it suitable for inspection or downstream response parsing. The supplied workflow builders include image and video saving steps, but the response helper and API execution body are not provided, so exact response fields, saved-file references, job-status values, and error formatting cannot be documented reliably. Inspect the returned response before assuming that it contains a particular field or that it confirms completed processing rather than another execution state.

OpenKatana · Run model workflow Usage Tips:

  • Keep the node inputs separate from the tool settings: connect or enter project_id directly, and place workflow selection, model settings, asset mappings, and request information inside arguments using the tool’s supported JSON structure.

  • Use installed model filenames that match the selected workflow. Image upscaling uses an upscale-model loader, interpolation uses a frame-interpolation model loader, and the shown restoration graph requires both a model and a VAE. These components are not interchangeable.

  • For qwen-edit, start with one reference image and the default resolution of 1024. The supplied builder permits 1–16 references and a resolution from 0–2048, but resolution must be 0 or a multiple of 32. The excerpt permits 0 without explaining its sizing behavior, so use an explicit supported resolution when you need predictable dimensions.

  • For repeatable qwen-edit experiments, keep the seed fixed while changing the prompt or references. Its shown defaults are seed 1 and 25 steps; accepted ranges are 0–18446744073709551615 for the seed and 1–100 for steps.

  • For model-interpolation, choose the input frame rate and multiplier together. The default settings are 24 fps and a multiplier of 2, producing a 48 fps output. The permitted multiplier is 2–16 and the permitted input frame rate is 1–120, but their product must not exceed 120 fps.

  • Treat interpolation’s fps setting as an input-processing choice, not just an output label: the graph passes it to the video loader as force_rate and then creates the output at fps × multiplier. Choose a value appropriate for how you want the source video sampled.

  • The shown video-restoration workflow uses a scale default of 2, with an accepted range of 1–4. Start at a lower scale for testing, since increasing frame dimensions increases the amount of image data to process. The graph uses tiled encoding and decoding and temporal chunking, but these do not guarantee that every video will fit your available memory.

  • For image-upscale, select the appropriate installed upscaling model rather than assuming that the restoration workflow’s scale setting applies. The shown image-upscale graph uses the model directly and does not include a separate resize step.

  • Confirm asset mapping before submitting. The builders refer to staged filenames such as source.png, source.mp4, and reference-1.png, but these internal names alone do not establish the public assets mapping format.

  • Check execution-service addresses when running across containers or computers. The API defaults use 127.0.0.1, which refers to the environment running the service call, not necessarily the computer where you are viewing ComfyUI.

OpenKatana · Run model workflow Common Errors and Solutions:

Unsupported workflow parameters: Unsupported `{kind}` parameters: {sorted(extra)}

  • Explanation: The model-workflow builder rejects parameter names that are not in the selected workflow’s allowed set. A setting valid for one workflow may be invalid for another.
  • Solution: Check spelling and remove settings that do not belong to the selected workflow. For example, do not automatically copy restoration settings into image upscaling or interpolation.

resolution must be 0 or a multiple of 32

  • Explanation: The qwen-edit resolution does not satisfy the builder’s alignment requirement.
  • Solution: Choose a supported value such as 512, 1024, or 2048, staying within 0–2048. Use 0 only when you understand its behavior in the underlying processing node.

Output fps exceeds the core CreateVideo limit of 120

  • Explanation: For model-interpolation, the selected fps multiplied by multiplier exceeds 120.
  • Solution: Reduce either value. For example, 24 fps with a multiplier of 4 produces 96 fps, whereas 30 fps with a multiplier of 8 would exceed the limit.

Specify an already installed BiRefNet model filename

  • Explanation: The supplied segmentation builder raises this error when its model setting is missing or empty. This builder requires an existing BiRefNet model filename.
  • Solution: If segmentation is available through your selected workflow route, provide a nonempty filename matching an installed BiRefNet model. The excerpt does not show automatic model downloading.

Unknown workflow

  • Explanation: The model-workflow builder reaches its fallback error when it cannot match the requested workflow kind.
  • Solution: Check the exact workflow identifier and confirm that your installed OpenKatana version supports it. Identifiers shown in the supplied model-workflow code include qwen-edit, h3-reference, model-interpolation, and image-upscale; the complete accepted list is not included.

unexpected inputs: ...

  • Explanation: The node explicitly rejects extra execution fields that are not declared as typed inputs. API argument names are not automatically valid node sockets.
  • Solution: Use only the exposed inputs project_id, arguments, and optional after. Put settings such as kind and parameters inside the tool arguments rather than adding them as node inputs.

Invalid JSON or malformed tool arguments

  • Explanation: The node sends arguments through argument-processing helpers before invoking the tool. Malformed JSON or an incompatible structure can prevent the request from reaching execution; the exact error text is not shown.
  • Solution: Use double-quoted JSON keys and strings, remove trailing commas, and verify the structure against the tool’s argument requirements. Do not paste a Python dictionary representation into a JSON field.

Numeric parameter validation failure

  • Explanation: The workflow builders validate several numeric settings, although the supplied excerpts do not show the validators’ exact error messages. Unsupported values can stop workflow construction.
  • Solution: Check the selected workflow’s limits: qwen-edit references 1–16 and steps 1–100; interpolation multiplier 2–16 and fps 1–120, with output fps no higher than 120; the shown restoration scale 1–4. Use whole numbers where an integer is required.

Missing model, media file, or required ComfyUI node

  • Explanation: The generated graphs depend on installed loaders, processing nodes, model files, and staged source media. A missing dependency can cause graph validation or execution to fail; the exact message depends on the execution service.
  • Solution: Check the reported missing component, install the appropriate dependency, and verify model filenames and asset mappings. Do not assume that placing a filename in the request installs or uploads that file.

Connection failure to an execution service

  • Explanation: An execution service may be stopped or unreachable, particularly when default loopback addresses are used across different containers or machines. The supplied code does not show the exact connection error.
  • Solution: Confirm that the required services are running and reachable from OpenKatana. If needed, supply appropriate endpoint and native_endpoint values through the supported tool arguments rather than relying on the defaults.

OpenKatana · Run model workflow Related Nodes

Go back to the extension to check out more related nodes.
OpenKatana
RunComfy
Copyright 2025 RunComfy. All Rights Reserved.

RunComfy is the premier ComfyUI platform, offering ComfyUI online environment and services, along with ComfyUI workflows featuring stunning visuals. RunComfy also provides AI Models, enabling artists to harness the latest AI tools to create incredible art.

OpenKatana · Run model workflow