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 · Morph images

ComfyUI Node: OpenKatana · Morph images

Class Name

OpenKatana_morph_images

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 · Morph images Description

Creates animated transitions between two images using feature matching and geometric morphing.

OpenKatana · Morph images:

OpenKatana_morph_images creates a gradual animated transformation between two local images using geometric image matching and piecewise-affine morphing. Rather than simply fading one image into another, it divides the images into corresponding triangular regions, gradually moves those regions, and blends their colors to create intermediate frames. It can find matching visual features automatically with OpenCV’s ORB feature detector and RANSAC filtering, or use explicitly supplied landmarks when automatic matching is unsuitable. Automatic matching follows visual patterns and geometry, not the meaning of objects, so it does not inherently understand facial anatomy or object identity. The process preserves image proportions by fitting images onto a padded canvas, checks that the triangular regions do not fold or collapse during the transition, and refuses unreliable transformations instead of silently producing invalid geometry. This makes the node useful for controlled image-to-image transitions and editable geometric animation, rather than semantic or generative video creation.

OpenKatana · Morph images Input Parameters:

project_id

A required STRING that identifies the OpenKatana project in which the morph operation runs. The underlying API resolves the project and checks its state before generating project-managed output, keeping the animation associated with the appropriate project workspace. Supply the identifier of an existing, accessible project rather than an image filename or output folder. The provided node interface does not specify a default value, a list of permitted identifiers, or minimum and maximum string lengths.

arguments

A required STRING containing the tool arguments, supplied as a JSON object. This is where you specify the morph settings; they are not separate sockets on this node. The underlying morph API accepts source and target as local image paths, plus optional width, height, fps, frames, and points settings. Width and height each default to 512, must be integers from 32 through 4096, and must be even; the images are fitted to this canvas with preserved proportions and black padding rather than stretched. The frame rate defaults to 30 and accepts integers from 1 through 120. The frame count defaults to 60 and accepts integers from 2 through 3600; more frames create more intermediate steps and increase processing and storage requirements. Approximate encoded duration is frame count divided by frame rate, so 60 frames at 30 fps produces approximately two seconds of video. The optional points value defaults to None in the API and represents explicit normalized landmark correspondence; omitting it selects automatic feature matching. The supplied excerpts do not show the required internal structure of the landmark dictionary, so consult the tool’s landmark specification before constructing it. For a basic operation, an arguments string can contain {"source":"/path/to/start.png","target":"/path/to/end.png","width":512,"height":512,"fps":30,"frames":60}. The node interface does not expose a default string, string-length limits, or a fixed options list for arguments.

after

An optional STRING accepted by the node’s shared tool interface. In the shown execution method, this value is accepted but is not explicitly read or forwarded to the morph API, so the supplied code does not establish a direct effect on image matching, animation settings, or generated files. Do not use it to supply image paths or morph configuration; those belong in arguments. Its name alone is not sufficient evidence of execution-order behavior, and no such behavior is demonstrated in the provided implementation. The execution method defaults it to None when omitted; no string default, allowed options, or minimum and maximum lengths are shown.

OpenKatana · Morph images Output Parameters:

result_json

A STRING containing the serialized response from the OpenKatana tool invocation, rather than a ComfyUI IMAGE tensor or a directly connectable video output. The underlying morph operation is documented as returning an editable correspondence receipt and a timeline clip, allowing you to inspect the morph correspondence and use the generated animation in the project workflow. The API explicitly converts the clip’s frames_dir path to a project-relative path, so interpret that location relative to the selected project, not necessarily relative to ComfyUI’s current working directory. On successful processing, the implementation writes numbered PNG frames such as 00000.png into a frames directory and encodes a video.mp4 file in the operation’s output directory. The first and last frames are the fitted source and target images respectively, including any padding. The complete response wrapper and receipt field structure are not shown in the supplied code, so inspect the returned JSON rather than assuming additional field names or a particular error-response layout.

OpenKatana · Morph images Usage Tips:

  • Start with the underlying defaults of 512 × 512, 30 fps, and 60 frames. This provides a practical preview before you commit to larger images or longer sequences.

  • Choose images with shared, clearly visible details and similar composition for automatic matching. Consistent textures, edges, and recognizable visual patterns give the geometric matcher more useful information than large blank areas or radically different subjects.

  • Use explicit landmark correspondence when you need deliberate alignment, especially for features such as eyes, mouth corners, or object edges. Automatic ORB matching is geometric, not semantic, and cannot guarantee that visually similar points represent the same anatomical or artistic feature. Verify the accepted points structure before supplying it.

  • Keep landmark pairings spatially consistent. Avoid crossing correspondences or moving neighboring points past each other, because the node checks every requested frame for triangular regions that fold or collapse.

  • Set width and height inside arguments to even integers within 32–4096. Choose a canvas aspect ratio close to your images to minimize black padding; the implementation preserves proportions rather than stretching images to fill the canvas.

  • Adjust frames and fps together. For approximately four seconds at 30 fps, use 120 frames; increasing fps alone while leaving the frame count unchanged shortens the clip rather than creating extra intermediate frames.

  • Test at a smaller canvas size and frame count before rendering at high resolution. The node processes triangular warps for intermediate frames, saves every frame as a PNG, and then encodes the video, so larger settings increase both work and disk usage.

  • Keep the source files unchanged until processing finishes. The implementation compares their content digests before and after generation and rejects the operation if either input changes.

  • Inspect result_json for the returned clip and correspondence information. Resolve the clip’s frames_dir against the project directory when accessing individual frames.

OpenKatana · Morph images Common Errors and Solutions:

Install the OpenKatana vision extra for image morphs

  • Explanation: OpenCV could not be imported in the Python environment running OpenKatana. The node needs it for feature matching and geometric image warping.
  • Solution: Install OpenKatana’s vision extra according to its installation instructions, using the same Python environment that runs ComfyUI, and restart ComfyUI afterward.

No reliable image features; provide explicit landmarks

  • Explanation: The automatic detector could not obtain feature descriptors from one or both images. Images with very little detail, large uniform areas, or weak contrast may not provide suitable visual features.
  • Solution: Try images with clearer shared detail, or supply explicit normalized landmarks using the supported points format.

Too few reliable feature matches; provide explicit landmarks

  • Explanation: Automatic matching found fewer than eight matches that passed its reliability filter. Detecting details in both images is not enough; enough of those details must also match between the images.
  • Solution: Use images with more overlapping visual structure and similar framing, or provide explicit landmarks to define the intended correspondence.

Feature registration failed; provide explicit landmarks

  • Explanation: RANSAC could not establish enough geometrically consistent matches. The implementation requires at least eight accepted matches after this filtering stage, and rejects a missing or insufficient result.
  • Solution: Choose a more closely related image pair or use explicit landmarks. This is a correspondence problem, so changing frame rate or output duration will not resolve it.

Use even morph dimensions

  • Explanation: The requested width or height is odd. Both canvas dimensions must be even in addition to meeting the supported size range.
  • Solution: Set both values to even integers, such as 512 × 512, 768 × 512, or 1024 × 1024, with each dimension between 32 and 4096.

Landmarks fold/collapse a triangle at frame …

  • Explanation: At the indicated frame, a triangular region becomes too small or reverses orientation during interpolation. This indicates that the correspondence would make part of the image collapse or turn inside out.
  • Solution: Revise the landmark pairings so neighboring features remain in a consistent spatial order throughout the transition. If automatic matching caused the problem, use carefully chosen explicit landmarks. Do not rely on reducing the frame count to hide an invalid transformation.

Morph triangulation left uncovered pixels

  • Explanation: The triangular warps did not cover every pixel of an intermediate frame. The node stops rather than leaving gaps in the generated image.
  • Solution: Review the correspondence for poorly distributed or problematic points, particularly near image boundaries. If you supplied landmarks, check them against the supported landmark specification; if the problem persists with valid correspondence, retain the settings and report the case to the extension maintainer.

Morph inputs changed during processing

  • Explanation: The content of the source or target file changed between the initial and final integrity checks. This can happen if another application or workflow overwrites an input while the morph is rendering.
  • Solution: Save stable copies of both images, use those copies as inputs, and avoid modifying or replacing them until the operation finishes.

Invalid numeric settings

  • Explanation: The implementation validates numeric settings before generating frames. The exact validation error text is not shown, but out-of-range or non-integer values can fail these checks.
  • Solution: Supply integer values in arguments: width and height from 32 through 4096, fps from 1 through 120, and frames from 2 through 3600. Width and height must also be even.

Invalid arguments JSON or unsupported argument names

  • Explanation: The node processes the arguments string before invoking the tool. Malformed JSON or names not supported by the morph API can prevent execution, although the exact parser error messages are not included in the supplied code.
  • Solution: Use a valid JSON object with double-quoted keys and string values, and no trailing commas. Put image paths in source and target and use the documented morph option names. Do not add out as a configurable API argument; the project API creates the output location.

unexpected inputs: …

  • Explanation: The node’s execution method explicitly rejects additional fields that are not declared in its typed interface. This differs from supplying supported morph options inside the arguments JSON string.
  • Solution: Use only the exposed node inputs: project_id, arguments, and optionally after. Place morph configuration inside arguments rather than adding extra node-level fields.

Image file cannot be opened

  • Explanation: The image loader opens local files with Pillow, so a missing, inaccessible, corrupt, or unsupported image can stop the operation. The exact error text depends on the underlying file or image-reading failure.
  • Solution: Confirm that both paths point to readable image files on the machine running ComfyUI. Open the images in another application to check them, and re-save them as standard PNG or JPEG files if necessary.

Output directory already exists or cannot be created

  • Explanation: The morph implementation creates a new output directory with existing directories disallowed. A pre-existing destination or insufficient filesystem permissions can therefore cause a filesystem error; the exact message depends on the operating system.
  • Solution: Confirm that the project workspace is writable and that the operation can create a fresh output location. Investigate any conflicting directory before removing it, and preserve previous results you still need.

Video encoding failed

  • Explanation: After saving the PNG sequence, the implementation calls a video encoder to create video.mp4. The provided context does not show the encoder’s dependencies or exact failure messages.
  • Solution: Check the detailed execution log and OpenKatana’s video-encoding setup, confirm that sufficient disk space is available, and verify that the output directory is writable. Check whether the frame sequence was saved before rerunning the operation.

OpenKatana · Morph images 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 · Morph images