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 > ComfyUI_LC_ModelBuilder_nodes > LC Anima LoRA Editor

ComfyUI Node: LC Anima LoRA Editor

Class Name

LCAnimaLoRAEditor

Category
LC ModelBuilder/Anima
Author
lonecatone23 (Account age: 0days)
Extension
ComfyUI_LC_ModelBuilder_nodes
Latest Updated
2026-11-14
Github Stars
0.01K

How to Install ComfyUI_LC_ModelBuilder_nodes

Install this extension via the ComfyUI Manager by searching for ComfyUI_LC_ModelBuilder_nodes
  • 1. Click the Manager button in the main menu
  • 2. Select Custom Nodes Manager button
  • 3. Enter ComfyUI_LC_ModelBuilder_nodes 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

LC Anima LoRA Editor Description

Adjusts Anima LoRA by block and layer, softens spikes, and outputs a model and savable LoRA.

LC Anima LoRA Editor:

LCAnimaLoRAEditor lets you adjust how an Anima LoRA affects different parts of an image without retraining it. It loads a LoRA, varies its influence across Anima’s 28 model blocks and layer types, and can soften unusually strong layers or dominant directions within a layer. This gives you a way to test whether a LoRA’s desired effect can be retained while reducing unwanted changes, such as an overly fixed face. The node applies the edited LoRA to an Anima model and displays a heatmap comparing layer strength before and after the edits. The same edits are also prepared for saving as a separate LoRA.

LC Anima LoRA Editor Input Parameters:

model

The Anima model to which the edited LoRA is applied. Connect a compatible MODEL so you can use the edited result in the rest of your image-generation workflow. This input has no numeric range or default.

lora_name

The LoRA to edit, selected from the files available in models/loras. Choose a LoRA intended for the connected Anima model: the node loads its layers and attempts to match them to that model. Available options depend on the LoRA files installed in your environment; no fixed default is shown in the provided source.

strength

The overall LoRA strength, comparable to the strength setting on a LoRA loader. It scales the edited LoRA applied to the model and is baked into the LoRA prepared for saving. A value of 1.0 uses the selected edits at their set strength, 0 removes their effect, and negative values reverse the applied contribution. Default: 1.0; minimum: -4.0; maximum: 4.0; step: 0.05.

preset

A ready-made 28-block strength curve. Full leaves every block at 1.0; each available Anima block-group preset sets its selected blocks to 0.3 and leaves the others at 1.0. Presets provide a quick way to compare which group may be responsible for an unwanted effect. The source shown here does not list the names or block ranges of the other ANIMA_PRESETS, so use the options displayed by the node. Default: Full. A connected blocks_in curve takes precedence over this setting.

blocks

A string containing 28 block multipliers, corresponding to Anima’s 28 blocks. The default is 28 values of 1, meaning no per-block reduction or increase. The tooltip specifies a 0 to 2 range for these multipliers: 0 removes a block’s contribution, 1 leaves it unchanged, and values above 1 increase it. You can adjust the node’s bars to build a custom curve. A selected preset supplies its own curve when one is available; a connected blocks_in curve overrides both the preset and the bars.

attention

A multiplier for self-attention and cross-attention layers across the blocks. Lowering it reduces the edited LoRA’s contribution in those layers, while raising it increases that contribution; it works alongside each block’s multiplier rather than replacing the block curve. Default: 1.0; minimum: 0.0; maximum: 2.0; step: 0.05.

mlp

A multiplier for MLP layers across the blocks. Use it to change the LoRA’s influence in these layers separately from attention, while retaining the shape of the selected block curve. A value of 1.0 leaves this layer-type multiplier unchanged; 0.0 removes the affected contribution. Default: 1.0; minimum: 0.0; maximum: 2.0; step: 0.05.

outside_blocks

A multiplier for matching LoRA layers outside Anima’s 28 blocks, such as an LLM adapter or final layer, if the selected LoRA contains them. It has no layers to affect when the LoRA contains none outside the blocks; the heatmap note reports how many were found. Default: 1.0; minimum: 0.0; maximum: 2.0; step: 0.05.

spike_soften

Selects whether to reduce unusually concentrated LoRA strength. Options are off (no spike softening), layers (cap layers that are much stronger than comparable layers in other blocks), directions (soften a dominant direction within an eligible plain LoRA layer), and both (apply both methods). Direction softening applies only to plain lora_A/lora_B layers supported by the implementation; it does not apply to every adapter type. Default: off.

spike_limit

Sets the threshold used by the selected spike_soften method. For layers, the threshold is a multiple of the median strength of the corresponding layer type across blocks. For directions, it is a multiple of the layer’s second-strongest direction. Lower values permit more softening; higher values leave more strong components untouched. This setting has no softening effect while spike_soften is off. Default: 2.0; minimum: 1.0; maximum: 10.0; step: 0.1.

blocks_in

An optional connected STRING containing a 28-block curve, for example from LC Anima LoRA Block Scan’s blocks output. When connected, its curve replaces the preset and the node’s block bars for this edit. This is useful for carrying a scanned or prepared block curve into the editor. It has no default value; its expected content is a valid 28-block curve.

LC Anima LoRA Editor Output Parameters:

model

The MODEL output is a clone of the connected Anima model with the edited LoRA applied. Connect it to the next model-consuming node to generate images and evaluate the changes. The original input model is not itself returned as the edited result. The heatmap note reports how many loaded LoRA layers were applied, which helps you assess whether the selected LoRA matched the model.

edited_lora

The LC_LORA output contains the edited LoRA data for LC LoRA Save. It carries the edited layers along with information about the source LoRA and edit settings, allowing the result to be saved separately for later use. This is LoRA data, not a generated image or a second model.

LC Anima LoRA Editor Usage Tips:

  • Start with Full, strength at 1.0, all layer multipliers at 1.0, and spike_soften set to off. Generate a baseline image before changing one setting at a time.
  • Test block-group presets using the same prompt and fixed seed. A preset reduces its selected blocks to 0.3; compare the images to see whether it reduces an unwanted effect while preserving what you like about the LoRA.
  • Use attention and mlp when you want to adjust a layer type across all blocks. Use the block bars or blocks_in when you want to target particular blocks. These multipliers combine, so several reductions can produce a stronger overall change than any one setting alone.
  • Try spike_soften set to layers when the heatmap shows an exceptionally strong layer relative to the same layer type in other blocks. Try directions for eligible plain LoRA layers when you want to test whether a dominant internal direction is driving the effect. Keep spike_limit at its 2.0 default initially, then adjust it gradually while comparing images.
  • Check the heatmap and its note after each edit. The outline represents strength before editing and the fill represents strength after the block, layer-type, and spike edits; the note also flags layers outside the blocks and reports how many layers applied. The displayed “strength kept” calculation does not include the separate overall strength multiplier, so judge that setting through generated images as well.

LC Anima LoRA Editor Common Errors and Solutions:

unknown preset: <name>

  • Explanation: This validation message occurs when the saved or selected preset name is not among the Anima presets recognized by the node.
  • Solution: Select a preset currently shown in the node, such as Full. If the workflow came from another version or a different architecture’s editor, replace its old preset selection with an available Anima preset.

LoRA file cannot be found

  • Explanation: The node resolves lora_name from the installed LoRA files. A workflow can refer to a file that is missing, renamed, or no longer in models/loras. The exact file-not-found wording comes from the file-loading environment and is not specified in the provided node source.
  • Solution: Confirm the file exists in your LoRA folder, refresh the available model-file lists if necessary, and select its current name from lora_name.

Invalid or unreadable LoRA file

  • Explanation: Loading can fail before edits begin if the selected file cannot be read as a supported LoRA. The provided source does not specify a single error string for this case.
  • Solution: Confirm that the selected file is the intended LoRA and opens normally in your workflow. Try a known-working Anima LoRA to determine whether the problem is specific to that file.

Invalid 28-block curve

  • Explanation: The editor expects a 28-block curve from blocks_in when that input is connected, or from blocks when using a custom curve. The provided source does not show the parser’s exact error message or accepted text syntax.
  • Solution: Use the node’s block bars or connect the blocks output of LC Anima LoRA Block Scan. If entering a curve manually, begin with the default 28 values of 1 and keep each multiplier within the tooltip’s 0 to 2 range.

some layers do not match this model: is it a Anima LoRA?

  • Explanation: This is a heatmap note from the node, not necessarily a fatal error. It appears when fewer LoRA layers are applied than were loaded, which can indicate that the LoRA does not match the connected model.
  • Solution: Check that model is an Anima model and lora_name is a LoRA made for that model or a compatible Anima architecture. Review the note’s applied-layer count before relying on the edited result.

Edited LoRA has little or no visible effect

  • Explanation: This may be a configuration outcome rather than an error. strength at 0, block multipliers at 0, or attention and MLP multipliers at 0 can remove contributions from the affected layers; a mismatched LoRA may also apply only partly.
  • Solution: Restore strength to 1.0, select Full, set attention, mlp, and outside_blocks to 1.0, and turn spike softening off. Then check the applied-layer count and add edits back one at a time.

LC Anima LoRA Editor Related Nodes

Go back to the extension to check out more related nodes.
ComfyUI_LC_ModelBuilder_nodes
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.

LC Anima LoRA Editor