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-lora-FindingLora > LoRA Loader (Finding LoRA)

ComfyUI Node: LoRA Loader (Finding LoRA)

Class Name

LoraLoaderFindingLora

Category
loaders
Author
shootthesound (Account age: 1559days)
Extension
comfyui-lora-FindingLora
Latest Updated
2026-07-23
Github Stars
0.01K

How to Install comfyui-lora-FindingLora

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

LoRA Loader (Finding LoRA) Description

Applies a main LoRA and a stack of LoRAs to a model at set strengths and outputs saved triggers.

LoRA Loader (Finding LoRA):

LoraLoaderFindingLora applies LoRA files to a diffusion model so you can use a particular style, subject, or other learned effect in your artwork. It can apply a main LoRA followed by a stack of additional LoRAs, each at its own strength. It also supports bookmarks, saved trigger words or phrases, and a searchable LoRA selection interface, making a large collection easier to navigate. The node modifies the model only; it does not modify a CLIP connection. Saved trigger text is made available separately so you can incorporate it into your prompting workflow.

LoRA Loader (Finding LoRA) Input Parameters:

model

The required MODEL connection is the diffusion model you want to modify. The node applies the selected LoRAs to this model in sequence, then passes the resulting model onward. Connect a model from an upstream loader before running the workflow. There is no numeric minimum, maximum, or default for this connection.

lora_name

This required selection identifies the main LoRA. Its options are the filenames currently available in ComfyUI’s loras folder list; the source does not define a fixed list or a default choice. The node processes this selection before any entries in the additional stack. If its value is None or empty, the main LoRA is not loaded, although valid stack entries can still be applied. Choose a file intended for the model you connected.

strength_model

This required FLOAT value controls how strongly the main LoRA modifies the diffusion model. Its default is 1.0, its minimum is -100.0, its maximum is 100.0, and its adjustment step is 0.01. A value of 0 skips loading and applying the main LoRA; other values, including negative ones, are passed to the LoRA loader. This setting does not control the strengths of additional stack entries, which have their own values.

lora_stack

This optional STRING contains additional LoRAs to apply after the main one. Its default is [], an empty JSON array, and the node’s stack interface is intended to manage up to 10 entries with a LoRA name and an individual strength for each. For example, [`{"lora_name":"example.safetensors","strength":0.5}`] illustrates the stored format, but you should select an actual available filename. An entry without a usable name is ignored; a missing or nonnumeric strength becomes 1.0; and an entry with strength 0 is not applied. Invalid JSON or a value that is not a JSON array quietly becomes an empty stack. There is no numeric minimum or maximum for the string itself.

LoRA Loader (Finding LoRA) Output Parameters:

model

This MODEL output is the connected model after the node has applied every eligible LoRA, starting with the main selection and then proceeding through the stack in order. An entry with zero strength, an empty name, or the name None does not modify the model. Connect this output to the next model-consuming node in your workflow; there is no CLIP output from this node.

trigger

This STRING output contains saved trigger words or phrases found in bookmarks for the main LoRA and stack entries, in that order, separated by , . Empty triggers and duplicate trigger strings are omitted; when no selected names have saved triggers, the output is an empty string. Trigger collection uses the listed names even when an entry’s strength is 0, so text here does not necessarily mean that its LoRA was applied. The string is available for you to use in a prompt workflow; the node does not insert it into a prompt automatically.

LoRA Loader (Finding LoRA) Usage Tips:

  • Start with one LoRA and the default strength_model of 1.0, then adjust its strength in small steps while comparing results. Set it to 0 when you want to test the workflow without applying the main LoRA.
  • Add stack entries only when you need to combine effects. Each valid entry is applied after the main LoRA, so the order of entries determines the order in which the model is modified.
  • Save trigger phrases in bookmarks for LoRAs that need them, and connect trigger to the appropriate text-handling part of your workflow. Check the emitted text before relying on it: only bookmarked names with nonempty saved triggers contribute.
  • Remove LoRAs you are no longer using from the stack. The node retains loaded LoRA data for paths referenced during a run and evicts cached data for paths no longer used by that node.
  • Use the node’s stack interface to manage entries where available. If you edit the stored JSON directly, keep it as an array of records with lora_name and strength fields; malformed JSON silently disables the additional stack.

LoRA Loader (Finding LoRA) Common Errors and Solutions:

LoRA file not found (filename shown in error)

  • Explanation: The node resolves each nonzero-strength LoRA name to a file. If a selected file is missing or its saved name no longer resolves, path lookup can stop the run. The exact message comes from ComfyUI’s file-path lookup rather than this node.
  • Solution: Recheck the main selection and every stack entry against the currently available LoRA filenames. Restore a moved file or select its new name, then rerun the workflow.

Error loading LoRA file (file path shown in error)

  • Explanation: The node loads each selected file using ComfyUI’s safe file loader. An unreadable, damaged, or unsupported file can cause loading to fail; the exact message depends on the loader and file.
  • Solution: Identify the path in the error, verify that the file is a working LoRA, and replace or remove it if necessary. Test it as the only selected LoRA to isolate the problem.

Error applying LoRA to model

  • Explanation: After loading a file, the node asks ComfyUI to apply it to the connected model. An incompatible LoRA or a model-related loading problem can fail at this stage; the exact error text is supplied by ComfyUI.
  • Solution: Check that the LoRA is intended for your model family. Test the main LoRA and stack entries one at a time to find the combination that causes the failure.

No visible LoRA effect

  • Explanation: The main LoRA or a stack entry is skipped when its strength is 0, its name is empty, or its name is None. A valid but low strength may also produce an effect that is difficult to notice.
  • Solution: Confirm that the intended file is selected and its strength is nonzero. Temporarily test one LoRA at a time at a moderate strength to check whether it affects your result.

Additional LoRAs do not apply, with no error

  • Explanation: The stack parser treats invalid JSON, a non-array JSON value, or a non-string value as an empty stack. It also ignores entries without a valid lora_name and processes no more than the first 10 entries.
  • Solution: Use the stack rows on the node when available. For manually edited data, verify that the value is a JSON array, each intended entry has a nonempty string lora_name, and the entries you need are among the first 10.

Trigger output is empty or missing a phrase

  • Explanation: The node emits trigger text only when it finds a bookmark for a listed LoRA name with a nonempty saved trigger value. It skips empty values and repeats of an identical trigger string.
  • Solution: Check the bookmark associated with each selected LoRA and save the intended trigger phrase there. Confirm that the selected filename matches the bookmarked entry.

Trigger text appears for a LoRA that was not applied

  • Explanation: Trigger collection examines the main name and stack names independently of their strengths. A bookmarked LoRA set to strength 0 can therefore contribute trigger text even though it does not modify the model.
  • Solution: Remove or change the unused LoRA selection if you do not want its trigger emitted. Do not use the trigger string alone as confirmation that every listed LoRA was applied.

LoRA Loader (Finding LoRA) Related Nodes

Go back to the extension to check out more related nodes.
comfyui-lora-FindingLora
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.

LoRA Loader (Finding LoRA)