Nodes/ComfyUI-Texturaizer/Get LoRA Data (Texturaizer)
ComfyUI Node

Get LoRA Data (Texturaizer)

The LoRA stack Blender picked, ready to apply

By LatentSpaceDirective·Created 2 years ago·Updated 8 months ago· 21
Get LoRA Data (Texturaizer)
  • data_optional
  • LoRAs
  • data_hash
directory_optional

Retrieves whatever LoRA setup Blender configured for a render - the README lists "full support for LoRA models" as one of Texturaizer's headline features, and this is the node that carries that configuration from the Blender addon into ComfyUI. Rather than hardcoding a fixed set of Load LoRA nodes in your graph, you read the LoRA list Blender exported and apply whatever's actually in it.

How it works

Standard Texturaizer getter behavior: directory_optional points it at the export folder, leaving it blank falls back to the addon's global directory, and data_optional accepts an already-loaded dictionary if you're chaining from Texturaizer_GetJsonData upstream rather than re-reading the export file per node. From that data it pulls out the LoRA configuration block - presumably a list, since Texturaizer's Blender side supports assigning different things (LoRAs among them) per object or material, meaning more than one LoRA can plausibly be active on a single render.

The outputs that matter

  • LoRAs - a dictionary, described in the tooltip as "LoRA data." The schema doesn't break this down field-by-field the way Texturaizer_GetIPAdapterData does for its own data, so treat it as a bundle you'll either feed straight into a downstream node built to consume it, or unpack yourself if you're wiring LoRA application by hand - most likely holding, per LoRA, at minimum a filename and a strength value, mirroring what a normal Load LoRA node needs.
  • data_hash - the standard debug-only fingerprint, not meant to be wired into your generation path.

There's no dedicated "Extract LoRA Data" or "Apply LoRAs" node in this pack the way there is for ControlNets (Texturaizer_ExtractCNData / Texturaizer_ApplyControlNets), so getting from this dictionary to an actual applied LoRA on your model and CLIP is left to you - either a custom node elsewhere in your setup that reads the same dictionary shape, or manual unpacking.

Installing it

ComfyUI Manager → search Texturaizer → Install → restart, or by hand:

cd ComfyUI/custom_nodes
git clone https://github.com/LatentSpaceDirective/ComfyUI-Texturaizer

then restart ComfyUI. No extra dependencies from this node itself - the LoRA files it references still need to exist in your ComfyUI/models/loras folder, and applying them still runs through your usual LoRA loader.

Where people get burned

Because the pack doesn't ship a matching "Apply LoRAs from dictionary" node, it's easy to wire LoRAs output straight into a node that expects a single filename-and-strength pair rather than a bundle covering possibly several - check what shape your downstream apply logic actually expects before assuming a type mismatch is a bug in this node rather than a wiring mistake on your end. And as with every getter in this pack: without the Blender addon actually exporting, directory_optional left blank is reading from a directory that has nothing in it until Blender configures and writes one.

CategoryTexturaizer

Inputs (2)

NameTypeDefaultDescription
directory_optionaloptSTRING
data_optionaloptDICTIONARY[object Object]

Outputs (2)

NameTypeDescription
LoRAsDICTIONARYLoRA data
data_hashSTRINGHash value for debugging purposes.