QwenImage Layered V2 Loader (compat)

This node and QwenImage Layered V2 Loader are the same code. Not "similar" - literally the same Python class, registered twice under different names in the pack's NODE_CLASS_MAPPINGS. The (compat) in the display name is the tell: it's a backwards-compatibility alias kept alive so early local workflows that referenced the old name still load without you hunting down which node changed.

So the entire article for the compat version is: it loads the V1 base model plus the V2 control LoRA, assembles DiffSynth's QwenImagePipeline with the official Qwen text encoder, tokenizer and layered VAE, and outputs a single pipeline socket (QWEN_LAYERED_V2_PIPELINE) that feeds QwenImage Layered V2 Decompose. All seven inputs - v1_model, v2_lora, allow_official_download, download_source, device, vram_mode, vram_margin_gb - and every install requirement are identical to the V2 node.

Which one should you add to your graph?

If you're starting fresh, use the V2 loader. It's the name the pack's README and example workflows use, it's the one that'll show up in Manager updates and future fixes, and the compat alias is exactly the kind of thing that gets dropped in a cleanup release. If you open a saved workflow and see this node sitting there, leave it - it works, it's the same weights and the same pipeline, and deleting and re-adding it buys you nothing.

The gotchas that still apply

Because it is the V2 loader, it inherits the full setup burden: DiffSynth-Studio must be installed in ComfyUI's Python environment (the pack's requirements.txt only hints at it), the official Qwen/Qwen-Image text encoder/tokenizer and Qwen/Qwen-Image-Layered VAE must be cached (ComfyUI's FP8 text encoder repack and ComfyUI layered VAE won't work), and allow_official_download defaults to False - meaning a missing official file is a hard error, not a silent download. All of that is covered in detail on the V2 loader's page. This node doesn't get an easier version of any of it.

Short version: don't type its name into the search bar if you can help it, but don't panic if it's already in your graph.