Inputs Diffusers Basic

If AUNInputsDiffusers is the full all-in-one for split diffusion models, AUNInputsDiffusersBasic is the same idea with the save-filename machinery cut out. It keeps the diffusion-only loading flow - standalone UNet, separate CLIP, separate VAE - plus the sampler settings and empty latent creation, and stops there. No MainFolder/Filename/name mode outputs, which is exactly why the pack points new workflows at this family: the Basic nodes plus a separate save node (like AUN Save Image V2) is the recommended, less legacy-heavy way to build.

What you get

Outputs: MODEL, CLIP, VAE, model name, sampler, scheduler, cfg, steps, latent, width, height, seed, batch size. That's the entire sampling contract in one node - a replacement for the three loader nodes (UNet + CLIP + VAE), Empty Latent, and whatever you were using to carry your sampler settings around.

Inputs mirror AUNInputsBasic but with diffusion-style loading:

• diffusion_name - the UNet-only model file.
• clip_name + clip_type - the CLIP file and its architecture (Flux, SD3, Hunyuan, etc.). The clip_type has to match the CLIP file; that's the one field to get right.
• vae_name - the VAE file.
• sampler, scheduler, cfg, steps, seed, batch_size.
• aspect_ratio / width / height with aspect_mode, plus megapixels and multiple for computing resolution from a ratio rather than typing pixel values.
• The SpeedLoRA trio (speed_lora, speed_lora_model, speed_lora_strength) for distilled fast checkpoints - and those want a low CFG, so the default cfg of 2 is intentional, not a typo.

The dropdowns read from the same model folders ComfyUI's core loaders use, so if you already load Flux (or similar) with stock nodes, the files you need are already listed here.

Installing it

One pack, one install:

• ComfyUI Manager: search "AUN ComfyUI Nodes", install, restart.
• Manual: cd custom_nodes && git clone https://github.com/loz2754/AUN-ComfyUI-Nodes, then restart.

Manual install and a ModuleNotFoundError: cv2 error? pip install -r custom_nodes/AUN-ComfyUI-Nodes/requirements.txt.

Common issues

The classic failure is clip_type not matching the CLIP file - double-check it before blaming anything else. And if the dropdowns show <no diffusion models found> or similar, that's ComfyUI telling you the relevant model folders are empty; it's a download-it-first situation, not a broken node. One more: because this is the Basic contract, don't expect save-prep outputs - if your workflow reads MainFolder/Filename from a fuller Inputs node, wire those up with a filename builder on the save side instead.