Inputs Hybrid

AUNInputsHybrid is the shape-shifter of the Inputs family: one node that loads either a classic combined checkpoint (UNet+CLIP+VAE in one file) or a diffusion-only UNet with separate CLIP and VAE files, and from there behaves essentially like AUNInputs - the full all-in-one with sampler settings, latent creation, and the save-prep outputs. It exists so you don't have to rebuild your Inputs setup when you switch between model styles. Have a combined SDXL checkpoint in one workflow and a split Flux setup in another? This node covers both.

How it works

The switch is model_source. Set it to Checkpoint and the node loads ckpt_name - a standard combined file, with clip_skip applied. Set it to Diffusion model and it loads diffusion_name (UNet only) plus clip_name/clip_type and vae_name as separate files, and clip_skip still applies. The rest of the contract - sampler, scheduler, cfg, steps, aspect ratio, seed, SpeedLoRA, and the filename outputs - is the same AUNInputs-style surface either way. SpeedLoRA works in both modes too.

So the practical story: build once, swap model_source when you change model families. The trade-off is that the node carries both model-loading widget sets (ckpt_name and the diffusion trio), so it's the busiest of the Inputs nodes. In practice you only touch the widgets relevant to your current model_source, and the others sit inert.

Inputs and outputs

Inputs: model_source, then the checkpoint or diffusion trio depending on mode, clip_skip, sampler/scheduler/cfg/steps, aspect_ratio/width/height/aspect_mode, batch_size, seed, megapixels/multiple, SpeedLoRA, and the save-prep set (MainFolder, ManualName, name_mode, prefix, date_format, crop, words).

Outputs: MODEL, CLIP, VAE, model name, sampler, scheduler, cfg, steps, latent, width, height, seed, MainFolder, Filename, prefix, date format, batch size, name mode.

Installing it

Same pack, same 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 installs and a ModuleNotFoundError: cv2 error? pip install -r custom_nodes/AUN-ComfyUI-Nodes/requirements.txt.

Common issues

The main confusion is leaving widgets set from the wrong mode - if you're in Checkpoint mode, the diffusion dropdowns don't matter, and vice versa. When switching model_source, give the new mode's widgets a once-over rather than assuming the old values carry meaning. And because this is a full Inputs-style node, the same "legacy directionally" note applies: the pack steers new builds toward the Basic nodes plus AUN Save Image V2, so treat the Hybrid's filename outputs as the older contract.