⭐ Star Split Sampler Info
Star Split Sampler Info — unpack everything StarSampler knows about your run
- split_info
- processing_time
- model
- text_encoder
- vae_model
- sampler
- scheduler
- denoise
- steps
- cfg
StarSampler (Unified) is the pack's big all-in-one sampler, and like a lot of powerful nodes it hands you one fat bundle of everything it knows about a run. That's great for passing along, terrible for looking at. Star Split Sampler Info is the wrench for that bundle: it takes the split_info output and splits it into nine individual, labeled sockets so you can read - or act on - each piece on its own.
What it unpacks
Connect the split_info output of StarSampler into the split_info input, and you get nine outputs:
processing_time(FLOAT) - how long the sampling took.model(STRING) - the model that was used.text_encoder(STRING) - the CLIP/text encoder; reports "Embedded in model" when it isn't a separate file.vae_model(STRING) - the VAE used.sampler(STRING) andscheduler(STRING) - the actual sampler/scheduler combo the run used.denoise(FLOAT),steps(INT),cfg(FLOAT) - the sampling parameters.
Why you'd actually use it
Two camps of use. First, debugging and logging: instead of squinting at a text blob, you get clean typed values you can wire into a display, a text writer, or a metadata node - great for building a "this is how this image was made" audit trail into your workflow. Second, conditional logic: those typed outputs are exactly what you feed into a routing or comparison node when you want the workflow to behave differently based on what actually ran. processing_time is the one people often overlook - a simple check on that lets you flag slow runs or log performance across generations.
It also pairs naturally with the other half of the split family: Star Split Sampler Option configures a split run, this node reports on whatever run happened. They're complementary tools for the same sampler.
Inputs and outputs
One input (split_info, type SAMPLER_INFO, from StarSampler's split_info socket). Nine outputs as above. It's a pure unpacker - nothing passes through, nothing samples. If you connect something that isn't a SAMPLER_INFO bundle, you'll get the fallback defaults ("Unknown Model", steps 20, cfg 7.0) rather than an error.
Installing
It's part of StarNodes:
cd ComfyUI/custom_nodes
git clone https://github.com/Starnodes2024/ComfyUI_StarNodes
cd ComfyUI_StarNodes
pip install -r requirements.txt
Restart, search star under ⭐StarNodes/Sampler. ComfyUI Manager users can search "Starnodes".
Gotchas
The one real catch: the split_info socket only exists on StarSampler (Unified) - a stock ComfyUI KSampler has no such output, so this node is useless there. It's a companion to the StarNodes sampler, not a general-purpose info node. And if your StarSampler version doesn't emit a full split_info, the defaults kick in silently; keep one eye on the model output when you first wire it up to confirm you're seeing real data.
Inputs (1)
| Name | Type | Default | Description |
|---|---|---|---|
| split_info | SAMPLER_INFO | Connect from StarSampler split_info output |
Outputs (9)
| Name | Type | Description |
|---|---|---|
| processing_time | FLOAT | — |
| model | STRING | — |
| text_encoder | STRING | — |
| vae_model | STRING | — |
| sampler | STRING | — |
| scheduler | STRING | — |
| denoise | FLOAT | — |
| steps | INT | — |
| cfg | FLOAT | — |