Bubba Merge Naming Helper

Merging checkpoints produces an awkward truth: the honest name for a weighted blend of two models is basically the whole recipe, and recipe names are unreadable. Bubba Merge Naming Helper is the little node that solves the naming half of the merge workflow - it takes a base name, an output folder, and an optional suffix, and produces a clean, sanitized relative filename prefix that Bubba Save Checkpoint uses for the merged safetensors file.

It exists because the pack's merge pipeline is otherwise fully automatic: Bubba Checkpoint Merge blends the models and even suggests a name (something like modelA_0.50-modelB_0.50), but you probably want your merges filed under your own naming scheme, in your own folder. That's what this node is for - the human decision point in an otherwise mechanical process.

The three inputs

• base_name - the core name for the merged checkpoint. The thing you'll actually recognize in the models dropdown later.
• folder - the subfolder under models/checkpoints where the file lands. Defaults to Bubba_Merges, which is a decent convention if you don't want your merges mixed in with your downloaded models.
• suffix - optional extra text, appended to the name.

Optionally, you can connect a checkpoint_merge (the payload from Bubba Checkpoint Merge) - when connected, the node can pull recipe context from it. The output is filename_prefix, a STRING like Bubba_Merges/my_merge_v2 that feeds Bubba Save Checkpoint's filename input, plus an info string for display.

The "helper" part is the sanitization: it cleans the name into something filesystem-safe - no path separators sneaking in, no characters that'll break on Windows, a controlled length. If you've ever merged in another tool and gotten a file with illegal characters or a name that loops back on itself, this is the node that refuses to let that happen. The pack's underlying utility also understands the safetensors convention - a Bubba_Merges/name prefix is enough for the save node to write models/checkpoints/Bubba_Merges/name.safetensors.

Why a whole node for this

Because merge workflows are iterative and batch-y: you'll blend at ratio 0.3, then 0.5, then 0.7, and without a naming plan you end up with merge (1).safetensors, merge (2).safetensors, and no memory of which ratio made which. A naming node that always gives you the same structure - folder/base_suffix - means every merge you produce is findable and comparable. Wire the merge_recipe or info from the merge node into a suffix, and your filename encodes the recipe.

Honest caveat: this node does one thing and does it quietly. If you're merging once and never again, you can type the filename yourself in Bubba Save Checkpoint and skip this node entirely. It earns its keep the moment you're iterating on ratios or producing merges for other people.

Install

Same pack, same install:

cd ComfyUI/custom_nodes
git clone https://github.com/bubbafett5611/Bubba_Nodes

or ComfyUI Manager → search "Bubba Nodes", restart. ComfyUI 0.27.0+, no extra dependencies.

The one trap is forgetting the folder: the prefix is relative, so if you set base_name to something with a path in it (merges/new, say), it's honored as a nested folder - which is a feature, not a bug, but it means a stray slash can file your merge somewhere unexpected. Keep the folder in the folder field and the name in base_name, and the paths stay predictable.