CONVERT (JOV) 🧬

CONVERT (JOV) 🧬 takes any value and converts it to a type you pick. String to int, int to float, float to a vector, a number to a boolean - the README's highlight list calls this functionality "conversion support for all ComfyUI types and some 3rd party types," and it's one of those nodes that quietly removes a whole class of "this node wants an INT and I only have a FLOAT" errors.

One naming note before anything else: CONVERT is the legacy name. In current Jovimetrix the node is VALUE (JOV) 🧬, same emoji, same idea, expanded - the modern version adds per-component X/Y/Z/W inputs, seeds, and raw value injection. If an old workflow references CONVERT (JOV) and it shows as missing, the current pack installs VALUE (JOV) 🧬 and you rewire.

What it does

Two controls:

• 🅰️ - the input value, any type. If it's already the type you want, it passes through unchanged.
• ❓ - the target type: STRING, BOOLEAN, INT, FLOAT, VEC2, VEC3, VEC4.

Output is a single *-typed list. Feed it a float and ask for INT and you get the rounded integer; feed it a string like "42" and ask for INT and it parses; ask for VEC2 and a scalar becomes a two-component vector. This is the plumbing that makes Jovimetrix's type-agnostic ports work together - you can normalize anything to anything mid-graph.

How it works

It runs the value through the pack's EnumConvertType machinery, which knows how to coerce between numeric, string, and vector representations. The interesting edge cases: strings parse numerically, floats truncate to ints, and vectors accept scalars by repeating them across components. There's no inference and no model - this is pure data normalization, instant even on batches.

Installing it

Part of Jovimetrix:

• ComfyUI Manager - search "Jovimetrix", install.
• Manual - git clone https://github.com/Amorano/Jovimetrix.git into custom_nodes/, then pip install -r requirements.txt.

No models. Dependencies: numpy, OpenCV, Pillow, matplotlib, cozy_comfyui. Needs ComfyUI 0.1.3+.

Where it fits

Reach for it the first time you wire a KSampler's steps (an INT) into a node that wants a FLOAT and get a type mismatch - this is the adapter. It also shines when you're driving schedule or batch logic and need everything on one numeric type. The main gotcha is the legacy-name confusion above; the second is that converting a non-numeric string to INT yields 0 rather than an error, which is a silent-ish failure if you don't notice. If you're building new workflows, go straight to VALUE (JOV) 🧬 - it does everything CONVERT did, plus the vector component inputs that make it genuinely one of the pack's most useful utility nodes.