Comfy-Org/ComfyUI · error · ValueError

Cannot convert non-finite value to number: {float_val}

Error message

Cannot convert non-finite value to number: {float_val}

What it means

Raised inside the string branch of the number-conversion node when float(text) succeeds but the result is NaN or ±Infinity (math.isfinite fails). This catches strings like 'nan', 'inf', '-infinity' which parse but cannot represent a usable workflow number.

Source

Thrown at comfy_extras/nodes_number_convert.py:64

            int_val = 1 if value else 0
        elif isinstance(value, int):
            float_val = float(value)
            int_val = value
        elif isinstance(value, float):
            float_val = value
            int_val = int(value)
        elif isinstance(value, str):
            text = value.strip()
            if not text:
                raise ValueError("Cannot convert empty string to number.")
            try:
                float_val = float(text)
            except ValueError:
                raise ValueError(
                    f"Cannot convert string to number: {value!r}"
                ) from None
            if not math.isfinite(float_val):
                raise ValueError(
                    f"Cannot convert non-finite value to number: {float_val}"
                )
            try:
                int_val = int(text)
            except ValueError:
                int_val = int(float_val)
        else:
            raise TypeError(
                f"Unsupported input type: {type(value).__name__}"
            )

        if not math.isfinite(float_val):
            raise ValueError(
                f"Cannot convert non-finite value to number: {float_val}"
            )

        return io.NodeOutput(float_val, int_val)

View on GitHub (pinned to 1c6d8d45b3)

Solutions

1. Replace the value with a finite number (e.g. a large but finite bound like 1e6 instead of 'inf').
2. Sanitize upstream computations that produce NaN so they emit a real default.
3. Use the node's plain number inputs instead of strings when you need large values.

Example fix

// before
NumberConvert(value="inf")
// after
NumberConvert(value="1000000")

import math
text = text.strip()
try:
    v = float(text)
except ValueError:
    raise ValueError(f"not a number: {text!r}")
if not math.isfinite(v):
    raise ValueError(f"non-finite value not allowed: {text!r}")

def is_finite_number_string(text: str) -> bool:
    try:
        return math.isfinite(float(text.strip()))
    except ValueError:
        return False

When it happens

Trigger: Input strings 'nan', 'NaN', 'inf', '-inf', '+Infinity', or float-overflowing scientific notation such as '1e999' (parses to inf). Only in the string branch, immediately after float_val is computed.

Common situations: Upstream math or data pipelines emitting NaN/Inf as text; placeholder sentinel values like 'inf' used for 'no limit'; hand-typed scientific notation that overflows float64.

Related errors

• Keyframe times must be finite numbers in seconds; got '{valu
• Cannot convert empty string to number.
• Unsupported input type: {type(value).__name__}
• Math Expression '{expression}' produced a non-finite result:
• Cannot convert string to number: {value!r}

AI-assisted analysis of Comfy-Org/ComfyUI@1c6d8d45b3 (2026-08-14). Data as JSON: /api/errors/bb7c702f039ddae5. Report an issue: GitHub.