FunUI Load Modifiers:
FunUILoadModifiers applies enabled FunUI sampling modifiers to a model before generation, allowing installed modifier modules to adjust how sampling behaves. It discovers the available modifiers, checks which ones match the model’s actual capabilities, and arranges compatible modifiers according to their declared ordering requirements. Rather than changing the original model directly, it works on a clone and replaces existing FunUI modifier patches on that clone to avoid stacking repeated installations. It also validates modifier settings and reports which modifiers were applied, skipped, or could not be installed. Guarding is enabled by default to help manage modifier failures, although some kinds of modifications cannot be guarded. The node works with any sampler, making it a preparation step rather than a replacement for sampling itself. Its creative effects depend on the modifier modules installed and enabled in your FunUI setup.
FunUI Load Modifiers Input Parameters:
model
A required MODEL connection containing the model you want to prepare for sampling. The node inspects the model’s traits—its detected capabilities or characteristics—to decide which installed modifiers are compatible, so a modifier is not selected simply because it is installed. It clones this model, removes existing patches belonging to the FunUI modifier system from the clone, and installs the selected modifiers in their resolved order. This preserves the original model connection for other workflow branches while giving the sampling branch its own prepared version. Connect a model loader’s model output here. There is no default model, numerical minimum or maximum, or fixed list of model options defined by this node; compatibility is determined from the supplied model and the installed modifier modules.
settings
An optional FUNUI_SETTINGS connection that supplies settings organized by module ID. Normally, you should connect the output of FunUI Modifier Settings rather than supplying raw text directly. At execution, the value must be a dictionary-like settings object represented as a Python dictionary, or None when omitted; another value type produces an error. These settings influence modifier selection, module-specific values, and any available control settings. The node checks the values for the selected modifiers again before installation, even if they have already passed through a settings-producing node. The default is None, in which case each selected module’s settings are checked without a supplied entry and any applicable module defaults are resolved by its schema. There are no node-wide numerical limits or fixed modifier options for this connection: supported values, ranges, and defaults depend on the installed modules. Settings for module IDs that are no longer loaded are ignored with a warning. If an installed control provider enables full control, validation problems are warned about rather than refused, invalid values may be retained, and modifier guarding is disabled.
FunUI Load Modifiers Output Parameters:
model
A MODEL containing the cloned model after the selected FunUI modifiers have been processed. Connect this output to the model input of your sampler so generation uses the prepared model rather than bypassing the modifiers. Existing FunUI modifier patches on the clone are removed before the new installation pass, so this output reflects the current selection rather than simply accumulating every previous FunUI modifier installation. Incompatible modifiers, modifiers rejected by ordering rules, and modifiers that decide they are off are not counted as applied. The node also attaches tracking information for dropped modifiers, including a copy in the model’s options so that standard model-cloning operations can preserve it. This output is a model prepared for sampling, not a generated image, video, or latent result.
status
A multiline STRING summarizing the preparation result. The first line reports N modifier(s) applied and lists the applied module IDs when there are any; the next line begins with model traits: and lists the detected traits, or none read if none were detected. Additional lines can describe individual modifier installation results, replaced FunUI modifiers, control-related selection changes, incompatible model requirements, ordering rejections, installation failures, and hooks that could not be guarded. When full control is enabled, the status includes a warning about disabled guarding and the risk of ending a run. Use this output to check what was actually configured before sampling. It describes installation-time results rather than guaranteeing that every modifier will complete successfully during generation, and some warnings, such as stale module settings, are written to the log instead.
FunUI Load Modifiers Usage Tips:
- Place the node between your model loader and sampler: connect the loader to
model, then connect this node’smodeloutput to the sampler. Connecting the sampler directly to the loader bypasses the prepared model. - Use FunUI Modifier Settings to supply
settings. In the supplied FunUI pipeline, the same settings output is connected to both FunUILoadModifiers and FunUISampler, keeping their settings source consistent. - Check
statusafter changing your model or enabling a modifier. A modifier that works with one model may require traits that another model does not provide; the status can identify those missing requirements. - Keep the default guarded behavior while testing unfamiliar modifiers or values. If your installation provides a full-control setting, enable it only when you understand the risks: it disables modifier guarding and permits problematic settings to pass validation with warnings.
- Change a small number of modifier settings at a time and compare generations using the same seed and sampling settings. This makes it easier to identify which modifier is responsible for a visible change; this node does not define universal strength or quality settings of its own.
- If you see
replaced N modifier(s) already on this model, treat it as an informational message, not an error. The node replaces existing FunUI modifier patches on its cloned model instead of deliberately stacking duplicate installations. - Read both
statusand the ComfyUI log when troubleshooting. Installation details appear in the status, while warnings about saved settings for unloaded modules are logged separately.
FunUI Load Modifiers Common Errors and Solutions:
FunUI Load Modifiers: settings must be an object keyed by module id, got …
- Explanation: The
settingsconnection supplied a value other than a Python dictionary orNone. For example, JSON text supplied directly is still a string rather than the settings object this node expects. The full message also says,Wire FunUI Modifier Settings into this input. - Solution: Connect the
FUNUI_SETTINGSoutput of FunUI Modifier Settings. If you are building an API workflow or another settings-producing node, ensure it supplies a dictionary organized by module ID rather than an unprocessed JSON string.
saved settings for '…' ignored: no module with that id is loaded
- Explanation: Saved settings refer to a module ID that is not present in the current registry and is not a reserved control entry. This is a logged warning, and those settings are ignored.
- Solution: Check whether the corresponding module is installed and loaded. If you no longer need it, remove its obsolete settings; if you do need it, restore the module and check the log for any loading problems.
Modifier settings validation failure
- Explanation: The selected modifier’s settings do not meet its schema requirements. The exact message depends on the module and validation helper; no single fixed validation error string is shown in the supplied code. Validation occurs here even when settings came from another producer, helping prevent invalid values such as NaN from silently making a modifier ineffective.
- Solution: Read the reported validation details and correct the affected value in FunUI Modifier Settings. Use values supported by that modifier’s documented type and range, and restore its defaults if necessary. Do not enable full control merely to bypass a validation problem you have not resolved.
…: failed to install -- ExceptionType: …
- Explanation: A modifier raised an exception while installing itself. The status names the module and includes the exception type and message; the failure is also logged and reported to the control system. The node catches this installation exception and continues processing the remaining modifiers.
- Solution: Identify the modifier named in the status, reset its settings, and try again. If it still fails, disable that modifier and review the log for the underlying exception. Check that the module and its dependencies are correctly installed before re-enabling it.
…: needs …
- Explanation: The modifier requires model traits that were not detected on the supplied model. It is excluded from installation because it is incompatible, rather than being applied despite those missing requirements.
- Solution: Compare the listed requirements with the
model traits:line. Use a model supported by that modifier, or select modifiers compatible with your current model. Changing a strength value will not supply a missing model capability.
Modifier rejected by ordering rules
- Explanation: Compatible modifiers are checked for their declared ordering requirements before installation. A rejected modifier appears in the status as its module ID followed by the reason supplied by the ordering system; the exact wording depends on that reason.
- Solution: Follow the reported reason and review the enabled modifier combination. Disable the conflicting modifier or adjust the selection to satisfy the stated requirements, then check the status again. Changing the visual placement of nodes does not replace the modifier system’s internal ordering rules.
…: … could not be guarded
- Explanation: A modifier installed a hook whose failure cannot be handled by the guarding system because there is no known neutral result for that hook type. This is a warning, not necessarily an installation failure, but a later failure inside that hook can still end generation.
- Solution: Test the modifier with supported settings before relying on it for a longer run. If generation fails, disable the named modifier and inspect the log. Leaving guarding enabled is still useful for other hooks, but it cannot protect this particular hook.
full control is ON: modifiers are not guarded, and a failing one will end the run. Any consequences of these settings are yours.
- Explanation: An installed control provider has enabled full control. Modifier guarding is disabled, and settings validation uses warning behavior rather than refusing problematic values. Failures during modifier execution can therefore stop sampling; installation exceptions are still caught separately by this node.
- Solution: Turn off full control through the settings mechanism provided by your installed control module unless you intentionally need it. Correct validation warnings and test modifiers individually before running a complex combination.
0 modifier(s) applied
- Explanation: This status is not necessarily an error. No modifier may be enabled, compatible, accepted by ordering rules, or successfully installed. A modifier can also decide it is off by returning no installation note.
- Solution: Check the remaining status lines for missing traits, selection changes, ordering rejections, or installation failures. Confirm that the desired modifier module is loaded and enabled in your settings. If you intended to sample without modifiers, no correction is needed.
