Krea2T Enhancer Advanced:
Krea2T-Enhancer-Advanced attaches a text-conditioning enhancement to a compatible Krea2 diffusion model. During model execution, its wrapper can adjust the model’s text-fusion behavior and the scale of its text-processing path, giving you a way to experiment with how text information is handled during generation. The node works on a clone of the supplied model, so it does not directly modify the model object you connected to it. Its enhancement is specific to models that match the Krea2 text-fusion layout checked by the implementation; with a different layout, the wrapper leaves normal model execution in place. The source establishes how the enhancement is attached and controlled, but does not establish that any setting will reliably improve a particular image or prompt.
Krea2T Enhancer Advanced Input Parameters:
model
The MODEL you want to process. Connect the model that will be used for generation, then use this node’s returned model downstream. The node clones this input and attaches its configuration and diffusion-model wrapper to the clone. There is no default, minimum, or maximum for this required connection. The enhancement only takes effect during execution if the model passes the implementation’s Krea2 layout check.
enabled
A BOOLEAN switch that controls whether the attached enhancement runs. The default is True; the available options are True and False. When False, the node still returns a cloned model with its configuration attached, but the wrapper calls the model’s normal execution path without applying the enhancement. This makes it useful for comparing otherwise identical workflows with the enhancement on and off.
strength
A FLOAT controlling the text-fusion enhancement. The default is 1.0, the minimum is 0.0, the maximum is 2.0, and the interface step is 0.05. The value is bounded to that range when the node is applied. At 0.0, the wrapper does not install its enhanced text-fusion forward behavior; values above zero enable that behavior and pass the selected strength to it. The supplied source does not show the enhancement calculation itself, so a precise visual effect or linear relationship between this number and image changes cannot be promised.
text_scale
A FLOAT passed to the model’s text-MLP scaling setup during execution. The default is 1.0, the minimum is 0.25, the maximum is 4.0, and the interface step is 0.05. The value is bounded to that range when the node is applied. 1.0 is the unscaled setting; other values request a different scale for this text-processing path. This control is separate from strength: setting strength to 0.0 does not, by itself, turn off a non-default text_scale. The supplied code does not include the scaling implementation, so its exact visual effect is not specified.
debug
A BOOLEAN that controls diagnostic printing. The default is False; the available options are True and False. When enabled, the node prints an attachment message showing whether it is enabled and the bounded strength and text_scale values. During a compatible execution, it can also print measurements such as progress, sigma, and text-fusion comparison values; for an incompatible model, it can print a skip message. These messages help you check what ran, but do not change the image by themselves.
Krea2T Enhancer Advanced Output Parameters:
model
The sole output is a MODEL: a clone of the connected model carrying this node’s configuration and diffusion-model wrapper. Connect it wherever your workflow expects the model used for generation. This output is not an image, an enhanced prompt, or a report of measured quality. Its attached behavior is evaluated when the model executes: if the enhancement is disabled or the model does not match the checked Krea2 layout, normal diffusion-model execution continues.
Krea2T Enhancer Advanced Usage Tips:
- Connect the returned
modelto the rest of your generation workflow. Connecting the original input model downstream will bypass this node’s attached configuration. - Start with
enabled=True,strength=1.0, andtext_scale=1.0. Change one numeric setting at a time while keeping your prompt and other generation settings fixed, so you can judge what changed. - Use
enabled=Falsefor an on/off comparison without removing the node from the workflow. For a neutral numeric configuration, usestrength=0.0andtext_scale=1.0together;strength=0.0alone does not rule out text scaling. - If you want to examine the two controls separately, first vary
strengthwhile holdingtext_scaleat1.0. Then setstrengthto0.0and varytext_scale. Treat the resulting images as your evidence of usefulness rather than assuming a particular setting always improves them. - Turn on
debugwhen checking compatibility or execution. Look for the attachment message and, if the model runs through the compatible enhancement path, its execution measurements. Turn it off when you no longer need console diagnostics.
Krea2T Enhancer Advanced Common Errors and Solutions:
[Krea2TEnhancerAdvanced] skipped: diffusion model does not match Krea2 text-fusion layout
- Explanation: This is a diagnostic message, not a crash. With
debug=True, it means the supplied model did not pass the wrapper’s Krea2 layout check, so the enhancement was skipped and normal model execution continued. - Solution: Check that the connected model is the intended compatible Krea2 model and that you connected this node’s output downstream. If you need to use a different model, leave the enhancement disabled or remove the node from that workflow.
No enhancement appears when enabled=False
- Explanation: This is expected behavior, not an error message. The wrapper sees the disabled setting and passes execution through without applying the enhancement.
- Solution: Set
enabled=Truewhen you want to test the enhancement. Keep itFalsewhen you want a comparison run.
No enhancement appears with strength=0.0 and text_scale=1.0
- Explanation: This is expected behavior, not an error message. The wrapper returns to normal execution for this combination.
- Solution: Increase
strengthabove0.0to test text-fusion enhancement, or changetext_scalefrom1.0to test the separate scaling control.
[Krea2TEnhancerAdvanced] attached ... appears, but the image does not change
- Explanation: The message confirms that the wrapper was attached to the cloned model; it does not confirm that the connected model is compatible, that the enhancement ran, or that a visible image difference will result.
- Solution: Confirm that the returned
modelis connected downstream,enabledisTrue, and the numeric settings are not both neutral. Enabledebugand check for a compatibility skip message, then compare runs with otherwise identical settings.
Model connection or required-input error
- Explanation: The
modelsocket is required. An absent connection, or a connection that does not supply aMODEL, prevents the node from receiving the model it needs to clone. - Solution: Connect a valid
MODELoutput tomodeland check that the workflow uses this node’smodeloutput afterward. The exact validation message depends on the host and is not specified in the supplied node source.
