Nodes/ComfyUI-ZImagePowerNodes/⚡| Z-Sampler Turbo ^G2 (Advanced)
ComfyUI Node

⚡| Z-Sampler Turbo ^G2 (Advanced)

Efficiently denoises the latent image using a process specifically tuned for the "Z-Image Turbo". This node takes a Z-Image Turbo model, an initial latent image, and conditioning parameters, and produces a denoised latent output ready for decoding into the final image. This advanced version includes extra parameters for more precise control and chaining of samplers.

By martin-rizzo·Created 7 months ago·Updated a day ago· 346
⚡| Z-Sampler Turbo ^G2 (Advanced)
  • latent_input
  • model
  • positive
  • divider
  • divider2
  • positive_stg2
  • positive_stg3
  • latent_output
add_noisetrue
seed1
steps8
start_at_step0
end_at_step100
force_final_denoisingtrue
initial_sample_sizefull_size
intensity0.0
intensity_bias0.0
turbo_creativityoff
Category⚡Z-Image

Inputs (17)

NameTypeDefaultDescription
latent_inputLATENTThe initial latent image to be denoised; usually an 'Empty Latent' for text-to-image tasks or an encoded image for image-to-image processing.
modelMODELThe Z-Image Turbo model used for denoising the latent image.
positiveCONDITIONINGThe main prompt/conditioning used to guide the generation process toward the desired content.
add_noiseBOOLEANtrueDetermines whether to add initial noise to the latent image. Recommended for most cases. Disabling this is useful for sampler chaining when the input latent already contains residual noise from a previous process.
seedINT11–18446744073709550000The seed used for the random noise generator, ensuring the same result is produced with the same value.
stepsINT83–20Number of iterations to perform during the denoising process.
start_at_stepINT00–100The step at which the sampling process should start, allowing for more precise control over the denoising process and enabling sampler chaining.
end_at_stepINT1000–100The step at which the sampling process should end. allowing for more precise control over the denoising process and enabling sampler chaining.
force_final_denoisingBOOLEANtrueDetermines whether to force a full final denoising step, resulting in a output latent with no residual noise. Recommended for most cases. Disabling this is useful when residual noise is required for the next process in a sampler chain.
dividerZIPN_SEPARATOR
initial_sample_sizeCOMBOfull_sizeThe latent image size used for calculating the initial noise for intensity correction. While smaller sizes result in a faster first step, they can lead to a less accurate correction
divider2ZIPN_SEPARATOR
intensityFLOAT0.0-1–1Initial noise amplitude used to enhance contrast and colors. A value of 0.0 is neutral; negative values create more muted images, while positive values increase contrast and saturation. This only takes effect when 'denoise' is set to 1.00
intensity_biasFLOAT0.0-1–1Custom adjustment for the intensity noise bias. Usually kept at 0.0; used to fine-tune 'brightness'. Note that its effect depends heavily on the prompt and image style, so it may not always act as a simple brightness control. Adjust it within the positive or negative range until it seems right to you.
turbo_creativityCOMBOoffBoosts model creativity for more diverse compositions while maintaining the general style. Be aware that this can lead to hallucinations and isn't recommended for inpainting tasks. The refined options add extra steps to try to correct the hallucinations and bring coherence to the image
positive_stg2optCONDITIONINGThis input is optional and can remain disconennect. It allows specifying a different prompt/conditioning for the second stage of the denoising process.
positive_stg3optCONDITIONINGThis input is optional and can remain disconennect. It allows specifying a different prompt/conditioning for the third stage of the denoising process.

Outputs (1)

NameTypeDescription
latent_outputLATENTThe resulting denoised latent image, ready to be decoded by a VAE or passed to another node for further processing.