Sigmas Nodes
Use this node to analyze the trends for the sigmas with various noise settings. The model input is required.
σ = the initial noise level at each step.
σnext = the target noise level at the end of each step.
σdown = how far down the sampler really goes, when noise addition is being used.
σup = how much noise is added after each step.
σ + σup = the maximum noise level at each step.
Δt̂ = the full step size: how much noise is removed at each step after noise is added (σ + σup - σdown). Ultimately, the effect on the overall sampling process matches delta_t below.
Δt = the step size: how much noise is removed at each step (σ - σnext).
αratio = how much the current latent is scaled by (as a multiplier) at each step prior to noise addition.
