ComfyUI-Detail-Daemon

DetailDaemonSamplerNode

This sampler wrapper works by adjusting the sigma passed to the model, while the rest of sampling stays the same.

Sampling/Custom Sampling/Samplers · entrée 1 · paramètres 10 · sortie 1

LyingSigmaSampler

A simpler version of Detail Daemon Sampler, with only amount adjustment and start and end values. Parameters: - dishonestyfactor: similar to detailamount in the Detail Daemon node, this adjusts the amount of detail. It is on a different scale though, for example, -0.05 dishonestyfactor is the equivalent of 0.5 in detailamount of Detail Daemon (or 0.95 of Multiply Sigmas). Negative values adjust the sigmas down, increasing detail. You probably want to stay between -0.1 and -0.01. Positive values would increase the sigmas, decreasing detail. - startpercent: when do you want the adjustment to start, in a percent range from 0–1.0, 0 being the first step, 1.0 being the last step. Recommended: 0.1–0.5 - endpercent: when do you want the adjustment to end, in a percent range from 0–1.0, 0 being the first step, 1.0 being the last step. Recommended: 0.5–0.9

MultiplySigmas

Simple node to multiply all sigmas (noise levels) by the supplied factor. It multiplies both the noise levels added and denoised by the factor, which somehow adds detail with a factor less than 1. It is stateless, meaning it calculates the sigmas fresh on every queue (other multiply sigmas nodes seem to calculate on prior run sigmas). Parameters: - factor: the amount that you want to multiply the sigma (noise level) by at each step. So, for example, if the first step has a sigma of 1, then using a factor of 0.95 would make this sigma 0.95. If a step has a sigma of 0.7, then a factor of 0.95 would make it 0.665. You probably want to keep this factor between 0.95–0.99. Lower values increase detail, but might also increasingly change the composition of the image, or introduce noisy grain. Setting it to 1.0 effectively disables the node. - start: when do you want the adjustment to start, in a percent range from 0–1.0, 0 being the first step, 1.0 being the last step. - end: when do you want the adjustment to end, in a percent range from 0–1.0, 0 being the first step, 1.0 being the last step.

Sampling/Custom Sampling/Sigmas · entrée 1 · paramètres 3 · sortie 1