Visualize Sigmas KJ(VisualizeSigmasKJ)

The Visualize Sigmas KJ node receives a sequence of sigmas and allows the user to highlight a specific range within the sequence by specifying start and end steps.

The Visualize Sigmas KJ node visualizes the noise schedule (sigmas) sequence from the diffusion model sampling process as an intuitive line chart, helping users visually understand the changing trend across sampling steps.

Node Functionality

It receives a sequence of sigmas and allows the user to highlight a specific range within the sequence by specifying start and end steps. Internally, the node uses the Matplotlib library to generate a chart with a dark gray background, clearly displaying the curve of noise level decreasing with sampling steps. This is very useful for debugging and observing sampling behavior. The node outputs the original SIGMAS tensor unchanged, along with an IMAGE tensor of the generated chart.

Usage Scenarios

When you use a custom sampler or adjust scheduling parameters and want to verify if the actual shape of the sigmas sequence meets expectations, you can use this node for visual inspection. When building complex workflows, if you need to isolate and analyze a specific sampling phase (e.g., the latter half of denoising), you can highlight that range by setting start_step and end_step. Additionally, the generated chart image can be saved as part of workflow documentation or experimental records.

Notes

Please note that start_step and end_step are integer step indices (ranging from -1 to 1000). When a parameter is set to -1, it indicates that the range marking feature is disabled for that parameter.

Visualize Sigmas KJ Node Source Code Link

The Visualize Sigmas KJ node is from the ComfyUI-KJNodes node package.