node                       output       shape                numbers
CLIPTextEncode             CONDITIONING [1, 77, 768]         59,136
EmptyLatentImage           LATENT       [1, 4, 64, 64]       16,384
KSampler                   LATENT       [1, 4, 64, 64]       16,384
VAEDecode                  IMAGE        [1, 512, 512, 3]     786,432
the sampler works on 16,384 numbers, the picture has 786,432: 48x more

sigma_max 14.6146, sigma_min 0.0292
20-step normal schedule: 14.6146 10.7468 8.0815 6.2049 4.8557 3.8654 3.1238 2.5572 2.1157 1.7648 1.4806 1.2458 1.0481 0.8784 0.7297 0.5964 0.4736 0.3555 0.2322 0.0292 0.0000
same as the 20-step list in the documentation to four decimals: True

denoise   steps run   list built with   first sigma
1.0       20          20 steps          14.6146
0.75      20          26 steps          4.2650
0.5       20          40 steps          1.5482
0.25      20          80 steps          0.6708

step  sigma     x to final   denoised estimate to final
0     14.6146   38.975       2.065
2     8.0815    20.889       1.891
4     4.8557    12.021       1.651
6     3.1238    7.327        1.364
8     2.1157    4.657        1.071
12    1.0481    1.948        0.627
16    0.4736    0.642        0.240
19    0.0292    0.004        0.000

network evaluations for 20 steps: Euler 20, Heun 39
