facedetailer Sampler does not work when connecting the image pin through the load image node even workaround with encode vae does not work.

[zappazack]

the facedetailer sampler works fine when the image is transferred from vae decode node 8 sample steps.

But if the image is transfered from the load imgae node following log. So its not possible to refine faces from already generated images!!

Log:
got prompt
Requested to load ZImageTEModel_
Model ZImageTEModel_ prepared for dynamic VRAM loading. 7671MB Staged. 0 patches attached. Force pre-loaded 145 weights: 383 KB.
Model ZImageTEModel_ prepared for dynamic VRAM loading. 7671MB Staged. 0 patches attached. Force pre-loaded 145 weights: 383 KB.
Model ZImageTEModel_ prepared for dynamic VRAM loading. 7671MB Staged. 0 patches attached. Force pre-loaded 145 weights: 383 KB.
Model ZImageTEModel_ prepared for dynamic VRAM loading. 7671MB Staged. 0 patches attached. Force pre-loaded 145 weights: 383 KB.

0: 640x416 1 face, 29.2ms
Speed: 1.3ms preprocess, 29.2ms inference, 3.5ms postprocess per image at shape (1, 3, 640, 416)

0: 640x416 2 hands, 45.8ms
Speed: 2.0ms preprocess, 45.8ms inference, 4.4ms postprocess per image at shape (1, 3, 640, 416)
Prompt executed in 3.05 seconds
got prompt
Prompt executed in 0.10 seconds

Please advice!

[Buumcode]

So its not possible to refine faces from already generated images!!

Refining faces with Z-Image-Turbo in already exists images works well.

Try this workflow:
Z-IMAGE-TURBO_ALONE_DETAILER__LOAD-FROM-OUTPUT__004.zip

For Krea2:
KREA2_ALONE_DETAILER__LOAD_FROM_OUTPUT_001.zip