Feature Request - ControlNet Pose Parameters Node

[ronan36880]

First of all, apologies if this issue should instead be moved over to the https://github.com/Acly/krita-ai-diffusion/ repo.

I noticed that commit https://github.com/Acly/comfyui-tooling-nodes/commit/9a9cbe78a5851a49da0b38bc9b17504837476e8f, which introduced the KritaStyle node, has the message High level krita parameters (style, control-net, ip-adapter), but it only introduced style settings.

Would it be possible to add a node similar to Krita Image Layer, where the user could pick vector layer in Krita and the custom workflow receives the a pose vector ready to input into ControlNet?

[Acly]

That should already work (with Krita Image Layer). ControLNet expects a pixel image with pose vectors rendered on black background.

I wanted to implement the high level UI used by the plugin for control layers too, as a more convenient way to select control-net model and parameters. I ended up removing the code again because it wasn't possible to detect which model architecture (SD1.5, SDXL, Flux...) to select at workflow generation time.

[ronan36880]

That should already work (with Krita Image Layer). ControLNet expects a pixel image with pose vectors rendered on black background.

This does work, but when using the workflow mode, I don't see extra UI elements that let me add a pose skeleton to a vector layer directly from there. I'm guessing that would be part of the high level UI you mentioned?