Here we go!

This is a quick one.

I’ve tweaked the workflow for the "UnNamedA - ANIMA realistic model," which you can find here: https://civarchive.com/models/2697655/unnameda-anima-realistic-model?modelVersionId=3243147

The creator has made a great model.

Note that he recently added a specific BF16 version of the model, explaining that it should now work well in ComfyUI (I gather he usually works with Forge); however, I based my setup on the workflow he shared previously, making the following changes:

- Swapped the standard "Load Checkpoint" for a "Unet Loader" within my Loader Subgraph, along with the corresponding loaders for CLIP and VAE.

- Changed the CLIP Text Encode nodes, as these models require the Lumina-specific ones when used in ComfyUI.

- Removed the "CLIP Set Last Layer" node.

- Confirmed that the shift-setting node shouldn't be used (at least not in ComfyUI).

- Tidied up the node layout a bit—though that’s just a matter of personal preference :-)

I’m not sure exactly which LoRAs work with this model, but I’ve included the loader for them.

I’ve verified that both 2.0 models work in ComfyUI with this workflow.

Have fun! :-)