The 'helloArt Oil' model does not want to work well in ComfyUI... Other more modern SD1.5 models work well, but with this one the image is saturated even with low CFG values, however, it works correctly in Forge. The problem escapes me.

The images I have published here use the model but to make them look good I have used CFG 1 and I have not copied the prompt from the examples.

With the Basic Workflow you can generate images using all of these models.

With the Forge Workflow I have included all the parameters taken from the metadata of your example images. It would be necessary to include 'After Detailer', which in this case would be unnecessary, and a couple of parameters that must be further investigated... Emphasis and NGMS.