My method is a rather simple one using stable diffusion automatic1111 webUI.

step 01: Generate image with 2 (or more) characters while describing ONLY one character in your prompt

In the example images, I used the first character lora: https://civitai.com/models/1686849/dandara-dandy-sombraes-emi-erokuro-series?modelVersionId=1909176

this will usually give me an image with one character resembling the lora I used and another random character with a mixture of similar characteristics.

step 02: Take the image to img2img and use Inpaint or Inpaint Sketch for the second character while only describing the second character in your prompt

In the example images, I now used the second character lora: https://civitai.com/models/1730116/emi-erokuro-and-dandara-dandy-sombraes-metalchromex-mascots?modelVersionId=1958047

And that's it...

I use Inpaiting to make other changes in my images like for small details and such, but it is quite handy in making different characters in the same image as you can see in the examples.