I taught myself all my computer skills. That doesn't make me a professional, but my approach might work for others who would like to give it a try.

Since I don't have the patience to get programs like Forge WebUI or Automatic 1111 running, I use Invoke AI for my images. The program offers a Windows installer and keeps itself up-to-date.

Unfortunately, all training options have been removed from the program, and I can't get the separately available "Invoke Training" to work. So I looked for an alternative, but the only viable program remains ComfyUI. It's not exactly beginner-friendly, but it also has a Windows installer and keeps itself up-to-date.

Thank goodness I found this workflow by "0l1v1aR0551" on Civitai that's as ingenious as it is simple. Just copy the file to the folder "ComfyUI/user/default/workflows/" and load it into the program. Then you can select checkpoints and LoRAs from the dropdown menus and finally click "Run". The program does the rest automatically.

When dealing with LoRAs and checkpoints, always check whether you are allowed to share the resulting merge with others or only use it for yourself (to create images via Invoke Ai).

Two of my favorite checkpoints currently allow sharing merges: Impossible Pie and Semi-Real by mommymia. Many thanks to "mommymia" for these great checkpoints and for allowing me to share the merges!

Since I wanted to try to make the merge a little more in the direction of realistic checkpoints, I mixed the two mentioned checkpoints with PhotoPedia XL by "C3RBERUS" and Hyphoria Real by "ecaj".

To execute the workflow, I now needed two LoRAs and opted for SDXL Detail Enhanced/Simplified by "ritanosonoac554" and Detail Tweaker XL by "w4r10ck".

As a result, I obtained a checkpoint that is very close to photorealism, yet still retains a touch of semi-realism. The checkpoint's style favors close-ups.