Skip to main content

Indie game storeFree gamesFun gamesHorror games
Game developmentAssetsComics
SalesBundles
Jobs
TagsGame Engines
(+1)

There are 2 versions of the Improved Random Portraits mod; the original by Kyler2 and an updated version by Zepeteus (https://itch.io/t/984434/randomportraits-and-portrait-pack-editor-for-the-10-version). They work on the same fundamentals, but there are some fixes and adjustments in the latter.

Images are tagged with the characteristics of the image using folders and file names. There is no functional difference between the ways of tagging as the mod just looks at the full file path.

There are 2 types of filters (my naming for clarity) used by IRP: Necessary and Booster. The Necessary filters will exclude any image that does not meet a criteria such as race, sex, or age. The Booster filters increase the weight of images that match the criteria such as skin color, hair color, tit size, ass size, or penis size.

If no images pass the Necessary filters, then the slave gets no image so image pack diversity is necessary to get good coverage. If there are still images, then the mod will use a weighted random selection to choose from all of the remaining images. The mod is designed not to give the optimally matching image all the time in order to increase diversity of picked images. So it is entirely possible for the mod to pass up a better match for a poor match.

The downside of the weighted random is that it can be overwhelmed by having too much diversity of images within the scope of a single group selected by the Necessary filters, thus proportionally increasing the chances of poor matches. There is not a good fix for this, but the mod has a settings file where users can increase the weights to bias it more towards better matches. But this will also result in less diversity of images appearing in game.

Tuning the settings can be tricky, but I'd guess that an increase by somewhere between 3 to 5 times the current values should show significant difference in outcomes. You can also adjust the proportions between the weights to emphasize some attributes over others.

Images start with a weight of 1 and weights are increased multiplicatively. So an image matching two attributes with weights of 10 will have a weight of 100, meaning it will be 100 times as likely to be chosen as an image matching no attributes.