I put together a quick loom showing how I set up a training on Scenario for a new consistent character. In this case I've only used five images. Anyone can try this - generator creation is completely unlocked for free plans. Thread with steps below 👇

I designed this character by blending a few LoRAs and guiding prompting directly on the platform. It's a tutorial for another day that I am happy to get into.

Then I clicked the + sign next to "Models" and clicked through to training.

I renamed my model (make sure it's set to SDXL) and uploaded my five images.

Then I like to adjust the captions. I tend to add a special token so that I can prompt it later to really lock in the character - this can be a really unique name (see some of @emmanuel_2m's example characters) or it can be a word with all the vowels removed. This step is optional, and if you are struggling with it just simply caption your images.

There are a few different settings I use for characters, but these are the most consistent. In this case "x" is the number of images you use to train. The app does 175 per image by default.

This model is still cooking, but I will show some results later!

The results!

@Scenario_gg I'm so hopeful for this. Have been tearing my hair out trying to make consistent characters in midjourney. Going to read the rest of the tutorials on the site then give this a go.

@Scenario_gg It tends to work pretty consistently!

@Scenario_gg I will check it and update