Don't Starve Influence

I am admittedly kind of a noob at this, but if you enjoy my humble work, any donations are greatly appreciated: https://ko-fi.com/insult_to_ninjary

Your generosity helps me stay motivated to keep improving and to keep trying new things!

Trigger Words:

Main trigger: d0ntst4rv3

Helper prompts: parody, style parody

Extra help: chibi

• The LoRA can potentially work with no triggers, but it may not be strong/consistent.

• For a stronger effect, use the main trigger: d0ntst4rv3. You can also weight it numerically or with parentheses - I usually do (((d0ntst4rv3))).

• If the (weighted) main trigger is still not enough by itself, next try the helper prompts along with the main trigger. They can be weighted as well - I usually do them all together, like (((d0ntst4rv3, parody, style parody))).

• If you want a higher chance of getting a small, big-headed character like the in-game characters are, you can also try adding chibi along with the other trigger words. It won't always work, but even when it doesn't totally work, I find it still gets characters a bit closer to the game's style, especially the head and face, so it's worth trying out.

• Please note that the main trigger here is only for the Illustrious model - it won't work for the SD1.5 model. The helper prompts may or may not work for SD1.5 (I never tried them), but they are not necessary as the SD1.5 LoRA works just fine with no triggers!

At a Glance (Illustrious):

• Base model: Something illustrated (I usually use my own Illustrious merge or WAI)

• LoRA weight: 1 (but slowly go lower or higher if necessary)

• Suggested parameters: Euler a, Automatic, 30 steps, 832x1216/1216x832, CFG 5 (but feel free to experiment!)

• Hires. fix: 1.5x, use a quality upscaler (like 4x-UltraSharpV2), denoising strength 0.5 - 0.3

• Extra: use ADetailer to fix faces and/or hands if necessary

Illustrious Version Description:

This LoRA will give your images a vibe similar to the Don't Starve series of video games. What this means is that you'll get kind of a 2D sketch-like style that's sort of creepy-cute and has a general brown/sepia tone. This is more of a general style LoRA than a character style LoRA, so while you may be able to coax out characters that look pretty close to how the in-game characters look, you'll mostly just get a general stylistic influence on characters, and a stronger effect on stuff like backgrounds. I'll go over some of the other quirks below.

First, it can be a little fuzzy and lacking in fine details. Like, if you zoom in or check some fine details closely, they can appear kind of muddy or garbled. I don't think it's too bad though, and if you do need any elements to be a bit more detailed or crisp, you can try inpainting them.

Second, I feel that outdoor shots produce the style much more recognizably than indoor shots. This is because, as you might expect, most of the training data consisted of outdoor shots from the game. Luckily, the LoRA does affect indoor shots pretty strongly as well, but you can tell the model is just guesstimating what a lot of indoor objects should look like, compared to outdoors, where it really knows how to make nature and buildings that emulate the style well.

Finally, it has some trouble keeping track of objects that move behind other objects, if that makes any sense. Like if there is a bed behind the main character, the bed might not line up properly on each side of the character. This seems to be a general Illustrious issue in my experience, but surprisingly, I haven't noticed this issue to be that bad when using this LoRA, compared to some of my other LoRAs or just using some Illustrious base models raw. It'll still pop up here and there to varying extents, but hopefully it shouldn't be too bad.

For more potentially relevant information, you can check out the original SD1.5 description immediately below. Below that, there's a troubleshooting section in case you're having any problems.

SD1.5 Version Description:

A while back I came across a few Don't Starve LoRAs, but they all seemed to be based off of character art. I started wondering what a LoRA would look like if it was based more off of gameplay-style art instead. And thus this LoRA was born!

When generating characters, you'll get some interesting results. You can definitely see an influence from the game's art style creeping into the pictures, but the way it manifests itself is a bit unexpected and surprising, at least for me. Characters won't really look like Don't Starve characters, yet they still retain the style of art that you'd see during gameplay, if that makes any sense. It's kind of hard to explain, but even if your results don't always scream "Don't Starve," you'll still get some cool-looking stuff!

Landscapes generate a bit more as you'd expect, provided that your prompts don't stray too far away from the types of settings you might see in the game. If you prompt for stuff that's pretty far outside of what was in the training images, your pictures won't come out bad, but you'll definitely have trouble holding onto the desired art style.

Here are some other usage tips for the LoRA:

1. I'm guessing it won't work too well with realistic base models, but you can try! Anime/cartoon models should work fine tho.

2. If you're frequently getting a small character, no character, and/or bad anatomy, you can try gradually lowering the weight of the LoRA.

3. Use a VAE, especially if your chosen base model doesn't have one baked in! It really seems to help with getting better colors and anatomy.

4. Upscale or use Hires. fix with a quality upscaler. If Hires. fix seems to be destroying your images, try lowering the denoising strength.

Troubleshooting:

If you're having persistent problems with anatomy, fine detail, and/or other jank, just try your best with all the usual fixes - different prompting (including negative prompt), try a different base model, use a different VAE, try different clip skip values, adjust the LoRA weight, use adetailer, inpaint, etc.

Outro:

If you make anything awesome, please consider posting them to the gallery! That goes for all of my models! I really like seeing what kind of stuff you all are able to make!

If you have any questions, feel free to ask! I will try my best to help! Also, if you have any ideas for future models, please don't hesitate to share them with me!