Fate/Zero | Style | IronCatLora

🔥 Recreate the iconic Fate/Zero style — subtle but powerful scene “re-grading.”

Making LoRA's is a thing I do for fun. So if you like what I do? You can help me out by just using the LoRA and leaving a review! 💖

📺 This LoRA has been exclusively trained on anime screencaps from Fate/Zero.

This LoRA adds from my testing "subtle" changes to your generations. Focussing on e.g. adding the cinematic palette, depth-of-field, lineart, skin and fabric shading et cetera from the anime. From my own testing it should be obvious when you use this LoRA with a weight of 0.75 with the trigger word icfzstyle, that the image should "re-grade" it into the Fate/Zero feel. While mostly not completely overwriting the scene.

It took some time playing with the captioning, image dataset, learning rates et cetera. But I feel in its current state, it can nicely add the Fate/Zero look I was looking for!

From my testing on a few checkpoints, I found nice results so far on WAI-NSFW-illustrious-SDXL& NoobAI-XL, it gives me that "Fate/Zero" feel when I generate images on these checkpoint with this LoRA. All sample images I've provided are generated with these checkpoints. Each image also comes with a compare image after the normal sample images, showing the LoRA enabled and dissabled.

👉 The LoRA works with other checkpoints as well, of course. If you want an even more "anime-look", you can try add the keyword anime screencap. I didn't yet include any examples with this keyword, but it should make you generations even more anime-like.

All WAI-NSFW-illustrious-SDXL sample images used the following setting:

• Hires. fix,

  • Upscaler: R-ESRGAN 4x+ Anime6B

  • Steps: 12

  • Denoising: 0.4

  • Scale factor: 1.5

All NoobAI-XL _(_v-prediction model) sample images used the following setting:

• Also use the following LoRA's

  • Add detail XL: /model/122359/detail-tweaker-xl?modelVersionId=135867

  • NAI vpred fix: /model/99619/control-lora-collection?modelVersionId=1965505

• Hires. fix,

  • Upscaler: R-ESRGAN 4x+ Anime6B

  • Steps: 16

  • Denoising: 0.35/0.4

  • Scale factor: 1.5