08.08.2025 - 1.1 version has been released, try it out! It should output even better results as it comes with a newer database and is trained under EPS 1.1 of NoobAI, making it work even better on both Illustrious and NoobAI based models.
Recommended Checkpoint:
WAI-NSFW-ILLUSTRIOUS-SDXL
- Blurry images with this Lora made Locally?
Use Detail Daemon at 0.25/0.30 Strength or DMD2
- Blurry images with on-site generator and this Lora?
Use either DMD2 or SDXL Lightning 2/4 Step Loras.
All images shown on the gallery are generated locally with Forge.
Hello. This is my fourth style Lora that is based on one of my most beloved titles of the 2000s and that title is Total Overdose. This is the kind of a videogame that I absolutely adore and love for what it is because nothing like this one will be made ever again.
This Lora is somewhat of a love letter for the game because I wanted it to happen, at last it will be preserved for generations, hopefully. If you're a lucky 90s gamer, there is a high chance that you've heard of it and played it, if not...
It is available on GOG if you play on PC.
Technical Info on how to use it:
Works best around with these base resolutions:
1024x1024 / 826x1216 / 1216x826 / 1536x1024 / 1024x1536
- Lora Weights: 0.7 to 1. 0.8 Is the most optimal value from my tests, inpainting boosts the quality even more.
- Sampling Method: 31 steps, DPM++ SDE / Euler A
- Scheduler: Karras / DIMM
- CFG Scale: Between 1.5 to 5.
(With release of 1.1, you can crank it up all the way to 7 and above)
Works even better with CFG+ Samplers if you're using Forge like me.
HiRes Fix Upscaler: Latent (Antialiased), but you can use one that you prefer by your own choice. 4x-Ultrasharp hasn't been tested because I took the liking to Latents a lot more and tiling takes time on my GPU.
I recommend to upscale it by 1.2/1.5x scale
For even better image quality, I highly recommend inpainting with Euler A Sampler and DIMM scheduler.
For prompting:
Positive Prompts: totaloverdose, 3d, realistic
Negative Prompts: artifacts, watermark, signature, artist name, extra fingers, missing fingers, fused fingers, malformed fingers, six fingers, deformed fingers, bad anatomy, poorly drawn hands, mutated hands, extra limbs, disfigured, distorted face, incorrect_eye_color, incorrect_hair_color, out of frame, border, frame, english text, logo, text,
This lora also comes with Ramiro Cruz with both of his appearances from Total Overdose and Total Overdose: Chili Con Carnage.
Total Overdose:
1boy, ramirocruz, 31 years old, muscular build, wristband, short spiky hair, facial hair, aged up, purple shirt with skull design, black cargo pants, sneakers, casual attire,
Total Overdose Chili Con Carnage:
1boy, ramirocruz, 31 years old, muscular build, wristband, short spiky hair, facial hair, aged up, red shirt, black cargo pants, sneakers, casual attire
If you like this lora, I'd really would love to see your images, one is that they're an actual support towards me as a very small lora creator.
Second is that they would help me greatly in improving this model further as 1.0 has it's issues and learning more about them will prove very useful to me. I appreciate it and thank you for your time. Have a great day/evening and enjoy generating with it. Thank you.
Total Overdose and Deadline Games will be never forgotten, forever and ever.
Legends never die.
Description
1.1 Dataset (533) - It should output better results from prompting ranging from scenery to characters. I tried it on three checkpoints - Hassaku v1.3 Style A - WAI-NSFW-ILLUSTRIOUS-SDXL 14.0 and WAI-SHUFFLE-NOOB (2.0 V-Pred)
This Lora was trained on NoobAI Epsilon-pred 1.1-Version.
I should point out right away that this was my second attempt at using a custom model for training that is not Illustrious 0.1. I must say that the training under NoobAI blew my expectations compared to IL0.1, which makes me very happy with how it turned out.
I did it blindly using my usual ILXL 0.1 training settings, and the best part is that it works with Illustrious-based models. I believe that's a win for both users who use NoobAI and Illustrious models in general.
From the perspective of this model, it is exactly how I envisioned this Lora to be from the beginning. It matches the game aesthetics even more than its predecessor on the 1.0 ILXL version.
I strongly recommend using weights between 0.7 to 1, 0.7 will make it more compatible with Loras and still keep it's soul, and in case of emergency: use hires fix and inpainting
{
"engine": "kohya",
"unetLR": 0.0005,
"clipSkip": 2,
"loraType": "lora",
"keepTokens": 0,
"networkDim": 32,
"numRepeats": 4,
"resolution": 1024,
"lrScheduler": "cosine_with_restarts",
"minSnrGamma": 5,
"noiseOffset": 0.1,
"targetSteps": 2374,
"enableBucket": true,
"networkAlpha": 16,
"optimizerType": "AdamW8Bit",
"textEncoderLR": 0.00005,
"maxTrainEpochs": 10,
"shuffleCaption": true,
"trainBatchSize": 9,
"flipAugmentation": false,
"lrSchedulerNumCycles": 3
}FAQ
Looks like we don't have an active mirror for this file right now.
CivArchive is a community-maintained index — we catalog mirrors that volunteers upload to HuggingFace, torrents, and other public hosts. Looks like no one has uploaded a copy of this file yet.
Some files do get recovered over time through contributions. If you're looking for this one, feel free to ask in Discord, or help preserve it if you have a copy.
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.