I made Marauder, a notoriously difficult enemy you can face in Doom Eternal, LoRA from Doom especially Doom Eternal. I think he's cool enough to be made to LoRA. Marauder is very difficult to made into LoRA because the level of detail of this guy is insanely high and my checkpoint that I usually made for benchmarking or testing model seems to bit struggling. But, I found out that the output is good enough to be released.
I expect some feedback from you to improve this model further in future.
Trigger Word (choose one):
doommarauder, marauder_doomSettings I used:
Weight: 0.75 - 0.85
Sampling Method: DPM++ 2M SDE Karras
Sampling steps: 22 - 25
Resolution: 768x768 or 960x960 (768x960)
You can use HiRes + Upscaler too. I suggest to start from 640, though any resolution will work (don't go too high because of tiling problem).
CFG Scale: 7
Checkpoints
Mature Male Mix --> perform bit worse
Airfuck's Brute Mix
Juggernaut
Realistic Vision --> I recommend this bit closer to the intended model
Suzumehachi (I didn't recommended this alone because it's like action figure-y but with bit off putting without any LoRA that suitable with this model) + VHS3 (The checkpoint suddenly become better choice due to VHS filter just does work very well)
Note:
The trigger word of this model is bit different than used to be and second one (after Cliff from Pokemon) because the word marauder is seems reference to something that already in SD and it's generic name that frequently used in several model or scenes like marauder sword, marauder land, etc.
Description
Here's some info about this model:
Trained on 120+ images that came from various source from Google Images, Art Station, SFM, Gmod model, Pinterest, In-game Screenshots came from DOOM Eternal.
Epochs 0f 15 with 10 repeats. Approx ~9300 steps going in.
Resolution trained on 640
FAQ
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.