Type | |
Stats | 1,222 4,505 |
Reviews | (149) |
Published | Mar 8, 2024 |
Base Model | |
Training | Steps: 3,000 Epochs: 18 |
Usage Tips | Clip Skip: 1 Strength: 1 |
Trigger Words | dosvga style |
Hash | AutoV2 E7CEA4C700 |
Version 4.1a notes
Hi again,
Worked for a while in a new version for the VGA style, it's almost a complete new model, double the dataset, different train parameters etc.
It's still no near pixel perfect but it's a little bit better in that regard, also this version brings a mayor improvement in flexibility an comparability with more base models (see examples)
Have fun!
Description:
This model is intended for pixel art and trained to emulate the old VGA dos games.
trigger: "dosvga style"
Usage:
I try to make my loras as easy to use as possible, so the examples are generated with simple prompts and no negatives to give a realistic expectation, I find that using BluepencilXL CFG 6-7 40 steps yields good results but you can tune with some tags (like videogames) to improve the results.
DreamshaperXL also works and I'm sure that there is better checkpoints for this but as long as it not a realistic one you should be fine.
Have fun!
dedicated to https://civitai.com/user/Jigen