This DORA is designed to recreate the style of Boys Factory by Ogawa Hiroshi. Read the usage notes for each version for details. HiRes Fix is recommended for the best results.
All sample images were created directly out of ReForge UI with a Hires Fix pass at 1.5x Latent (Bicubic) upscale with ~20 steps and 0.57-0.62 denoise. The NoobAI version may look best with a higher denoise.
NSFW NOTE: This dataset contains some NSFW images. If you do not want NSFW make sure you actively prompt away from it.
Now that there is support for DORAs on most generation platforms, all future releases will be in the DORA format. These can be used exactly the same way as a LORA.
Please support the original artist wherever possible so they can continue to produce their art. Without them, LORAs such as these would not be possible.
This model is also backed up on Huggingface and on CivitAI: https://civitai.com/models/487524/ogawa-hiroshi-boys-factory-style
USAGE NOTES:
For both versions, it is recommended to use full 1.0 strength. Both versions have some special tags that you can use:
"2020s" - generates the most recent illustrations by the artist that were released on their Pixiv page. You may want this in the negative for more detailed shading or a more analog look.
PONY:
When using Pony, no trigger words are needed other than "source_anime". This LORA was trained on AutismMix. If you are using a different checkpoint, you may need to use a higher strength to maintain the style.
Recommended positive prompt:
score_9, score_8_up, score_7_up, score_6_up, source_anime
Recommended negative prompt: (you can adjust "hands" and "fingers" as needed)
source_pony, (hands:1.1)
NoobAI:
The NoobAI version was trained on NAI Epsilon-pred 1.1-Version, please use this checkpoint for best results. It should also be compatible with V-Pred versions at a much lower weight, but this has not been tested by me. When using the NoobAI version, you should use the trigger word "boysfactory-style".
Recommended postive prompt:
masterpiece, best quality, boysfactory-style
Recommended negative prompt: (you can add "hands" and "fingers" as needed)
worst quality, bad quality, manga, screenshot, sketch, jpeg artifacts
TAGGING NOTES:
The dataset contained over 180 images that were cleaned up (where possible) by hand.
Score_9/"masterpiece" and score_8/"best quality" are the highest resolution works by the artist with the best shading and most details. These are mostly from the artists original period of activity.
Score_7/"good quality" and score_6/"average quality" contain lower-resolution works and the newer Pixiv sketches that have more digital shading.
Please do not violate the content policy