RealCosplay 1536px

CHECKPOINT
Original


Updated:

RealCosplay is a multi-style model, not limited to realism alone.

If you like this model, you can support my work on Unifans.Or give my model a ❤️

Unifans: https://app.unifans.io/c/sennke (sample workflow, more sample image)

X: https://x.com/SennkeEknn

------------------------------------------------------------------------------------------------

1536 vs 1024 【The default style is still anime style】

✅Better small face quality 更好的小脸质量

✅Better skin detail 更好的皮肤细节

✅Better understanding of hand movements 更好的手部理解

⚠️Style diversity is slightly weaker at high resolution 高分辨率时风格化较弱

⚠️30+ steps is better, but the speed will become slower 需要更高步数

Overall, 1536px version is my favorite. I may take a break from it for a while.❤️

------------------------------------------------------------------------------------------------

Recommended settings:

  • CFG: 3~3.5

  • Steps: 28 ~ 40

  • Sampling Method: [Euler a✅] and [DPM++ 2S a✅] have been verified (⚠️Others are likely to generate color noise or strange patterns. They may not support illust2.0 1536px model, such as DPM++ 2M⚠️, Euler⚠️, etc.)

  • Resolution: Train on 1536px. supports 512~1536 resolutions . In my test, 1024px has better logic and adherence. 1536px has finer texture details.You can generate at 1024px first, then Hires.fix to 1536px. Total pixel count within 2.35 million is acceptable, for example, the resolution of 1824x1248 also very stable.

  • postive prompts:

    photograph,cosplay,illustration,

  • negative prompts for photograph

    illustration,anime,sketch,comic,cartoon,toon,oil painting \(medium\),flat color,outline,3D,2.5D,2D,unreal,unity,blender,anime,smooth skin,semi-anthro,impasto,

------------------------------------------------------------------------------------------------

Prompt Guidelines(trained trigger)

  • Real style:photograph,cosplay,

  • Anime style:illustration,anime,

  • Ethnicity: european, asian,

  • tip:The prompt "realistic" will generate semi-realistic images, not photo-realistic images.

    -----------------------------------------------------------------------------------------------

Safety Guidelines(same as IllustXL 2.0):

  • Prompt:***** content, r-18, nsfw, explicit, sfw, safe for work

Version Detail

Illustrious
90780
51
Full finetuning from Illustrious XL 2.0 trained on 7000+ images and Tag+NLP mixed caption

Project Permissions

Reprinting is strictly prohibited

    Use Permissions

  • Use in TENSOR Online

  • As a online training base model on TENSOR

  • Use without crediting me

  • Share merges of this model

  • Use different permissions on merges

    Commercial Use

  • Sell generated contents

  • Use on generation services

  • Sell this model or merges

Related Posts

Describe the image you want to generate, then press Enter to send.