SD3Medium Realisitc Photo EK-ArtBooster

This workflow uses the SD3 Medium with the LoRA EK-Art Booster.

https://tensor.art/models/751446481946364197/EK-Art-Booster-SD3-2.0

The human limb is still a problem.

Metadata:

1. Checkpoint SD3 Medium

2. Prompt

   • t5xxl positive prompt: As far as I understand it, t5xxl helps the model better understand relationship between objects and reduces problems such as color bleed.

   • clip_l and clip_g positive prompt: clip_l and clip_g are the CLIPS used in SDXL and acting as the `translator` between text and image. So I think this is the main place where you can place your prompt.

3. Negative prompt (by default): 2d, 3d, unreal, animate, cartoon, anime, cg, manga, bad anatomy, bad hands, three hands, three legs, bad arms, missing legs, missing arms, poorly drawn face, bad face, fused face, cloned face, worst face, three crus, extra crus, fused crus, worst feet, three feet, fused feet, fused thigh, three thigh, fused thigh, extra thigh, worst thigh, missing fingers, extra fingers, ugly fingers, long fingers, extra eyes, 2girl, amputation, disconnected limbs, fused legs,

3. Output Format: 1000 * 1500 (the portrait format) with Hires Fix

4. sampler: dpmpp_2m

5. scheduler: sgm_uniform

6. model sampling shift: 3.0.

7. Steps Basic: number of diffusion steps for the basic image

8. Steps Hires: number of diffusion steps for the Hires Fix

9. Art Booster LoRA weight: LoRA weight for Art Booster