lllyasviel / sd-forge-layerdiffuse

[WIP] Layer Diffusion for WebUI (via Forge)
Apache License 2.0
3.9k stars 334 forks source link

no transparency even with sanity check values #91

Open TanvirHafiz opened 8 months ago

TanvirHafiz commented 8 months ago

woman, messy hair, high quality

Negative prompt: bad, ugly

Steps: 20, Sampler: DPM++ 2M SDE Karras, CFG scale: 5, Seed: 12345, Size: 1024x1024, Model hash: 1fe6c7ec54, Model: juggernautXL_version6Rundiffusion, Token merging ratio: 0.2, layerdiffusion_enabled: True, layerdiffusion_method: (SDXL) Only Generate Transparent Image (Attention Injection), layerdiffusion_weight: 1, layerdiffusion_ending_step: 1, layerdiffusion_fg_image: False, layerdiffusion_bg_image: False, layerdiffusion_blend_image: False, layerdiffusion_resize_mode: Crop and Resize, layerdiffusion_fg_additional_prompt: , layerdiffusion_bg_additional_prompt: , layerdiffusion_blend_additional_prompt: , Version: f0.0.12-latest-150-g3cdae096

no transparency whatsoever

image

the girl looks the same but with a slightly different smily. I wonder why, I copies the generation data and using the juggernautXL checkpoint!

protector131090 commented 7 months ago

same here

jingwu2121 commented 6 months ago

Hi, did you manage to solve that? I meet the same issue, wondering if it is because of the lack of some plugins or bad setting? Screenshot from 2024-05-28 16-13-27