Home Creators Posts Import Register Favorites Logout

Downloads

Content

In order to test the integrity between LORA and LyCORIS I created both versions time ago, time to share it now. So this post it's for the LyCORIS version, check the other post for LORA version, different trigger keywords, different datasets.

By the tests, LyCORIS could be more flexible but easily breakable regarding likeness combined with different quality tags against a LORA version, it really depends on the checkpoint used.

Trigger Keyword: c7n6y3r-smf

Prompt example to use it:

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

Prompt:

4k, high-res, masterpiece, best quality, sharp focus, (cinematic lighting),  soft lighting, perfect body,  (detailed skin:1.2), <lora:cindycrawford_lycoris_smf_v1-000004:0.9>, 1girl, c7n6y3r-smf, solo, realistic, brown hair, long hair, looking at viewer, brown eyes, lips, parted lips, teeth, upper body, mole, portrait

Negative prompt:

NG_DeepNagetive_V1_75T,(greyscale:1.2), paintings, sketches, (worst quality:0.5), (low quality:0.5), (normal quality:0.5), lowres, normal quality, ((monochrome)), ((grayscale)), deformed body, disproportionate body, 2girl, 2girls, 2 girls body, dirty body, body paint, body marks

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

Sharing Epoch:

4 [ cindycrawford_lycoris_smf_v1-000004.safetensors ]

Recommendations and Specifications:

Sampling Steps: 25

GF SCALE : 7

Checkpoint CyberRealistic

As a LyCORIS it should not need ADetailer but it can also be helpful on some corrections on low res (512x768) full body or mid body renders.

[ https://github.com/Bing-su/adetailer.git ]

Images rendered using the next checkpoint(s):

Realistic Vision V5.1 - V5.1 (VAE)

CyberRealistic

Files

Comments

No comments found for this post.