Type | |
Stats | 822 |
Reviews | (86) |
Published | Nov 29, 2023 |
Base Model | |
Training | Steps: 1,600 Epochs: 3 |
Usage Tips | Clip Skip: 1 Strength: 1 |
Trigger Words | j3nc0n |
Hash | AutoV2 73E03EE2E2 |
READ BEFORE DOWNLOADING
This LORA was created using very low-res and blurry images.
Therefore the quality and likeness is not as good as it could be.
You can still get some good images out of it but might take a few tries.
Worked best with epicRealism in my testing.
ADetailer kinda loses detail in the face, not sure what's happening there.
Try a strength of 1-1.2 to get better results.
UPDATE: Running it with lcm produces more consistent results.
I used realesr-general-x4v3 to upscale the images.
If anyone has a suggestion for a better workflow to work with low res images, please let me know and i'll update the model.
I have added a wide range of cherrypicked images in the showcase.
Took a few tries to get these, hope you have better luck.
------------------------------------------------------------------------------------------------------
LORA trained on 70 publicly available images of Jennifer Connelly.
Trained on base SD 1.5, should work with most realistic models BUT with varied results.