as title says, this is designed to mess up the image in the pursuit of realism. It's intended tol break your whitebalance, add grain, and blur and other organic things Ai avoids usually.
Help it along with some "bad quality, jpg, grain" type prompts.
(depth of field) in the negative helps a LOT for the look
Updated with a v02 that works a bit better. Can even function with minimal prompt help now.
Description
I actually cant remember what extra I did, but I know this is trained on a bunch of extra crappy pics. Works better than v01.
FAQ
Comments (17)
Nice! Love it! Just the LORA I needed for a project.
My new favourite Lora. Very well done and thanks a lot!
Wow, this is phenomenal! I am very new to all of this but I have it setup and working, may I please have an example prompt and sampler settings to steer me in the right direction? :)
Thank you and great job!
You can tap the little i icon on the bottom right of the example images to see the prompt. Or click through all the way to the image to see the prompt on the image page. Have fun
@I_dont_want_karma_ That's very good to know, thank you so much. :)
Exactly what i needed Thanks a lot!
Why is this LoRa not appearing on the lora section ? (automatic1111)
This is my first comment (like to lurk haha) but I had the same issue, This should fix it for you, it did for me!
https://www.reddit.com/r/StableDiffusion/comments/10s20eg/comment/jwj51lb/?utm_source=share&utm_medium=web3x&utm_name=web3xcss&utm_term=1&utm_content=share_button
is there a 1.5 version
no. I tried making one but 1.5 just isnt detailed enough to catch the finer noise details and such
I use a combination of depth of field slider at -2 to -3 and old siemens at 0.2, gives a nice low quality look on SD1.5. It works really well on the checkpoint called "serenity"
I often get confused if this is a negative or not, as the popular "bad hands" etc naming convensions go. In case anyone's also confused, put it in the positive prompt with a positive weight
This is my favourite LORA by far. I hope it gets continued development.
Still performs really well! Are you planning on updating the model?
Who?
Model is fixing all my problems wow. any plans for updates? :)
Any chance of getting this in Z Image Turbo or Krea 2? This is one of my all time fave Loras, but SDXL is a bit limited compared to newer models.
Details
Files
badquality_v02.safetensors
Mirrors
bad_quality, jpg, grain.safetensors
badQualityLora_v10.safetensors
badquality_v02.safetensors
badquality_v02.safetensors
badquality_v02.safetensors
badqual.safetensors
badquality_v02.safetensors
badquality_v02.safetensors
badquality_v02.safetensors
badquality_v02.safetensors
badquality_v02.safetensors
badquality_v02.safetensors
badquality_v02.safetensors
badquality_v02.safetensors
badquality_v02.safetensors
bad-quality-lora-sdxl.safetensors
badquality_v02.safetensors
badquality_v02.safetensors
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.