Hey There Create your own sliders at https://sliders.ntcai.xyz
Change your generations into a comic book illustration . Examples at 4-5 strength
Trained for 6000 instances with sliders https://github.com/rohitgandikota/sliders
batch size 12, 500 steps, 1.5 hr on 1x a6000
Trained on https://civarchive.com/models/215796/top-rated-turboxllcm-super-mario-merge
Support my work on patreon and get access to over 400 unseen loras with many new ones daily. Plus, easily train your own, no hardware needed (coming soon)!
https://patreon.com/NTCAI Any proceeds go to funding my compute addiction.
Description
FAQ
Comments (9)
How did you convert to Safetensor? I'm just starting to get into making sliders
I have a fork that adds safetensors here: https://github.com/ntc-ai/sliders-conceptmod
@ntc WAT! Awesome, thank you!
I've been trying to get this to work, but it fails to run even the default training prompt,
raise ScannerError("while scanning a double-quoted scalar", start_mark,
yaml.scanner.ScannerError: while scanning a double-quoted scalar
in "trainscripts/textsliders/data/config-xl.yaml", line 3, column 17
found unknown escape character 's'
in "trainscripts/textsliders/data/config-xl.yaml", line 3, column 21
PS C:\sliders-conceptmod>
Is there a custom script I'm overlooking?
@TigonTX Oops, I modified the flow to use it as a library and I probably broke the intended path.
I can't look at debugging it right now because I'm running behind on various tasks, rohit's repo works well and you can pull the specific commits if you need any of my changes.
Again, sorry for the bug! Edit: his repo is here: https://github.com/rohitgandikota/sliders
Could you perhaps share the settings for one of your example images? I have tried dozens of combinations of settings in A1111 and cannot come close to getting anything like them. Regardless of the Lora strength or VAE, the results I get are harsh and while somewhat comic-like, they're not soft and detailed like yours. Thank you very much.
hi sorry about the issues, they are comfyui images.
https://civitai.com/images/6353336
This has very similar settings
I switched to a111 recently so it shares the generation info on my latest models. They are the same.
@ntc Ah, thank you very much, I will try it out.
I used his comfyui images and there are no tags inside wich describe the pictures, even it's not img2img.
I mean with tags like "________bg________, comic book illustration" as the only positive tag, you don't get any picture showing a woman in a city, it would be very buggy and bad looking. Also the pictures are NOT img2img. Your pictures also have cfg scale 1 and steps 12 only, the LORA here is set to 5, but even set to 10 it does absolutely nothing.
Oh, btw the tag "comic book illustration" gives the same results, with or without your lora.
It seems this is a simple Lora just doing nothing.
Would you please share at least one workflow wich is working please, so we all could finally use your LORA, as it is very "special" to get used to it.
Regards,
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.