SMOOTH BOOSTER v.5 Update!
More tweaks based on feedback:
Should have a better effect on Illustrious 2 Checkpoints. Still works well on any XL/Illustrious/NoobaAI Checkpoints!
Should do a better job of maintaing the main colors even on higher weights instead of adding light sources of different colors.
Hopefully less blurry, distortions and oversaturation on higher weights for the most part - just don't over do it by going over 1.5 unless it gives the results you want! X_X
You can probably risk even weight over 1.0 and get good results - That, of course, depends on the Checkpoints and Models used.
The same rule applies: The more Loras you use the lower the Lora weight should be!
Less than 2 Loras: Lora weight 0.7 ~ 1.2
More than 2 Loras: Lora weight 0.3 ~ 0.7
SMOOTH BOOSTER v.4 Update!
More tweaks based on feedback:
Should work well on any XL/Illustrious/NoobaAI Checkpoints!
Improved use on Realistic Checkpoints: v3 would sometimes add some anime characteristics to realistic faces (bigger eyes, weird eyes). It should work better now.
Slightly more realistic textures to anime style: Its not much but texture gains a little more realism on anime/cartoon style images. The results should still be very similar to v3 so you can just use your preferred version ^^.
The more Loras you use the lower the Lora weight should be!
Less than 2 Loras: Lora weight 0.5 ~ 0.8
More than 2 Loras: Lora weight 0.2 ~ 0.5
SMOOTH DETAIL BOOSTER!
A detail tweaker/booster inspired by the work of many awesome creators like Shed_The_Skin's Detail Slider LoRA, w4r10ck's Detail Tweaker XL, and many, MANY, others.
This gives an overhaul improvement to color, lighting, shadows and adds more details.
This Lora was made to be used on with Smooth Mix and Smooth Embeddings, but of course, they should work well with any other Checkpoints/Embeddings/Loras. Just try out your favorites and see how it turn out!
I recommend using:
Strength: 0.25~0.8
CFG Scale: 3~5
Adetailer
Hires. fix: Remacri
Have fun!
Description
FAQ
Comments (15)
V4 is wonderful! V3 was great as well, but depending on strength it sometimes mushed the fingers together. V4 is much more forgiving in that respect.
I really like your new version, SMOOTH BOOSTER v4 – it’s excellent!
I’m using it at 0.4 and it works amazingly well. Just wanted to say thank you! ♥
In which directory of comfyui is this file placed?
ComfyUI\models\loras
феникс
v4 is really good. I'm still figuring out how to use it, but I like it a lot so far.
I used v3 extensively, and I had figured out how to tune my settings such that I could apply v3 three to four separate times to an image at full strength (lowres->hires.fix->img2img polishing pass (optional) -> sd upscale). During the final SD upscale, I would actually throw in another detail lora (at 0.5 strength) as well.
And I guess the question is: what does this accomplish?
Most illustrious images here on civitai have bokeh'd/blurry fairly unremarkable backgrounds, but I could get stuff that approached Flux-levels of detail and clarity. And the repeated application of the lora gives subjects/characters a very slight almost 3d quality (it's really interesting). I experimented with other detail loras, but they all warped images or were ultimately just subpar in some way. v3 was the only one that I could stack like this (and apply at such a high strength).
You can't push v4 quite as hard, and you can't pair it with other detail loras quite as easily (the image will warp); I've noticed during my final upscale that it will warp noses and mouths, so I often have to take the resulting image into photopea and clean things up manually. Regardless, v4 produces really beautiful and detailed images.
I think v4 is probably better for people that gen onsite because the images that it produces from lowres and hires.fix are definitely superior to what v3 can do. However, if you gen locally, and you do multiple upscales, things even up quite a bit.
V4 is a fantastic step forward. @DigitalPastel continues to amaze me.
OP LoRa
This model is a standout right from the start. It has a really distinctive feel that makes every output pop with personality, and it’s clear the creator put a lot of thought into shaping its vibe. What I love most is how consistent it is—no matter what direction you take it, the results carry that same spark that makes you want to keep experimenting.
It’s rare to find something that feels this polished and reliable on the first go, but this one nails it. There’s a balance of creativity and stability here that makes it easy to trust, while still inspiring new ideas every time you use it.
Overall, it’s an awesome release and definitely worth diving into if you’re looking for something fresh, expressive, and dependable.
Does this work with V-Pred models?
Will there be an updated Pony version? Or is the V4/V3 version compatible with it?
Yea is the v4 compatible with pony ?
Sorry for the late response. It should have an effect on Pony. Try using at weight 0.5 at first then see if you want more or less weight. :)