• GᖇᗩᑭᕼIᑕ ᑭOᖇTᖇᗩIT •
This template creates artistic portraits with an eye-catching graphic style.
It is a mix of slightly messy drawings, sketch-like lines and a very interesting use of color. This style is inspired by pop art, it is also characterized by vivid patterns and round cheeks.
I hope that this style, which I personally love, will be appreciated by you too.
UPDATE FLUX:
Beware of realistic and photographic tags as they cancel out the style of LoRa.
Recommended tags: " RAW DRAWING , POP ART , PATTERN ART , PINK (or other colors) CHEEKS "
Description
Hi guys! Do you want to try Graphic portrait in FLUX version? HERE IT IS! 😊💜
You can use the suggested tags together or separately depending on your preferred style.
FAQ
Comments (10)
Why does this lora use tons of VRAM in comfyui with FLUX 1D compared to others that are much larger?
I'm sorry you're having problems. If you can double check your workflow because it's impossible that it consumes more VRam than other models, I'm very sure of this as it was created with the minimum possible rank and in fp16.
What I can suggest is that if you have trouble with Dev you can try different models than 1D like the one I recommended and tested as an alternative. It definitely works faster.
@Lara_De_Martin Yeah its really weird as I can see you made it tiny and I can load up 6 other really large 300mb+ Loras all together in the same workflow and it doesnt go over my 24GB of VRAM but as soon as I use this one even alone it pops over the 24GB and goes into my RAM which makes the gen from 1iteration a second to about 1 iteration every 18 seconds lol
@Vyviel It's very strange, I can't understand. The model has no particular settings compared to others made in the past or later. You've probably already done it but if not I suggest you update comfyui, for example I had encountered slowdowns in generation and updating solved it.
@Lara_De_Martin Thanks for that just if anyone else has the same issue I finally found a solution as I found another Lora that did the same so finally with comfy UI i found if you add the following tag and reserve 2 gb of vram or more it may vary with people it wont go over the limit and put the data into slow RAM. --reserve-vram 2.0
@Vyviel Thanks so much for the tip :) it can surely be useful to others.💪💜👍
Interesting. So it generates stylish faces. I guess I would need another lora trained on my own face to make it work. No? How to stylize my own photos that is?
I tried to reproduce the style that I had generated for the XL version of the model. The model is designed to apply its style not only on faces, but also on other subjects or shots. If you want to use it to create your own Avatar, you don't necessarily need a LoRa of your face, I recommend using the model on IMG2IMG with at least two controlnets, or IPadapters. (I made my avatar here on Civitai like this, with the XL version)
Relay nice and lovely creations
🙏☺️ thank you so much 💜
Details
Files
Available On (2 platforms)
Same model published on other platforms. May have additional downloads or version variants.