Support my work on Patreon and Ko-Fi and get access to tutorials and exclusive models.
Rate and leave a like if you enjoyed it, and follow for new models!
LoRA trained to generate a person in public surrounded by people, such as photographers. Works with other LoRAs and various models.
UPDATE - new training and small improvements
Updated training: used LyCORIS instead of LoRA, trained on larger dataset with more realistic images, cleaned up tags.
Updated output: less likely to have a camera on character's hands, more detailed crowd, a bit more consistent, you can better increase the crowd or photographers by increasing the trigger tag's weight.
Main tags (use one or both for different effect): public, photographers
// OLD VERSION (1.0) //
Main trigger word: public Other tags used photographer,phone screen,camera,foreground
You can use your usual CFG and weight 1.0
See weight comparison 0.0 vs 1.0, minimal effect on style and composition:
Description
New training: LyCORIS instead of LoRA, larger dataset, cleaned up tags, more realistic images.
Output: less likely to have a camera on character's hands, more detailed crowd, a bit more consistent, you can better increase the crowd or photographers by increasing the trigger tag's weight.
FAQ
Comments (23)
Could it create just pov hand holding a phone and photoing, without crowded photoers?
not this lora. it was trained to generate crowd or crowd photographers.
do you mean like a selfie or something else?
@psoft Not selfie actually, I mean a first-person perspective photoing phone screen. In most cases it refers to the viewer himself is using his phone taking photos of the girl.
Similar to this lora: https://civitai.com/models/43965?modelVersionId=48602
But instead of taking photos, this one more like checking the phone....
@bluestarr I see, so it would be similar like that lora, but instead of looking at pictures, it would be the camera app, showing what's behind the phone.
ngl, it seems complex. I don't know how hard it's going to be to train a model to replicate basically the same image in one. I'll see what I can do
@psoft That's a imaginable big challenge, tribute to you.🥳🥳
In my thoughts just hands and phone screen in correct position is ok. No matter what's on the screen, girls can be copied to it by photoshop, and that's what I have been doing when using the lora mentioned above, lol.
yay v3.0. The best gets even better.
终于等到你
Hello, Here's a suggestion: you can import some fixed faces into the crowd and load them as needed instead of regenerating them every time. Otherwise the faces of the crowd would be strange.
八成以上的大模型都没办法生成摄影师，浪费几个小时，泪目
are you using lycoris adequately?
@psoft of course,I can only successfully trigger the cameraman in Perfect World, it adds clothes inexplicably, using "ChilloutMix" is just normal trees, in other 2d, 2.5d checkpoints are not triggering the cameraman, even though I set weights up to 1.8
破案了，在很多真实模型使用“public”这个触发词只会生成树木背景，只用“photographer”就可以了
thanks! good to know
Brilliant idea, would be great if this was a Lora
hi @Toastyghosty the first version (v1.0) is a lora. v3.0 is locon because I used more photos and it was harder to maintain an anime look otherwise. v1.0 still works well as a lora.
@psoft Version 1 is a lyCORIS too using Drawthings so unfortunately will not work for me
@Toastyghosty , v1 is not a lycoris. I rarely use lycoris and at the time I created the v1 lora, lycoris was just starting to be used. so NO, v1 is not lycoris
One of my all time favourite lycoris, but just now I found out this lycoris will entirely put into shutdown regardless of weight once paired with style lora derivates from artist named ThePit.
Aside from that, everything else is nice.
Thanks for the feedback!
Does it happen with other style loras you’ve tried? As that one could be overtrained to the point that it conflicts with this one
@psoft Yep, any derivate style or merger using "ThePit" style like the gopheralpha and that one style which I can't quite remember right now that have this weird/unique interaction. All of them also came from different creator/uploader so the chance of every single one to be over trained shouldn't be that high, right?
Been using your public lora for months now so it also rather surprising when I found out about this just yesterday.
Edit: Other style seems to be fine & doesn't trigger weird interaction. Been trying the MetalOwl, Kunaboto, Rampage, Fellatrix, and many others which only need slight fiddling here and there but it should works just fine.
@Stray_Seaman I see. but it's indeed curious how that happens with different loras of the same style
it's pity that this model can’t be used in draw things
你可以为这个模型制作pony版本吗
Details
Files
Available On (2 platforms)
Same model published on other platforms. May have additional downloads or version variants.