Update: Please if you see any of my models somewhere else report or let me know! Thnx
Trained on base model xl 1.0 + Flux1D + ZIT + ZIB + SD1.5
Update: Pony version added
Update: Illustrious version added
Update: zimage Turbo version added
Can be combined with Skin Tone Style XL (Human skin color)
Please read about version info box on the right.
Feedback and ratings are much appreciated.
Feel free to check other similar models:
16mm (Super) Film Emulator style XL
Anamorphic format film style XL (Hollywood Cinematic style)
Cinematic text title (on screen) style XL
Cinematic "color grading" motion picture film style XL
Cinenaut XL (A TRUE CINEMATIC DIFFUSION)
Cinematic Photography Style XL
Cinematic Kodak Motion Picture "Film Still" Style XL
Cinematic Film Stock Footage Style XL
Cinematic Vintage Film Style XL
High Angle (from above) cinematic style XL
Hyperrealism 3D Cinematic Style XL
Kodak Film Grain (Cinematic) Photography style XL
Infrared Cinematic Photography style XL
Lens Flare Cinematic Reflection style XL
Low Angle (from below) cinematic style XL
Music Video cinematic style XL (Including 60's 70's 80's 90's, etc.)
Point of view (pov + fps) cinematic subjective camera style XL
Rotated View (Dutch Angle) Cinematic Style XL
Realistic Photographic Film Stock Footage Style XL
Set design style (cinematic scenic design) XL
Split Diopter style (Deep Focus) XL
Television advertisement cinematic Film style XL (including fashion models)
Text message on screen bubble (cinematic style) XL
Description
Illu v1.0
lower lora weight if needed.
FAQ
Comments (27)
Update: Pony version added
Update: Illustrious version added
Any chance for noob?
Not at the moment. Sorry. But you can try adding any xl lora's, they should work fine.
Great work!
A thought though:I hope that I dont overstep, if I ask why Flux realism still look unrealstic? Seems we cant get read of this waxy skin ? or is this a matter of adjusting weights?
Tons of great detail (and I still strongly appreciate the tools you bring to the table), but the "material" still seems waxy to me (Am I wrong?).
It depends on many factors.
First, some other models can produce better realistic deep texture images.
Flux1 when released, the free version is a distilled model (trained to mimic the predictions of the larger model while maintaining much of its accuracy) one of many reasons why it is not a 100 percent awesome.
As a result, hard to maintain balance on large data set of images without compromising some issues.
Anyways, I always try to update my model(s) with better version(s) when I can if I can.
But for now, Flux has it's own limitations. Best thing is to combine some lora models together thus giving you an overall better detailed image quality. (As long as you manage balancing weight correct)
@EauDeNoire I completely understand. I have indeed noticed this limitation in flow from the start. I was just not sure on how much in the model could be modified and improved. But yes, the way to describe with tweaks and loras is the way now... But work like yours makes this process more fluent and fun.
I used to run SD 1.x in FP32 and I could get very realistic skin using the base model. The problem was that the model practically never managed to assemble coherent bodies. I just find it interesting that the first most primitive of the community-accessible models did skin better than the models that have come since. My only guess is that the companies making them are intentionally trying to make the results look artificial because of deepfaking.
what is everyone seeing? none of the images here look realistic or natural. am i crazy or is this just the best option right now
its not the greatest, it gets the job done tho. and yeah its our best option
You can't just slap it on in a basic workflow... use it with an inpaint workflow along with the skin detection YOLO model and it looks fantastic
@chollman82141 Can you share it?
@chollman82141 Hello??
Share what? Just look up inpaint workflow or learn to use ultralytics models... Do people really not know how to do inpainting? Do you guys not at least do a high res face inpaint pass?
@chollman82141 fym share what, wowwww just look up a impaint workflowww wowwww. Talk about lazy the man talks about yolo models the not all people and cleary me and this dude don't/didn't know about and his response is look it up as if all workflows contain them. Now that I'm wiser I assume he's talking about ken of the yolos like yolo face detection and etc. How are people supposed to learn with "just learn to use the model". Might as well should've stayed vanished like the avatar with how irrelevant you just were.
@Thatonecreator Add 'subsurface scattering' to your prompt.
I completely understand your frustration with being told to look up things you haven't come across before when using ComfyUI. It's a great way of making images but a pain in the butt to learn.
But in all fairness, his response was completely valid. He didn't have to respond to you, his advice was correct, and we all had to learn these things at some point - pointing you in the correct direction gives you far more information than trying to squeeze an overview of technical details into a comment.
He doesn't know what workflow you use, which techniques you use, you didn't give him enough information to give you a brief answer.
Perplexity has a free tier and is a great way of getting answers to these kinds of questions that are specific to your setup. I hope that helps.
@thelustriai In no way was his response remotely valid. Your saying "he didn't have to respond to me" when in actuality he should have never commented anything at all if he had the "look it up attitude". What do you expect us to do, browse through hundreds of workflows to may or may not find what he's talking about when all he had to do was either provide a workflow or say he doesn't have one. I honestly don't care if he made up some excuse as to not give the workflow cuz that would have been better than "look it up". Your take on him not knowing what workflow is used is invalid cuz he was clearly asked for a workflow from nanalaya for guidance. It is the best way to learn from an already working workflow, learning the nodes in it that make it work, than to look up a workflow that may not even work, use not needed nodes, or not even do what chollman is talking about and you cannot argue me otherwise. How would you feel if you where trying to make a strawberry cheesecake donut and somebody came to you and said "man this creme paste i used for it makes it taste so much better" so you ask what was the paste? "Look it up". Sounds goofy right, well so does that guy. Also i'm not the one who needed it the most it was nanalaya
@Thatonecreator There are literally template workflows for this in the ComfyUI menu.
Or - I just entered Z Image Turbo Img2img in the search bar above and then filtered for workflows. This one was right there up the top: https://civitai.com/models/2169096/zimageturbo-face-detailer-upscaler-workflow
Your metaphor about the creme paste doesn't fit at all. It's more like asking someone what brand of creme paste they're using and refusing to look at the tube of creme paste on the table next to you.
Also, I gave you two methods for getting what you want in my first post, and two more in this post. You have acknowledged neither. You're complaining about Chollman not handing you the answer while I'm literally handing you the answer.
I don't expect gratitude on the internet, but the lack of it says plenty about you being less interested in learning and more interested in arguing.
@thelustriai Alr bro idk if you just don't comprehend english but i'm not going back and forth with you
@Thatonecreator I'm proud that you could admit what you don't know and that you were wrong. That takes a lot of courage. Just remember that being so entitled won't get you far when you're an adult one day.
@thelustriai Lmao wtv you feel the need to do to help urself sleep at night lil bro
Any possibility of a smaller version of the Lora? I use AI Horde for generation, and loras are limited to 400mb each.
Check for other similar lora's. Some are even less than 100mb.
Que paso? Antes si salia el estilo realista y ahora ya no. Por favor arreglarlo plis plis plis
Looks pretty much like plastic to me. Nothing realistic.
dont use it on main sampler, use it on 2nd sampler res3_s bong_tangent at 0.3 denoise
Is this a NSFW maker? I tend to generate very sloppy clothed characters, but this LORA stips them totally naked and gives them a very pronounced hourglass figure. I mean, i it has its perks, but it also changes their face shapes, which basically makes my character LORA´s total useless. Thanks for this pleasant surprise though. ;D
No, it's not a nsfw model. Lower lora weight of the model. (As mentioned) Add nsfw in negative prompt. Double check your settings.











