RTM (realistic model textures) is a series of models that I have trained with various consistent characters from previous Loras in SDXL and Pony. Although FlUX can achieve realism quite well, RTM helps me to get the same lighting every time, the same character, the same textures and the same photographic quality of an average selfie and not a professional. The models posted here follow the guide to realistic and non-cartoon characters that I have previously posted for other Base Models. These FLUX trainings are made at the request of my discord community and do not have a specific trigger due to the captions but if you want to call each character you can look at the details of each model. Otherwise under the temperature
Description
Some people told me they couldn't download the model. I post it again because I can't repair the other post.
FAQ
Comments (8)
I don't understand, is this for the face only? why is it mentioning textures?
Because the dataset contains a series of images in a support folder that I use to give it realism. I usually do the same thing but I change the realistic images for cartoon images and when I train I distinguish them between toon textures and cartoon textures. This is the reason.
@_arkh_ Why...
not post this in the description of the lora?
Why do we have to guess what you are doing?
There are a grand total of 2 promo images. Only one has a prompt, and it also uses another lora in combination with yours. So how do we even know your lora is doing the work in that image?
Also, you still are not explaining what your lora does. Flux can do great skin out of the box. How specifically does your lora improve upon what Flux already can do? Show us.
I am not trying to be rude here, but these vague lora descriptions on Civitai are tiring. If you want to see your loras get more downloads, I strongly suggest being a lot more descriptive about what your lora does and also tell people how to use it. This should be standard practice for every upload. This will help you just as much as it does the user.
@tbsmax I don't think you're rude, now I've posted the description, I've posted the comparison and as for what are the differences between what flux can do and what this LORA can do with flux I can say this LORA is much more natural. When you ask for scenes with depth of field and a bit of motion blur, FLUX exaggerates a lot, whereas with RTM you can dampen it to a much more natural point. I have left a sample in the images of the model and you can also make the comparison yourself by generating images with the description of my generations of the previous version of this model. Sorry for the confusion, I have published the model for the people who have sponsored it, I did not think of clarifying it to other people who were not them because I did not think it would be interesting for them.
How are you making such realistic looking video?
It's Runway's Gen-3, it's the only video generator that achieves that realism, I've tried others but none are as good at making people talking
@_arkh_ Have you tried kling?
@5c0f4n0 Yes, I've tried Kling and it has good results in general, but when it comes to making talking faces it's a bit chaotic. It's still better than dream machine in this regard, but not better than gen-3