• Content

  • Why don't we get into private deepfakes?

Isn't a bit tricky ? Because what assurance have you that people won't use model who didn't have given their consent to be related with porn ?

I mean their is already some people uploding cosplay porn with the deepfake of the original actress...

    Amazing idea but...

    I personally feel deepfakes will get a huge regulatory crackdown very soon. With the release of StableDiffusion, It has now never been easier to create deepfakes. It's absolutely nuts how you can pick any instagram photo of a local girl.. brush her clothes out as mask, add "naked body" in the prompt and it perfectly undresses the woman in the photo. Works also the other way around by pre-training the model with 10 - 20 photos and unleashing your wildest fantasy on that person.

    I'd say tread carefully with this since revenge porn is at an all time high at this point. As much as I'd love a feature like this, you would contribute to an evergrowing societal problem that haunts many females today... making it more accessible to create deepfakes, download and spread it around.

    An option would be to start with an SLR-curated face gallery that are approved and to finetune the technology. Then after a quick lawyer-check if there's a grey area to operate in this... maybe try out user-submitted content as well.

    I wanted to make a topic about that a while ago but for the above mentioned reasons I didn't.

    It could be a revolution in the mainstream porn industry but it has to respect some conditions

    We could think of creating VR scenes exlcusively for deepswapping contents. For each scenes, we have the choice of picking a selection of faces for that.
    Every women who participated in that project need to give their consents.

    How could that be done?
    My ideas is that after each SLR shooting sections, take a close facial shot of each models if they agree to be part of the project. It could include models that normally shot solo and only G/G.
    With time, you could accumulate a lot of faces and those faces could enter in the choices of girls for the scenes made for deepfake.

    There is a lot of room for abusing this, but what GaryLopez suggested sounds better. If there is a database of girls who have given their consent for this initiative, it should not be a problem.

      Kieshi I like this idea. Basically we sign girls and guys all over the world to use their faces for videos.

      I still think we can be fine with user photo uploads given nothing of that will be shared on SLR. It's the same like Photoshop where you can put any face you want.

          doublevr oh men as soon as pictures of people that didnt sign papiers hit the ram on your servers you have a huge liability problem. if i understand it right "I still think we can be fine with user photo uploads given..." because you where in touch with it. And even software dev´s of opensource deepfake stuff get in the high quite often.

              1. Deepfakes never really looked that good or realistic ever, so I don’t have any faith they will in VR.

              Of corse, if users could select any pictures they want, it is way better. If they do it for a personal use, there is no reason to blame.

              once deepfakes become sufficiently advanced, it will be the end for a company like SLR.
              Instead of 20,000 unique videos, you will have a small number of videos and hundreds of unique "configurations"

                catflap Who do you think will render these 20,000 unique videos or supply data for them?

                  Smart, that's why I like SLR for always experimenting and innovating.

                  catflap You make it sound like it's easy. Do you even know what kinda powerful servers are needed to edit an 8K VR video on-the-fly based on the users configuration? Who is going to provide the VR player for you to make it all convenient? I don't think it's gonna make SLR obsolete.. if anything this use-case can strengthen their popularity even more if done right, fair price and compliant with the law.

                      damian catflap didn't give a time requirement on what he said. He said"once deepfakes become sufficiently advanced" which could be 10 years or more from now. I do think he has a point.

                        doublevr I think photorealistic CG faces would be much safer. doublevr I don't know where you live but in the US you will gets sued so fast if any girl finds out a guy from her school used her pictures for deepfake porn. It don't matter if you are legally allowed or not.

                          Intriguing idea! If executed well, this will give us exponentially more talent to choose from. I'm sure a lot of OnlyFans girls, cam girls or even some instagram models will be interested. It is much less intimidating sending a few photos than shooting hardcore porn.

                          It also creates a win-win situation if you allow the girls to promote their OF/cam accounts on their profile pages. They'd get a lot more subscribers that way... I'm sure a lot of men would love to talk to girls they could "fuck".

                          And they can help promote SLR on their channels. "Want to know what it's like to have sex with me? Well now you can. Sign up to SLR and experience what it's like to have sex with me. Use the code AliceSLR for a 7 day free trial"

                          A technology which, as far as I can tell, is somewhat behind visual deepfakes at the moment is voice replication, and I think that the audio aspect of a scene is as important as the video.

                          Most research I've seen in that area is more along the lines of providing text to a model that then reads it out in a voice it's learned- but it feels to me that it should be significantly easier to use prerecorded audio and shift tones and pitches to match the target voice. This will probably fall short in terms of things like speed and rhythm of a voice, so not sure how convincing it would be but I think it's worth exploring.

                          I think the idea of having contracted models for deepfakes is the correct one both legally and morally, so it would be good to take some voice samples as well as high quality images from angles optimised for the best deepfake results.