• SLR

  • Any AI/ML stuff you would like to have on SLR?

doublevr sweat

Bro idk how many times I've suggested this but I said directors should have a water mist bottle to spray the hotties down by the end of the scene, so they look sweaty.

Sweaty sex is the best, both bodies wet from sweat smashing together creates magical musical melodies bro, and mist the hair, to many times, maybe in 96% of "virtual porn" videos ends off with the hotties hair still fixed up, makeup still on fleek, hair and maybe makeup should be in a mess after a good smashing.

Sweaty, hair messed up, sex smell in the air, we need that in vr.

Edit
We always end off with the scene the female telling us bla bla bla, bro a great pounding "sex" will knock a girl out, after great sex bro, do females talk in normal voice, maybe not...they just came, lots of exercise moving and grinding, after all that and Cumming she shouldn't even speak, she should be catching her breath, talking in tongue and lays next to you, arm on your chest...and fall sleep, that's usually how great long sex ends, and we have scenes that last over an hour sometimes 2. 🤦‍♂️

Just sum a thought for directors to consider.

    probably controversial, but could AI remove tattoos?

    Script AI that recognises non POV action.
    Random wank script AI generator which detects certain scenes as action?

    @xoxo why don't we do sex toys voice control: harder, harder, softer, cumming now

      I would like to be able to apply skin color and facial characteristic masks to the actors.
      Like change Melody Marks into a tan Japanese gyaru. Or make Vina Sky into a blue Na'vi . Or give Chanel Camryn anime eyes and mocha skin. Or make the stunt cock match my skin color. Or give Jewelz Blu a moustache.

      doublevr
      I would personally go crazy for the following features:

      1. Color Adjustments: Change hair color, eye color, and skin tone of performers. Like changing Jewelz Blu's hair to a different color, giving a performer a specific eye color, change PT dick color, etc.

      2. Face Morph: Change lips, eyes, nose, teeth, skin, and make up of performers. Very useful for replacing lip filler with regular lips, selecting a different make up style, giving a performer a nose job, giving the performer straight teeth, removing acne or pimples, etc. AR filters can already do this currently.

      3. Voice Personalization: AI generated performer voices that say your name or short custom dialogue at specific timestamps.

      4. Face Swap: Use AI to swap a performers face with another face using reference images uploaded by the user to SLR drive or something else.

      5. PT re-lighting: Use AI to automatically color correct the user's PT video using their current environment's lighting setup. This should automatically mimic the user's light source and color-correct the video accordingly.

      6. Voice Search: Use voice to search within the headset app (not sure if this is already there).

      7. 2D to 3D conversion: Convert user uploaded 2d videos in SLR drive into SBS 180 3d videos (using AI generated depth maps?).

        I'd like to use these features on the website and the headset app (if possible).
        I would also like to download the AI enhanced video.

        doublevr voice control would be good too for camera positioning too with headset view (eg. "recenter here" after just looking to the new location, "rotate down", etc...)

          xoxo

          There's definitely some low-hanging fruit on the ML front around recommendations and rankings. eg cosine or jaccard similarity over video-video audiences for recs, Wilson scoring for rankings, etc

            Would it be possible to de-mosaic JAV with AI? Thats what really turns me off watching JAV. Having an automatic fix to that would be awesome!
            Another great feature with PT would be to locate the position of my cock and zoom/tilt/rotate the video in a way it is alligned with the actor. Doing this manually all the time with every new scene kinda kills the imersion.

              The end goal is purely synthetic ai generated vr porn. Anyone can generate their dream scene with their dream type of girls. Gotta train a diffusion model on every vr video in existence.

              I think 2D non-pov porn conversion to 3D pov porn would also be possible not too long now. We can already turn 2D videos into '4D' and re-frame or move the camera, with the 3D data 'behind' the frame generated.

              Automatic height/distance and scale adjustment. Some videos especially older ones have these issues, and some male models are a lot taller than me and the action feel too far away.

              Body morphing, no more debates about big tits or small tits or busty or petite. Adjustable talent for each person's preferences and suddenly every video becomes a good option for everyone. If it's an ethics question, maybe talent can opt in for the feature.

              synthetic 360 degree views. Should be relatively easy, because it can be static, just generating the extended sections around the viewer. DALL-E was able to do kinda do this ages for an image ago and video editors would use a screenshot, generate the extended parts and then put them together in a NLE.

                11 days later

                Tone down the moan feature
                Would be great if we had a button to reduce excessive moaning porn acting sounds. Drives me crazy sometimes

                thatoculusdude Not exactly AI related, but i've been hoping for SLR to make another "Breaking the 4th Wall" type scene with a more popular model. It's a nice change of pace to have the model speaking directly to the user instead of the male talent.

                  What I prefer is when the skin, especially on the face, looks really natural and not too smooth,

                  Hairsational By the time a diffusion model capable of it is available, computational power would already be multitudes higher. Well probably, considering we can already generate playable games in real-time on consumer hardware.

                    AI can scan every scene and compile a list of identifiers/tags like:

                    • Clothing type/colours/styles

                    • Hair styles/colours

                    • Eye colour

                    • Even obscure things like fingernail colour or long / short, fake tan, everything

                    So one day in future someone can tell SLR they are looking for a scene with a short small boob girl with blue eyes and dyed red hair with no tattoos and the search will bring up such a video if it exists

                    What can AI/ML do with voice/audio from a video?

                    It may be interesting to see what AI can do about scripting instructions from JOI videos for use with haptics. Not the ones where they're jerking off a dildo, but using AI to recognize the voice commands from the girl like "faster", "just the head", "long, full strokes". Then executing those commands on the toy.

                    Or giving the option to remove speaking or moaning from the stunk cock.

                    Maybe selectively customize the pitch/timbre of the model's voice so that she's lower and more sultry or higher and chipper.

                    It may have been mentioned before, but what about manipulation of what the model says so that you can insert a user's name or remove any mention of "step-"? I guess if you're going that far you may as well look at changing the whole "story" of the scene by changing the model's words, but after a certain point it would look like 1970s dubbed kung fu movies.