Porn Studios wanna make money. Everyone does. AI has eaten our jobs, but it cannot eat our desires for hot sex. No matter what happens, we will still have dicks. Earth's population is 7 billion, so there are still 3.5 billion dicks to serve on this planet. A lot of money to be made :-). And someone needs to fill the gap. Who better than SLR :-)?
Every once in a while, dicks needs cock-milking with babes having blonde hair, huge tits, dirty talking & huge ass. Hell, SLR made passthrough finally work. (https://discuss.eroscripts.com/t/slr-can-we-do-3d-object-segmentation-for-passthrough/86243). There is a special section on passthrough in SLR Originals. Quest 3 finally made colored passthrough a reality. SLR shifted from a "VR" company, to a more general "sexual-wellness company". They promised to roll out their own stroker (https://www.sexlikereal.com/blog/380-the-new-slr-interactive-sex-toy-is-coming-soon)
Now that 2025 has drawn to a close, it is time for some reflections. AI has made a lot of progress. We need to ask ourselves: what's next? Are we done?
1) It appears that green-screen full-body passthrough OUTPERFORMS running on-device segmentation models.
While green screen is a cheap setup in long run, i guess the business still pays the cost of EACH male talent dressing up in a SEPARATE "special-suit"? Just curious haha. (you cant share vacsuits between different talents due to hygiene issues)
2) A key problem still remains: how to generate pass through for normal /old slr content. Natively applying per-frame segmentation failed miserably. For eg, https://www.sexlikereal.com/scenes/passthrough-sitting-cowgirls-2-vr-compilation-32189. See the flicker-here.
Recent advances appear in models like Meta's SAMv3. They perform video-object segmentation, so they can retain temporal context problem via "streaming memories". No more flickers.
3) SLR introduced subtitles for most of the "jav" content. This relies on running speech to text model like OpenAI's whisper. This craps out in "multi-group scenes" like (https://www.sexlikereal.com/scenes/dreamteam-26815), when MULTIPLE girls are speaking into the "SAME microphone".
We need the ability to: 1) pinpoint which girl is saying EXACTLY WHAT even if MULTIPLE girls are speaking at the same time, in different languages. Turns out, this is known as "audio source separation", https://ai.meta.com/blog/sam-audio/ , which has now become a reality. The biggest hits on SLR are "group scenes". Time to bring this tech in VR porn.
4) If a user gazes at a "particular" girl in the headset, her sounds will be filtered over "all the other girls", and result in seamless transitions. Your gaze in 3D is already tracked in apps like Heresphere. (i know @doublevr doesn't like the guy lol )
5) Interfacing with touch haptic devices: We now need sensation of "touch" on human body. For eg, devices like (https://www.bhaptics.com/tactsuit/tactsuit-pro/) allow a user to wear a vest, which simulates touch via redirecting output to a "vibrating motor". There are 24 vibrating motors spread all over your body.
Now, Imagine a girl running her hands over you in a massage scene, and you wanna feel every movement. You would need a "script" to "route" the sensation of touch to correct motor in a vest. There are two ways to solve it:
use spatial audio: which tells which speaker the girl is closer to (left/right) and how far from the speaker (based on volume). A second solution is to track her hands, and feet (in footjob), and give that output to the motors.
Right now this idea is used in gaming. This Tech will get cheaper. Time to build the software stack FOR VR PORN.
5) 6DOF scripts: Devices like OSR need 6 dof scripts. Right now, there is a script generation tool which generates only one axis (the stroke direction). With recent work (VGGT), it is possible to "infer" 3d orientation of a dick in real-time. Six axis script will improve experience. However, not everyone has Do it yourself OSR.
6) Handy, Kiroo Keon have an issue: they are strokers and NOT MILKERS.
After "one orgasm", dick gets soft. And you cannot use handy with a soft dick (trust me, i tried, a lot!!). And men need multiple orgasms. The solution out of this hiccup are "MILKERS" like Nimblestroker (https://shop.exploratorydevices.com/) (which allow interfacing via connectivity modules). Need support for those devices too.
7) Edging Mode: Edging mode DOES not mean making scripts with pre-defined slow-movements.
What it means is the ability to detect when YOU are gonna have an ORGASM, and automatically REDUCE stroker speed. Devices like this exist (https://darkmatter69.com/products/nogasm-denial-made-to-order?srsltid=AfmBOoobd84b9eQKloFTn3XEGwsu14FYhMUOOGanXKMbtxE76GSCSbx3), but no interface with SLR.
8) Auto PMV creation: The gold standard is NoodleDude's PMV (https://www.noodledude.io/videos/ultimate-tiktok-pmv). They are famous due to two suble reasons:
- Perfect beat synchronization
- A single frame is split into three separate screens (like Tiktok videos, with each containing different girl). For eg, there may be three different girls, each riding-dicks. There are methods which can "temporally align actions".
So you can synchronize the videos in such a way , that three different girls ride the dick in exactly same motion. Trust me, it is amazing for immersion. Check: (https://arxiv.org/pdf/2007.04515)
- A feature like "auto pmv creation" could be rolled out. User selects the best videos he likes, and the system creates a pmv specifically for him.
9) Multi VR watch session: SLR rolls out "rooms" where multiple users can watch the same porn, and share live audio/video feed.
10) The market not yet captured: those who don't yet believe in VR. But, Brazzers /Naughty America are dead in the grave they dug for themselves. Only time has to pass before their own users pivot.
Woah, such a long post. This is the end (for now). I will go back into the silo for now, and think next set of cool ideas lol.
till next time,
ccing those who make the revolution possible :-): @doublevr @AlexSLR @alexnash @RealcumberScripts