Video source gives following error " This content is not publicly accessible "
Can you have a check on the authorities
Many thanks
Video source gives following error " This content is not publicly accessible "
Can you have a check on the authorities
Many thanks
Fixed. You can also suggest another source that could be used for testing purposes, the one currently selected is fairly random.
Test this one then
Davina Davis In One More Rep.funscript (345.5 KB)
generation time: 3m12s
I would also add the following things to his analysis of the current AI capabilities:
the inference is fully deterministic (meaning that generating a script multiple times for the same video would result in the same outputs every time). At least with the current architecture. This is wonderful news for doing comparisons as the AI model improves, because you can be confident that it would not produce variations in the output.
for hentai and furry animations it is still in its infancy, where itâs anyoneâs guess whether it would even correctly pick up what is going on. This is obviously not a current focus (most of yâall are just gooning to people not animated shit, iâm clearly in the vast minority of users on this website xD), however me and other furs are going to be providing training data for this over time as well, in hopes that it will eventually improve also on this end. But just to give you an example, this is a short animation that would definitely be impossible for traditional vision trackers to handle, including point tracker models like Co-Tracker. Itâs literally a turntable fucking animation, a very rare sight even as far as animations go. And aside from the detection being generally off by some frames, it actually picks up the rhythm extremely well even in its current state. Check it out, this was just one click run, no edits, took maybe 10 seconds to generate:
While I have my own opinions about the pro tier pricing, there is no denial that this tech has real potential. I hope a lot more people will showcase their results with it so that it gives everyone a better understanding of the current capabilities. Ultimately, both developing this app and training the AI are NOT cheap things. This is a one man project, so your hard earned money go directly to funding this personâs work and life, not some fucked up sex tech corpo that couldnât give more of a shit about your feedback and treats you like a cash cow to milk. You can either spend that 5 bucks a month (annualized to 60) on this project, generating however much your PC can handle while helping a genuine hard working individual, or kinda just piss them away generating AI scripts from those greedy corpos at like what 3? 4? bucks per script? with maybe a marginally better AI gatekept and no guarantee of quality. These are results you can see right here, right now, the community is more than happy to show you their generated scripts and I hope that @k00gar will also make the free tier restrictions a bit more lax for this beta period to give people more confidence in the project and the models.
So please donât just read that 150 price as a be all end all, think for a second and understand what youâre getting, even at the free tier (leaving OFS in the dustbin of history where it belongs, for one).
A few things I came across - this is noted by using 2.2.1 and 2.2.3/2.2.4, sorry if something is done already, - wishlist/remarks:
Ok , that was nearly spot on in 2D, but what whit a vr movie
if that one is as the one above it is an instent bay to pro
Popcorn_Trap_For_Pretty_Gamer.0deg.funscript (154.7 KB)
Popcorn_Trap_For_Pretty_Gamer.funscript (138.2 KB)
Two versions: one with default settings, and in the other I changed the VR view angle to 0, because the AI scripts what it âseesâ. The main weakness is that it tries to capture movement even when there is actually no action happening. Fortunately, those can easily be deleted. You can provide additional source videos for wider testing coverage.
Hera are some another type of vr that might be an batter example
Come to discord and post bus and feature requests please. He takes them there.
Extra Load - Abigail Mac 4K.funscript (572.7 KB)
t&a Solo Edition - Kagney Linn Karter 4K.funscript (265.3 KB)
Tried, didnât turned out well, so Iâm gonna keep it here or nowhere, feel free to copy&paste if you like.
So what about some other formats, like 3D Animation and Hentai?
3D Animation?
Different 3D style but with multiple character types?
Normal Hentai?
Koikatsu Style?
this, I have been using fungen for the last 2 months and been in the discord and decided to pay for the pro for a year, it is only 30 for a year atm I believe, It is worth the different using the pro model, I hope it can be refined and maybe even have a feature like F.A.P.S for PMVS etc, an ultimate all in 1 funscript maker.
Whats your name on Discord? Have not seen you post anything on Discord. Oh and I am for sure not your copy-paste mailman.
YeYeBirdie Sex Academy EP02.funscript (96.1 KB)
Japs8005_milf-deep-blowjob-compilation-1_1080p.funscript (160.2 KB)
Nagachichi Nagai-san 2-1080p.funscript (138.9 KB)
[Takoiyaki] Raiden x Yae Miko â Horny.funscript (76.6 KB)
v2.2.7
Nagachichi Nagai-san 2, weakest of them, Iâm not sure whether the issue is slight animation flickering or censorship using larger blocks
FunGen wasnât even intended for or tested on 3D animations/Hentai, but apparently the AI doesnât really care what type of motion itâs processing
source: https://discuss.eroscripts.com/t/kana-arimas-ecchi-debut/320856
â Kana Arimaâs Ecchi Debut â - Day Version.funscript (22.7 KB)
Hentai: mixed results, in some cases it follows the movement, in other cases it doesnât
Looks and works great, thanks for your work.
Even bought the Pro 1 Year. However, will there be better models for JAVâs (especially VR JAV)? Sadly the tracking even with the Pro Model does not get some scenes right (even though the screen is quite clear).
Ran these through Multifun Player to get a better idea and Iâm honestly pretty impressed. Will have to play around with this software for sure.
They definitely came out a little extra aggressive, but itâs actually really impressive it managed some of those.
I am extra curious how the software handles multi axis because it could be a game changer for adding multi-axis to existing bangers of single axis scripts.
No audio with free tier or am I missing something?
The best thing you can do is provide the names of those scenes along with their timestamps, or even cut out clips from those parts and send them to the developer. That way, they can investigate exactly what the issue is, and the next model may perform better.