Thread for AI generated videos.Thread Rules:>No prominent AI glitches>No spamming>No blackedsee>>>/g/ldg for guides and technical support>Low end cards: https://rentry.org/AnimAnon-AnimDiff>High end cards: https://rentry.org/wan21kjguidehttps://rentry.org/wan22ldgguidePrevious: >>29835688
>She just gives up halfway through
>>29844903she's switching over to a titfuck.
>>29844903>>29844915naw she's having a seizure
>>29845023
In which a woman has a peculiar post-hypnotic trigger.
>>29844875nice
>>29845023>>29845025more
>>29845426completely fake boobs that's crazy
>>29845354If you're taking requests, I'd always be down for more Asuka. No rush though.
>>29845653Very nice. Wish it was a lot easier to get good foot stuff out of the image gen models and WAN.
>>29845025
>>29845874
Does anyone have the muscular tanned woman getting a facial from a couple threads ago?
"Maybe you could be more sexier sucking dick at work".Number One Work Ethics.
nanobanana pro is unlimited for today on some sites, anything knows something to do with it?
>>29845879This is turning me on way more than feet should. But Misato is a real one and this is a great rendition of her.Requesting a similar video with Misty if able. Much appreciated
>>29844875Delisharoo !
This is the ideal female body. You may not like it, but this is what prak fertility looks like.
>>29845354NEED more Miko gens, I fucking LOVE young girls with big fat asses
Ok were do I start if I wanted to make ai videos of women sucking my dick?
>>29847051BIGGER
First time degenerate here, ended up downloading grok, never used it before, using chatgpt for daily shit and chatgpt codex for code reviews other than that living a normal life not different than pre llm So I am fucking shocked what kind of shit you can generate with grok video, obviously it’s still censored and no actual nudity other than accidental nip slipThat being said, can someone tell me in simple terms is where a way to have grok level video generation not no censorship? I have 5090 I guess it’s relevant, only 32 ram but can buy more if it’s needed if it’s possible Thanks
>>29847452read the links in the OP.
where the fuck do you find decent workflows. this shit is so gatekept
>>29847596https://civitai.com/models/1309369/img-to-video-simple-workflow-wan21-or-gguf-or-lora-or-upscale-or-teacache
welp this failed>>29845788yeah was making some more sooner or later
>>29847452It's tough to say 'grok level' with no censorship. There are some things grok does significantly better than local models. It's far more plug and play, and seems to have better overall prompt adherence compared to Wan, the best local option. It's also got way better built in audio capabilities than anything you can use with WAN, though I personally think Grok and MMAudio both turn out crap I'd rather leave off. It's good enough for some people, though.Wan has the benefit of being completely uncensored and open-weight, meaning there's a lot more opportunity to make really high-quality gens of specific things and actions if someone's put out a quality lora for it. If you just want to have people strip or fuck, WAN blows grok out of the water.Worth noting, you should be looking into Wan 2.2, not 2.1, with a 5090. Vid attached genned in about 7 minutes on a 5090.>>29847378Oh, nice. I know someone had been asking for Wendy anal for a while. Hope they see your gen.
>>29847596It's not gatekept, 4chan is just a shit place to learn how to do it.Go to youtube, CivitAI. Hell, go to reddit. People don't want to spend a bunch of time explaining technical stuff on a board where everything is wiped after a day or two, and writing out a general purpose guide in a new rentry is a lot of work.My suggestion: Look up a youtube tutorial on how to set up ComfyUI for Wan 2.2. Follow it until you get bored or frustrated, which hopefully happens at some point after you actually have comfyui itself fully set up. Then go to civitai and start trying to make the most popular workflows work. Copy other people's prompts and lora selections. Google or GPT it when you run into errors. If you run into errors that you can't find a fix for, drop the workflow and try a different one.
>>29846884100%! More please!
>>29847835Grok gens look a lot like Kling, I wonder if Elon spied on them
>>29847916I think it's just a matter of both of them being models too large for consumer hardware. I'm willing to bet that they basically just threw money at Grok imagine. Having it interconnected with their huge LLM for context and ballooning the model size to something in the hundreds of gigabytes could brute force you into something that outcompetes smaller models in at least SOME use cases, if not most.
>>29847999
>>29847835Now that I have a 5090 as well, 7 minutes seem kinda long. Are you doing something special in your workflow or genning at insane resolutions? On a warmed up Smoothmix V2 workflow, I can generate a 113 frame 640x960 video, 3 steps high, 3 steps low (~11.5s/it) in about three minutes with the slow-ass frame interpolation taking up at least a third of the time.
>Mom... I've been doing internet porn. Me and a couple of my friends from school, we told you we were on the debate team. But after school, we go to my friend's basement, and we shoot videos of ourselves in different positions, and different little stunts. We give them our own little nicknames. It's me, and Mari, and then that boy that works at, um... his dad's company. And, um... we shoot videos, and then we put some different music behind it and we, um... post them on the internet.
#DontDropTheSoap
>>29848127Yeah, I'm genning at around 960x720 before upscale. If I'm actively experimenting, I kick it down to a max dimension of either 720 or 540 and get my gens in 2-3 mins. But the high resolution is necessary to cut the blurring and improve eye stability once I'm happy with what I'm getting. At that point I'll just add a bunch of randomization to it and let it run for eight hours, then just delete the gens that have issues afterwards.
>>298481792/10, the girl that sits down doesn't get a belly bulge
>>29848179>they're all sitting on onepatrician
>>29844871that's just beautifulis there more like it
>>29848255That resolution doesn't seem much higher than the one I mentioned, but sometimes those things make a huge difference. What do you get in terms of seconds per iteration?
>>29848332Looks like around 25-30 seconds per iteration on what I'm running right now. Not sure why mine's slower. Might be the checkpoint I'm using, or the number of loras. I also don't think I'm using sageattention.
>>29848411That's quite a bit slower. Would really recommend sageattention and also replacing the umt5_xxl_fp8_e4m3fn_scaled clip ran on the CPU (think that's what the workflow comes with) with a plain umt5_xxl_fp16 and running that on the GPU (default), if you haven't done so. The latter might really save some time, seeing as the prompt is different for each run with your dynamic prompts.
>>29847967>>29847976sauce?
degenerate loop
>>29847983>>29847973Their tongues could transform into alien tongues that resemble sounding beads and go down the urethra to finish off the scene.
>grok thread starts getting more popular>all the worst spammers disappear from /vdg/ and are now spamming /grok/really activates my almonds
>>29848701/vdg/ used to be waaaay more active than this even before grokposting though.
I gen'd this a while ago with wan2.2 and no loras with the intention of flashing her tits ("woman removes her shirt" or something like that) but it ended up inventing a massive bra instead. Is there a way I can proompt to get the bra like this again in the future?
>>29848486I was still running on the cpu (with plain umt5_xxl_fp16), but switching to default doesn't appear to have sped anything up. Actually, come to think of it, I bet the issue has something to do with my ram speeds. I installed four 32gb sticks a while back to try to run larger LLMs locally, and had to dial my ram speed way back as a result. I'll still see if I can't fix sageattention, but I'm betting that's the culprit.
>>29848707I assume you're talking about back then when people were posting hailuo, pixverse and kling stuff.
>>29848723no, even as recent as september we were still way faster than we are now.
>>29848719Interesting. Running Clip on the GPU really gave me a huge overall speedup whenever I changed the prompt, but I guess all setups are slightly different.I don't know if RAM speed really impacts the seconds per iteration all that much, when everything fits within VRAM. Will probably make switching models slower. I'm currently running two 64GB sticks at 6000 MT/s, CL34-44-44-96. Wanted to have the option of expanding the RAM, but that's most likely not gonna happen with the current prices for RAM kek
>>29847738this went worst
>>29848723>>29848731Would be add on to greasemonkey to delay moderation scans would be nice
>>29848870I'm running a 9800x3d, so it might just not be a bottleneck on my system. Or maybe it's just not a bottleneck if you aren't running sageattention.And yeah, I didn't realize that you generally can't get full speed when you use all the slots. I have 6000MT/s ram sticks, but I'm stuck at something super low, like 3800. Don't remember exactly. I honestly might pull two sticks out and sell them so I can go back up to 6000MT/s.Playing with local LLMs was fun for a bit, but API-based options are just too far ahead for it to be especially worth it.
>>29848898wot?
>>29848906Exactly. :p
>>29848575very nice
>>298489039950X3D here, so it's probably something else.Damn, 3800 MT/s is rough. Kinda glad I only went for 2 sticks.
>>29844871Grok might be chucked but good lord look at the water physics it can generate.
>>29848411Please do Semiu from GachiakutaHer thick lips are perfect for sloppy toppy
no idea who these people are
>>29848543Somehow this too I can jerk off to
>>29849345>>29848168>>29848486More asuka please.
If you use the Fusion lora and go above the steps they say to stay at the gen times are longer than if you didn't use the lora and you still get the movement nerfs :^)
I downloaded pixverse for Android and generated 1 tiny clip from a Jordan Mcewen image. But now the app is demanding money for more gens.I'm very hesitant to download a mod version of pixverse from the many shady apk cracked websites out there.Does anyone know of a mod or cracked apk ai gen app?What apps do you use? Do you gen from your phone or laptop?
If I watch ads on pixverse, I can go back to generating tiny clips after earning some credits. But I don't know how to reveal her giant B( . Y . )BS.
I must view more ads in order to remove the watermark. What a greedy dev -.-'
I typed in: The frog on the girl starts rubbing his face against her bodyAnd this anticlimactic video is all we get :(
>>29849671install an android emulator, create a disposable gmail and install modded apps
>>29849769I can do that from my phone?
Lame.I'll never be an elite coom brain ai gen wizard like the rest of you :'(
Motorboating penis in rose's pussy. Or maybe this was a nightmare all of this time?
>>29846972nice
Some of you have some horrible taste.
>>29844871What are there no women that look like that?WHY?WHY?This is torture!
>>29849777LMAO
Anyone have luck running wan on Amd windows?Getting fucked by HIP error even after SDk installed, not sure if pytorch build is installing correctlyWish I realized amd was this bad before I got the card, my vram is good but this compatibility is awful
>>29848168Peak
yo
>>29850555
>>29849567cute indie animation set in a weird themepark with ai avatars and stuff.
>>29850267>that orgasm face>that after bliss face with each trustMUH
>>29849612Perfection
>>29851044Those are weird tattoos.
>>29850555>>29850561Now do a few where she's super into it
>>29844871Requesting breast expansion.
>>29851062she needs a white dick for that
>>29851224Eagerly awaiting moar
>>29851224That's a lot of effort. What did you use?
How do i stop WAN2.2 I2V from adding pink shit to drawn asses? As soon as the character does anything with their asses it makes a little heart or flames come out of the ass. Sometimes it just makes the ass cheeks glow neon pink.I'm not even mentioning her ass in the prompt.
>>29851550like what the hell man
>>29851554What's your prompt? I used to get that with pussy reveals sometimes.
>>29851580>(at 0 seconds: shot of a woman, she has a flirty expression).>(at 1 second: the woman is acting suggestive and horny).>(at 2 seconds: she shows off her body in a flirty way).Just trying to get something super generic going that works with anything so I can just dump a load of random r34 pics into it and see where it goes.
>>29851604What loras are you using?
>>29851608None. I guess the Enhanced Lightning v2 preset in Wan2GP counts as using a lora.
>>29851595fuck capcom?
>>29851618Yeah, I dunno. I don't think wan2gp has nsfw built in, so maybe try putting in the generic nsfw lora. You can also try the smoothmix quantized models which have lightning and nsfw built in. I'm assuming you're low vram since you're using Wan2GP.https://civitai.com/models/2083588/smooth-mix-wan-22-gguf
>>29847051We need way more of this
That bored expression when the man isn't looking. lmao
why is /gif/ so fast all of a sudden?
>>29847835Which accelerator do you recommend for wan 2.2 and how many steps and/or mag cache speed are you using?
>>29851985>>29852204Wife. Love asuka so much. Is it possible to give her muscular chun li legs?
>>29848168Requesting an upscaled remaster of this gem of a gen.
>>29851643Yeah that's not Headmistress Fiora, that's Cockgobbler Fiona from Shrek
>>29852228That would stop being Asuka and just be some fat bitch with a wig
>>29852414>>29851214
>>29852183I believe the accelerators are made for people with low end cards trying to gen higher resolutions. A 5090 can do that out of the box so not really necessary. From what I seen on reddit most 5090s all gen at same speed regardless of "speedup" tweaks. Resolution is the biggest factor for gen speed. You can do 480p in like 2 min. native 4k resolutions will take about an hour for 5 seconds of video. There are upscalers and whatnot as well that may be quicker.
>>29852438Exactly
>>29852448Got it. And may I ask how many steps you are using for your BJ vids and if you are using mag cache to skip steps?
>>29852470I'm not the guy that posted it so can't help you there. Im just another anon with a 5090. I've never touched the mag cache setting and for BJ vids use one of the Loras off of Civi. Wan 2.2 is pretty intuitive and you can probably even do a BJ without a Lora if you prompt it. For 2.1 you definitely need a Lora.
>>29852490No worries, thanks. I'm just trying to find the right balance between speed and quality. I know I can generate these quickly with less steps but in my case the quality drops significantly for my taste.
What's the best "big ass walking away" lora for 2.2 that I see you guys using? Or is the base model able to understand the concept?
>>29852448Accelerators are purely for speed at the loss of quality. Higher resolution is more of a matter of VRAM and RAM if offloading.>>29852470There aren't many people using Tea Cache or Mag Cache because there are barely any steps to skip when using a lightx2v WAN workflow. You would use those if you were running 20 to 40 step generations without any distillation LoRAs. BJ anon uses Smoothmix which has lightx2v baked in so assume 4-8 steps but you can probably just grab a workflow video off his Civitai if you wanted to check.
>>29852556Is it true that lightx2v destroys motion that the base model is capable of? As in, you will get less motion when using it?
>>29852556Sorry, I probably conflated accelerators and distillators and didn't realize they were different things. I'm using Wan2GP and he has lightning i2v baked in, is that the same as the lightx2v WAN workflow that you mentioned? I apologize in advance if I sound like an idiot (spoiler: I am)
>>29852228Seconding this request, more thick Asuka can only be a good thing
I want to generate women pissing their space suits, which models and LORAs do I need?
>>29852567Yes to a degree. Lightx2v for WAN2.2 does indeed stifle motion a bit and there have been updates by the Lightx2v team to address that but whether or not you notice the lack of movement or action is dependent on what you are generating. Most people tend to mitigate the issue by adding LoRAs for the specific movements they want anyway so the speedup is generally preferred because it really does save a lot of time. Also some people prefer using Smoothmix since it is a bunch of trained LoRAs merged into WAN 2.2 for movement and appearance with lightx2v baked in or Dasiwa which is of a similar vein.>>29852650Probably the same. Lightx2v is the team and they tend to name their distillations based on whatever they are implementing or whatever version. Lightx2v, lightning, seko, etc. Usually found as a LoRA extracted from the trained distillation model either done by the team itself, Kijai, or Comfy. If it lets you run WAN2.2 in 4-8 steps, assume it's using some variation of Lightx2v. WAN2.1 had Causvid and Accvid before we got Lightx2v.
>>29852414>>29852462>>29852438>fat bitch with a wigI mean its cool if you dont want to, i love normal slim asuka aswell. But i clearly said "muscular" and "chun li" wtf about that is fat bitch??
>>29852745Thank you very much for the explanation. It really helps noobs like me.
need more muscule girls
>>29852745Sorry, one more question. To clarify, do you know if BJanon is generating his vids using lightx2v?
>>29851754
>[Baywatch theme intensifies]
>>29852925Smoothmix has lightx2v baked in and BJ anon uses Smoothmix. His Civitai is Alberist. I had recommended grabbing one of his vids off Civitai earlier to get a workflow but Civitai is shitting itself like usual so it might just be easier to wait and ask the dude when he is here. Now that I think about it, his uploaded vids might not even have a workflow anyway since he does some post processing.
>>29852183>>29852448Hey, just wanna say since none of the responses have been me: I actually do use a speedup because I use the SmoothMix checkpoints. They have it built in. It DOES help, even for higher end cards, because it directly affects the number of iterations you have to do. I find speedups have negligible tradeoffs in some situations, and are occasionally even helpful. (Too many low noise steps can cause 2d gens to look weird and blurry.) My experience has been that speedups largely hurt the model's prompt adherence, which can be supplemented by using more loras for the kinds of things you want. Basically, if you're just trying to show a sex act with minimal camera movement or other action (like I usually do), speedups are worthwhile. My gens are generally 6 steps total. (Compared to the 40+ steps I'll use if I'm trying to do something more complicated with regular Wan and no speedups.)I don't use mag/tea cache, though. By my understanding, those do have significant degrading effects.Overall, it can be a bit of a crapshoot. Sometimes Wan just doesn't want to do the thing you ask, speedups or otherwise. I spent like an hour the other day trying to get a gen where someone threw a cell phone they were holding in the starting frame. No matter what checkpoint or step count I tried, it wouldn't work. They'd usually turn around and generate a phone to throw in their other hand. The best way to do it is to try lots of workflows, checkpoints, loras, etc.
>>29852177India is awake
>>29853101Yeah, my vids don't have a workflow, but there's no secret sauce to it. I use the Smoothmix V2.0 workflow. I only really alter it to add stuff to randomize my gens. https://civitai.com/models/1847730?modelVersionId=2264611
>>29853016Is there an online cache of her or similar bbw characters somewhere?
>>29853796Fuck yes, more asuka!
>>29853793One with Shrekdick?
>>29850155who? point em out.
>>29853860yea been making more but i've just been working on trying the loop workflows it's bleh
>>29853393>that pussy gripWe need ginoids NOW
>>29851214Perfection.
>>29853872I've already genned all the initial images for her. Maybe next time.I do have turtle dick, though.
>>29853793>>29854078Please do Asuka langley.
>>29853936
>>29846884Who's the artist that creates the thicc videos?I need a source.
>>29849375dats sum nice h2o
>>29853796>>29853982Incredible work.
>>29854436Thanks.I really just think it's slop but it's fun to mess around with. gonna add sex to this some how, but not tonight.
>>29851224amazing! I want to see where the story goes
>>29854629Lovin' it
>>29847965awesome
>>29849751she grows a third arm
>>29847980rin is always best girl in everything
>>29847961i like this, good one
>>29848168the way the dark haired girl licks the dick at 0:02 is so fucking lewd
>>29851643damn that's some hq shit, what did you use to make it?
>>29849375>Grok might be chuckedSkill issue.
>>29849751Here you go
>>29855089Steal my look starter pack:Model: https://civitai.com/models/1995784?modelVersionId=2260110Workflow: https://civitai.com/models/1847730?modelVersionId=2264611Current MVP Lora: https://civitai.com/models/2121078/big-cock-oral-wan22-i2vAll works for anything from anime to realism.
>>29854629loop test not to bad
y'all sleeping on deviantart, all the best ai generated video comes from there. can be pretty hit-or-miss thoughsauce thirdcurve iirc
>>29855404How is there no motion and image distortion for the loop transition?
>>29847999>>29848026You wouldn't happen to frequent /pcg/ would you? I floated this concept a while back...
>>29854629>>29855404Astounding as always. Do you take requests?
civittybitty just deleted their entire profile on civit AI, anybody have their stuff saved?
>>29856282wtf why?
Have a couple that I just generated in the past day or two
>>29856414And 2/2
>>29852204Do you have a longer version of the first part?
>>29856575No, it's genned at once so that's as long as it is.
>>29844871Any guide for literal retards? I mean step by step to do I2V
>>29855212Do you use anything other than oral ?
>>29856773Check the links in the OP.
>>29855087The telltale sign of a cock-hungry slut
>>29857003fucking love maids
>>29855768It's 100% still there but it's harder to see when it's gened at such a low res.In MM audio news it seems the video has to be 24 frames per second to sync properly, so you would have to take the OG 16 fps gen interpolate to 24 fps, MM audio it, take that and interpolate to 32 fps... all for that sweet AI slop sound
>>29857090That seems really annoying as frame interpolation takes almost as long as both KSamplers....or is there some magical frame interpolation, that is both better and faster than RIFE I don't know about?
>>29856358I recognize all these transition loras except for the handjob one at the start, did you use a specific lora for that one or just the normal handjob lora.
>>29856799I use other loras, but nothing's really vital. I switch it up a lot. Even big cock oral can be replaced with Ultimate deepthroat. Unless you're asking if I gen othe things besides oral. In which case yes, but not nearly as much.
>>29857321Just a handjob lora, wan2.2 can do transitions by default so it just worked.
>>29857273Nah interpolating a finished 16 fps video to 24 fps took like a few seconds in RIFE, the RIFE that happens in the normal gen workflows do a lot of other shit as far as i know and it seems RIFE is the fastesthttps://civitai.com/articles/20293/darksidewalkers-wan-22-14b-i2v-usage-guide-definitive-editionthis article talks about RIFE, FILM, GIMM-VFI interpolation and a bunch of other stuff. the loop workflow I use has GIMM-VFI and that thing takes it's fucking time
>>29857405Okay, thanks. Maybe I just underestimated how much it's usually doing.
>>29857376nta but thats.. hot. More. Maybe one without the cock, so just her showing off her mouth/throat
>>29857367continue, really nice stuff
>>29857090Not to shit on what works for other people, but AI audio is just so far from good enough for me. It pulls me out immediately. I think I'm more attuned to uncanny audio than visuals.
please more transformation stuff any and all
>>29853735Thanks for the tips. I did some testing last night with Wan2GP Enhanced Lightning workflow and I got a significant improvement in generation time with minimal quality loss. I could generate a 480p video in 2-3 mins with 4 steps where as before I was trying to brute force it with native wan2.2 i2v and using accelerators, 25-30 steps, and magcache, which still took me 20-30 mins to generate. I'm floored at the result.in summary, if you're a noob like me, make sure to use Wan2GP's built-in lightning enhanced model instead of the lightning accelerator loras with native wan2.2 i2v.
>>29846734is that michael jackson?
this is pretty decent for genning audio for nsfw vids, not perfect but getting therehttps://civitai.com/models/2137833/nsfw-dead-simple-mmaudio-rife-interpolation-setup-for-wan-22-i2v-14b?modelVersionId=2418378
>>29857668I agree with you I think it's trash as well. Maybe some day it'll get better.
>>29857090>>29857886These gens but with Rei and Mari pls
>>29857668I find it to be okay for just sounds. As soon as any voice, even if it's just moaning, is involved it feels very off to me.
>>29857535
>>29858000yea, the issue with MM Audio is that it crowbars in moans way more often then not, even when it's in the negatives. would have to edit out moans manually and that's on top of you even getting anything good
>>29856282>cum is not colorfulyou had one job
>>29844875when I see outputs like this, I don't even know wtf I'm doing wrong anymoreI have a 4090 and have been experimenting with these models for a couple yearsI've never been successful with getting anything even close to as good as this >>29844875yet my text2image and image2image outputs using a number of other models (both A1111 and ComfyUI) have been great.but with T2V and I2V models my outputs are kinda trash and generation takes longer than it should. I've used a number of different models and comfyui workflows, followed guides including the one linked in all these threads on rentry for wan2.1 and 2.2can anyone provide a full comprehensive guide to set up comfyui, get the models and workflows AND with concrete examples (prompt, exact steps/cfg/etc settings in the workflow) to produce an output like this?
>>29858133Just use any default workflow and type "woman walking away"?
>>29858133Start with the workflow, Wan 2.2 S. I2V is very user-friendly: https://civitai.com/models/1847730?modelVersionId=2264611And this guide: https://civitai.com/articles/20293/darksidewalkers-wan-22-14b-i2v-usage-guide-definitive-editionThe 3rd tricky part is the models you are going to use, some models will be a bit faster than others, so its going to be a bit of trial and error.
>>29856282God fucking damn it. I knew I should have saved that shit.
>>29858133WalkAnon uses Wan2GP, a one click installer on Pinokio that takes care of all the dependencies for you, works right out of the box, and has an easy to use UI. He's even said that he still uses Wan 2.1 instead of 2.2, and for those walking vids he uses one of the walking away loras on civitai. I've copied his workflow to great success and I have a 4080.https://github.com/deepbeepmeep/Wan2GP
>>29857367How about her belly swelling like a balloon?>>29855404>>29857090>>29857886>anusNice.
>>29858690>>29858133I use this walking away onehttps://civitai.com/models/1363473?modelVersionId=1550982If it doesn't bring you to the right one it is the i2v 2.0 for WAN 2.1, I prefer the 2.0 over 3.0 version because of the slow motion, after I gen (at 480p) I use a pirated version of Topaz Video AI to upscale the resolution and up the FPS from 16 to 60, here is the original for reference (used handbrake to turn it into a webm because the mp4 file is 27.6mb)
>>29858764Love your stuff, Mr. Walker, sir. Will you be taking requests in the next thread?
>>29858764for reference I genned this on the 480p WAN but made it output for 720p (I sometimes do that if I don't have a lot of pictures to go through)>>29858766Sure, just characters are fine and if you want them to be at a location/type of clothes(keep it simple)/type/color of panties, say them in the requests I don't mind changing some things up
best version of anya taylor joy i've seen. not good enough to make her missionary. always ends up with her vagina looking like a beefy penis.
New thread: >>29859084