Thread for AI generated videos.Thread Rules:>Only videos made in Local AI video generators are allowed such as WAN or LTX or others>No videos made in Popvid or Grok or any other similar low-quality generators>No AI Celebs >No Prominent AI glitches for no reason>No Spamming>No Blacked>No Shitposting>Wan 2.2 Guide:https://rentry.org/wan22ldgguide>LTX 2.3 Guide[________________________________]Previous: >>30919571
I can't wait to see..>Arguments >Really shitty gens>People asking me to make my own as if I'm not allowed to criticize >The good ole 'no u'>Poop, gross trolling vids.>I have to sift through absolutely insane psychopathic mentally retarded shit.>I'm forced to look at the most retarded shit in the world.I just want to look at tits.
>>30922200the rules say you gotta post 3 clips when starting a thread lest you incur janny's wrath.>>30922218just shut up and post vids.
>>309223034060ti 16GB Minimum5060ti 16GB for decent gen3090 24gb for comfy gen5070ti 16gb for comfy gen + futureproofing
>>30922267>"And we go with another autistic psycho misinterpreting words">"No u"
>>30922354?
>>30922200Damn, no LTX guide yet?
>>30922313Any AMD options? I've got a 9070XT.
>>30922596I don't think there should be one, and I'd also recommend removing the WAN2.2 guide. As good as it was in the beginning, it's now the way more confusing option compared to just installing ComfyUI Desktop with the official installer and just messing around with their templates. There could be some workflow or model recommendations at most, as I don't think using stock WAN2.2 is what I would recommend in the context of this thread.>>30922677Install ComfyUI Desktop with the installer from their website, and test wether or not WAN or LTX work with the inbuilt templates. The installer will automatically grab the correct pytorch with ROCm for you. I dunno whether or not they fixed ROCm yet, as it was quite prone to OOM a few months back.
if you're a complete noob just get pinokio and install wan2gp.
>>reeeaaaaly nice.
>>30919569reeeaaaallly nice.
>>30919735
>>30922200>low-quality generatorsOnly at some times it's high quality.
>>30923024Yeah, Seedance 2.0 is really impressive for example, but it can't really do actual porn.
>>30922697But don't you need the models from civit to make nsfw stuff? What about the light loras and all that
>>30923514They're not strictly necessary, though strongly recommended.I think model and workflow recommendations would be completely fine, instead of installation instructions that are pretty certain to confuse some less computer savvy people. Not sure who should come up with said recommendations, as that can be somewhat controversial, especially when it comes to DaSiWa stuff.
MOREBREASTEXPANSION
LESSBREASTEXPANSION
>>30923712GAY
>>30923758wrong link
>>30923686GAY
>>30923804>"Yeah, that'll show em. I'll just repeat what they said."
may I humbly request tastefully muscular women with huge tits?God robbed me of that perfection IRL since any woman who cuts hard enough to show their musculature also loses most of their titfat along with the rest.But I will defy heaven.
>>30924103What's your definition of tastefully muscular?
>>30924138NTA. I think for we're doing here, if you can look at it and say, yeah, I'd like to taste that, then it's tasteful.
>>30924138I would do exactly this to Beid.
>>30924138extremely likable abs
>>30924158>how the hell did you get all o' that in dem jeans?
men want only one thing
>>30924103>defined abs >big tits truly a blessed combination
>>30924138I'd say this is the limit
>>30924103sorry, I only have futa
>>30924103how big do the tits need to be?
fuck
somebody could try to make AI video of Ms Apple from The Pit comic?
>>30924500
>>30924505FUCK
>>30924500>>30924513shit like this makes me go feral
Drama between 10eros creator and Dasiwa creator
>>30924531Go on
>>30924559Its still ongoing, basically Dasiwa stole the DMD Loras 10eros made
>>30924500>how bigYes.
>>30924591lap pillow with my face in that abs
>>30924566You mean, the DMD loras that were published freely on the internet for all of us to use?
>>30924566Where's this drama even happening?
>>30924437yo more?
>>30924625moar maid service
>>30924437>tfw no onahole company will ever make an onahole with silicone head and gel like boobies like vid rel New fetish unlocked regardless
>>30924628
Can I get a boobjob version with this?Thanks. No need for audio.https://d.uguu.se/MKHioJth.png
>>30923758MOREGAY
>>30925206coming right up
I still dont know how Distilled LORA works. That shit is like magic to me
>>30923958amazing
made this yesterday as a test for walking with LTX and LTX is surprisingly shit at this type (probably prompt skill issue)
>>30925622Forgot, anyone know how to get her to not walk as slow? lol
>>30922267>Man From Volta fucks for the honor of the mask
>>30925232Damn if only that cock was flopping around uselessly
>>30925622Despite multiple loras, I haven't gotten anyone to walk decently in LTX.
>>30925681Only time was me testing prompting with chatGPT and it was a selfie walk (ignore the subject matter I literally ask ChatGPT to have her talk about DBZ as a test)
>>30925631
>>30925766Big titty blonde talking about DBZ with her nips out, and you think I'm going to ignore any part of it?
>>30925681Adding "sexy dance" or "sexy move" lora might help but not sure how "sexy dance" would work without music. The weird boob movement is because of the BE lora, was playing around seeing if it worked with music.
>>30925779Got scared af when she turned around
>>30925779thanks, have some boobs
>>30925825Well, I put beautiful woman in the prompt but LTX aint entirely stupid and knows its a youngling
>>30925837Maybe short asian or petite?
>>30925837Yup, did beautiful asian woman.. LTX wasnt buyin it and wasn't worth 10 gen runs just to post that slop.>>30925681There is also "bouncy walk"https://civitai.red/models/1361346/bouncy-walk-ltx-wan-i2v-14b?modelVersionId=2999921
>>30925914This was one of my fantasies as a kid
>>30924494More like this.
>>30925959
>>30925766wtf is dragonball?
>>30926259DragginBallzz
>>30924500>>30924513Any specific prompt? Loras? I tried trickling rolling, beading, I just got rain and the sweat turning into pearls.
>>30922313>5060ti 16GB for decent genGood to know. Someone in a SillyTavern thread a while back was chimping the fuck out some shit that its not possible to gen anything at home.
>>30926318I gen on a 4070 12gb, they are only right when you are comparing to the ungodly checkpoints that need like 120gb of VRAM, you can gen video as low as 4gb (it's fucking awful but you can do it)
>>30922313I have the last one, but I'm reading decent genning wants at least 24GB. The LTX2.3+eros I use is shit.
>>30926333>LTX2.3+erosTry one of the GGUFs. They are smaller and probably offer better quality (and speed) for your card.https://huggingface.co/QuantStack/LTX-2.3-GGUF/tree/main/LTX-2.3-dev
>>30926305Pretty sure there is a wet body lora. Saw it on Civi week or two ago.
>>30926333i have 16gb of vram with 64gb of ram and i have gen with FP8 and INT8 models with loras. Avoid BF16/FP16 models. They are for workstation GPUs
>>30926333I avoid using the Sulfur/Eros/whatever models that have baked in shit cause you dont know what strengths they use or what conflicts it will cause with other loras. The only baked in one I will sometimes use is Smoothmix.
>>30926333What version of 10Eros are you using? I'd just give the fp8mixed version a try. With ComfyUIs dynamic VRAM that might just work and maybe won't even be that much slower, while giving you something better in terms of quality. Anatomy isn't quite there yet with 10Eros, though, as also stated by the author, so be prepared for fluids to come from pretty much everywhere when you prompt for them for example.>>30926383I think 10Eros is fairly good. Not having to stack tons of loras can really help retain a lot more quality.
>>30926053Moar!
>>30926686NTA. I found the the full safetensor of 10Eros_v1.4 is actually faster than its quants. I've also found, though, that quants give more dynamic gens. 16GB VRAM, btw.
>>30926993Full version as in the 46.1GB bf16 version? That really shouldn't be faster. Maybe the quants already didn't fit in your VRAM, so you had swapping either way, just less dequant.
>>30926686Fingering lora works somewhat okay.
>>30926858Same here I want a whole load more of these.
>>30927013Oh, no lol. The fp8 for sure.If i had to guess, maybe because by the time the separate text encoder loads, the quant doesn't save that much space, but the 10Eros_v1.4 safetensor is AIO so that reduces load times.
>>30926858>>30927062Thanks and appreciate the love guys but just got back from Vay-Cay and have some projects to get done around the house today. Those 3 were all I managed to get done yesterday (couple of them were really fighting me on the popping/cumflation/audio sync) parts. Next weekend might try and get Director and Qwen installed so I dont have to do as many gen runs.
>>30926283More like this
>>30927086Do love to see it would love to see more scenes maybe the woman herself loving it then struggling and then begging for it to stop etc I have good pc specs id love to dabble with all this type of stuff but unsure how id even get to doing it.
>>30927150I'm probably gonna setup a Catbox for the library of past gens and could post longer stuff there too. Hard to get much more than a few thrusts and a line or 2 with 4mb. Also when I go over 400 frames (18 seconds-ish) in base LTX the video tends to go a bit off-script with prompt/voice so hopefully Director can help with that.
>>30927150Also did try a few with the woman having lines but most gens when the woman is talking LTX will conflict that with the popping and not do the pop action at the end but again maybe Director will be the fix for that.
New title and rules seemed to have slowed down the thread a lot.Realised I did a very grok like gen here, but apparently it's entirely possible to get the same jank out of LTX as well.
>>30927175any new stuff coming soon?
>>30927944Well ..kinda procrastinating on the projects today so might hop on the SlopGen2000 today but not sure. Tomorrow is a "school day" for me so if I throw anything up will be in the next 6 hours or so.
I followed the Wan 2.2 Guide and did some tests and all the videos were man made horrors beyond my comprehension.Do you use the workflow in the guide to create these videos? Cause the look fucking amazing and it's way far from what I'm generating...
>>30928230Wan 2.2 can't do NSFW out the box. You need to either add individual loras or use an all-in-one baked in model like smoothmix. https://civitai.red/models/1995784/smooth-mix-wan-22-14b-i2vt2v?modelVersionId=2513182https://civitai.red/models/2299142/smooth-mix-wan-22-i2v-v20-gguf?modelVersionId=2587351
>>30928247Thanks bro,I haven't used LoRAs node yet because I wanted to see if there was a huge difference, but I did download the models in the guide and will check the difference.Will provide when I manage to get some good content.Any additional help appreciated
>>30928275Lot of the gen stuff is trial and error figuring out the right strengths/prompts for what you want to achieve. Only other tip off hand I can think of is use a Lightning Lora with the base Wan 2.2 model, but some models you should not use the Lightning if it is baked in (Smoothmix has some iterations that have Lightning lora baked in).https://huggingface.co/lightx2v/Wan2.2-Lightning/tree/main
Can someone generate a quad amputee femboy or futa with big dick being fucked and enjoying it?
Kinda curious would the AI be able to make an animation from these stills from this inflation manga? [Matinee] Semen-Pumping Sybian-Lamp ?
>>30928452Dunno why it shouldn't work. Just give it a try,
>>30922267You think many are able to do that in a short time frame. Get a life, loser.
>>30927944Here's some cookies. The POV ones fought me a lot and didn't turn out great. Also whatever rocket scientist believes the IC lora helps with 2 people talking during sex is fulla shit. Here is the garbage it spit out lol.
>>30929178
>>30929181The missionary ones always had her fall off the cock, then explodes or splat or whatever.
>>30929190Meant the POV ones. Also it always had the cock on backwards. Ugh
>>30929194Trying to get the "explode-ee" to talk never worked. It was either they both talk or just the Futa. Doesn't help that usually the Futa is always looking at the camera and other woman is not.
>>30929200
>>30929206
>>30929211
>>30929215
>>30929220Was kinda rolling on the reverse cowgirl prompt so tried to squeeze this one in with same prompt. Did not work out.. but amusing
>>30922218thread rules: get butthurt if popvids are posted. lol jk
Surpising LTX 2.3 still cant do - Titjob (Male thrust)- Female Masturbation- Tentacles to this day
>>30929932?? 1 and 3(sorta) you can do, what you can't do is breast/ass gropingUse a sex lora for missionary/doggy on a lower strength to get the thrusts. If you don't lower strength you get a random vagina popping out of her chest
>>30929940What about female masturbation ? I found this https://civitaiarchive.com/models/2586130?modelVersionId=2905250 but its really bad for anime, it instantly turns to realistic
>>30929953I have seen a masturbation one but never tried it so I have no idea sorry
>>30929977Its even harder if the male do the masturbation on female. We need more loras
>>30929178Oh fuck... Finally get the idea of the fetishUnbirthingFuck that's gonna itch like hell, my toaster of a laptop can't gen anything.
>>30928581there's an idea or two.your dialogs are of the worst.nice pussy, ugly face.who cares about seeing the man's face?
>>30930297Thanks for the feedback, I guess. Dialogue is intentionally retarded, hence why they both laugh about it.
>>30930079divine
>>30928298Are most people running base wan 2.2+nsfw lora or hsing dasiwa or smoothmix (plus loras also even though I think most loras are trained to work with base?)Im torn, with 10eros 1.4 ltx 2.3 is getting competitive but some wan 2.2 loras are still better, and it just gets fluids, breasts better.Anybody get id-lora to work well on ltx 2.3 even on a 2 pass workflow? Beginning to wonder if a 1 pass workflow is better generally anyway…
>>30926993iirc safetensors are faster unless you spill out of vram into ram, thats where gguf will win.
>>30928811 ??
>>30930691kek. Sorry. :'( :'(:'(:'(:'(:'(:'(:'(
Can anyone give my wife a big futa cock?
>>30930739I could but I won't.
>>30930540I use LTX 95% of the time just because I like the sound and it is 2x faster. Wan is much better at doing expansions (breast/stomach/muscle/ass/etc) and sometimes tweaking a starting image if it isn't quite right for LTX but aside from that I'm all in for LTX now. Have had zero luck with ID or IC loras but I usually have like 6 other loras going on so I'm sure its some conflict with other loras.
>>30929710Should have had the camera pan over to people at a table with point signs at the end. I'd give her a 6.7 on form.
>>30930739
Last week's leftovers.
>>30930833I’d love to see her sat on that or be the one with the huge cock, I can’t post a picture on the thread tho
>>30930739The real your wife's big futa cock is the friends you made along the way.
The secret to getting them to say "creampie" instead of "crempee" seems to be "cream-pie", fyi.
Where do you guys post your videos, other than here and civitai? I found r34gen. Can't post on twitter, right?
>>30930841I have TeleGuard if somebody want to Ai edit my wife and give her a big futa cock3B6E7BCDY
>>30930848Yeah, writing things out, so they're phonetically easier for LTXs audio stream is almost mandatory for some words.>>30930851I almost exclusively post my stuff here. Uploading stuff to my Civitai is reserved for gens I really like or poured a bit more effort in.
>>30930851I only post here. This is all just time wasting for me while I'm at the desk anyway during my work week. Don't care about buzz or whatever - absolutely no tangible benefit to posting anywhere more permanent.
>>30930874Same for me, except I don't do it at work.
>>30930886WFH, babe. Helluvathing.
I should take Taiga out of the randomizer, or at least reduce her chances of hitting. I'm never really happy with anything I produce with her. The schoolgirl thing doesn't do it for me, either.
>>30930866Fax. LTX sometimes pronounces easy things like Jesus as Jess-Us
>>30930874You can use buzz to train loras. Haven't done that yet, but it's an option. Think the main benefit to posting your better gens on Civitai is that you're at least guaranteed to get some reactions, even if they're just thumbs up or hearts or whatever.
>>30930838Shoot, I didn't even notice she's not actually speaking the first lines. Seems like a problem with 10Eros, that early dialogue in the scene is just narrated.
I just want to faceswap and remove clothes, I'm a simple person.How do I do that?For clothing removal therenwas a website called deepsukebe some years ago, but I want to do it offline now.
>>30930950I think there's a pretty good instructional video for face swapping. They used some pretty big name actors as the demo for it, I guess because of how easy it was to get footage for it maybe. Somebody may have a better memory of this than me - I think it was called Face Off?
>>30926305Found it.https://huggingface.co/Lightricks/LTX-2.3-22b-IC-LoRA-Water-Simulation
>>30930951Yah if i remember right that was one of the first workflows/videos promoted for AI generation because it was SFW and showed a side by side animation of an animatronic on one side and people replicating the animatronic dancing on the other side and you could put whoever's body face in place of the animatron.
>>30930950I mean a quick and dirty method would be use one of the iBlink loras and just prompt what you want her body to be. I've done breast expansion and muscle when using the iBlink before.FYI this isnt Iblink lora
>>30930967>>30930952>>30930951thanks frens I'll try this one
>>30931003Here is iBlink. If you just want nudity think there is a shirt removal/nudity one.. there's like 10 iterations of it..
>>30931017>>30931003I'll look into it, thank you.I just want images, my hardware is not good enough for video anyway.that's why I mentioned faceswap (can pick an image from ant porn and add the face I want) and clothing removal (get a photo of the person I chose and have all the clothes removed, like deepsukebe used to do
>>30930911at least on civitai I can actually get some fucking requests filled
>>30931084Learn to gen.
>>30931084Guess so. If you're willing to pay, at least.
>>30931063Ah got it. You would want something like this then. Base model is like 28gb, but there may be GGUF versions.https://huggingface.co/Phr00t/Qwen-Image-Edit-Rapid-AIO
>>30931086If I wasn't a poorfag I would.
>>30931101Post PC specs.
Man, sometimes it just does exactly what you asked it to do on the first try.
>>30931113Saw some "seed seeker" workflow/node over the weekend but its basically hacking comfy to find a good seed so would probably make comfy super unstable.
>>30931121I wasn't being sarcastic, lol. That was exactly 100% perfectly what I prompted for.
>>30931109>i7-4790>750ti 2gb>8gb ram>1tb hddgaze upon my specs and despair, I do
>>30929200Biggest problem with these aside from the bottle rocket is the bulge never moves or gets any deeper.
>>30931148Ok, this is truly awful, yes. Check out Runpod, I guess. You could build your workflows locally, save them, then upload them to Runpod for execution.
>>30931156It's a conflict with stomach bulge lora and explosion effect. Its an either or for now until I play around with Director.. plus LTX does stomach bulge for shit, Wan is much better.
>>30931156If you saw all the shit in my prompt to get the effect you wouldn't believe it even spit out a gen lol
>>30931235Any chance you could blow a girls head off with a facial, perhaps from two or more girls? (think of it like a firing squad)
Sometimes you gotta ask the real questions.
>>30931267Yah tried the head thing and its a nogo with prompting alone in LTX/Wan, at least intentionally.. sometimes it will give you body horror but only when you don't want it. Would probably have to try image edit and transition it. Another project for director at some point. I think one gen it did make the head come off but it just kinda rolled off lazily and onto the ground.. not at all the effect I wanted. Closest I got was this which was made in Wan 2.1, which seemed to let you get away with a bit more graphic stuff than 2.2
>>30931290A-non?
>>30931295I've tried head inflation and also a nogo. Think the head is like a hard object for the AI that cannot be altered much (have had success with cheeks/eyes/throat bulge and some head distortion but very hit/miss). It will do stretchy head, to an extent.
>>30931325Haha, yeah. I didn't even realize, was just excited to have a decent gen.
had to compress the shit out of it to fit in <4MB
>>30931290
>>30931437That's amazing. Good work
>>30931444Can I alternate places?
>>30931148>750ti 2gbBut hey, that was some state of the art stuff, like 10 years ago : P
>>30931458Pretty sure that's against the rules.
>>30931455God damn.
>>30931511Yes, not quite the voice, but I was pretty happy with it overall.
>>30931478GTX 1080 8gb was released 10 years ago
Sometimes the fails are better than any success could hope to be. Can't stop laughing at this.
>>30931934What do you use for text-to-speech?
>>30931944That was just LTX itself, as I found the voice it came up with okay-ish.
Welp, that's probably about it for me today. Hope someone enjoyed something.
Whoop, had one more come through the pipe before I shut down.
>>30931455circuits and mana?
>>30931685>her heels r balls
gen loli ap
>>30932111That's Saber from Fate. It's relevant to the plot.
>>30932121Dude, what you think all this anime stuff is? Obviously
>>30932155robots dont run on mana. What runs on mana doesnt require circuits
>>30932158i dont hear any ap#
Needs more gens from The Amazing Digital Circus...
>>30931457have some more
>>30932797
>>30932605You're not getting shit
>>30933071More minigirls
>>30933258
>>30933291>>30933296>>30933302Very niceThanks
>>30925622Hey welcome back Walkanon.
Trying out LTX for the first time. Seems like a steeper learning curve than it was with WAN. Spent all day trying to get something that isn't a complete mess or tits that actually looked like tits. Anybody have any tips, loras or workflows they'd like to share?Mostly interested in doing breast/ass expansion or big bouncing tits/asses. Face drift and general distortions seem a lot more common from what I've seen and motion doesn't look as fluid, but that's probably just a skill issue on my part idk.
>>30933696Heyo, was always lurking and I found out Krea 2 is fucking amazing at upskirts/voyeur stuff so I have just been non stop genning images
>>30931261I didn’t know Chucky and Misty had a daughter
>>30933851Krea 2 is amazing in general. Pretty easy to achieve some fairly realistic images, if that's what you're going for, and prompt adherence is also pretty good.
>>30933967Had a 3 day vacation from here?
>>30933977Nope. This is my best gen so far. I really like the result. The stickyness and the dripping of the liquid make my cock hard
>>30929932No Footjob Lora too
>>30931455How did you get it almost looping ?
anymore popping vids getting made ?
>>30933071New fetish unlocked.
>>30934004Any particular workflow to keep the somewhat 2d anime style not get raped by ltx?
>>30934197No.
>>30933772From my experience LTX is not at all good with expansions. Wan does it smoother and more accurately. Especially when clothed it will merge the clothes being removed with BE so when the top comes off the breasts grow with the clothes. You could try "her breasts slowly grow throughout the entire scene" but again very hit a miss mainly because LTX is very seed dependent so may take 10 gens to get 1 good one.
>>30934197Only popvids in another thread lol
>>30934368Also I had to run multiple loras with BE (bestbreasts, macromastia, BE) so they looked 1/2way decent, otherwise they come out bimbo looking fake. With LTX your prompt has to be like a novel and very specific and you have to explain every single movement in detail. (heavy, breasts, breasts have natural movement, etc).There is no ass grow or stomach growth for LTX. I have tried to do other body parts with the general growth lora but again very hit and miss though admittedly did not test it too much
>>30934379LTX can also be quite literal. Sometimes when I put "her breasts grow" it will just add another set of boobs so will have to change prompt to "her breasts become larger/bigger/10x their size), etc.
>>30934338Don't think there's any trick to it, except for heavily limiting the scope of movement. It'll turn into 3D real quick if it moves too fast, be it the whole scene if the camera moves fast, or just hands after a sweeping movement.It's a lot easier to go with an in-between style, that makes it less noticeable. The typical Illustrous anime style with very detailed shading.
super cringe
Having an issue with Comfy. I downloaded the desktop app and have tried to download some extensions and for some reason it has trouble applying them. When I check the extensions installed it says there's nothing. Has anyone here come across this issue? Any help would be appreciated. Thank you!
>>30934596Just install comfyui manager and get all custom nodes, through that.I'd suspect that the custom nodes you downloaded have some requirements.txt inside with the required python packages. You'd most likely have to install them, either globally or in the venv. Would recommend going with the comfyui manager over doing that, though.
>>30934606Doesn't the comfyui manager already come installed with the windows desktop app?
>>30934619I have no idea. Never used the desktop app.
>>30934629What do you use? Wait a minute does anyone use windows application....?
>>30934371By popping think he meant the inflation bursting ones.
>>30934642I just clone the repo and set up the venv myself. That way I can go with whatever pytorch or cuda version I want.That's not as easy as the ComfyUI desktop application, though.
>>30933903That's that LTX2.3 Q6. It gives more dynamic faces, but they definitely come out with a, shall we say, "look" to them.>>30934066It's FLF, which the stock LTX does an ok~ish job of. To tune it up I have a workflow that replaces the last few frames with interpolations to the starting frame. Mixed result. Here's that same gen without the interpolation step.
Boy, sometimes it's a process. I ended up having to take the audio from a clip I liked and chop that up into 3 different pieces, Qwen the cum out of the base image (which of course Qwen for some reason doesn't know what a butthole is but whatever), and then piecemeal the original prompt and run all of that through LTX Director.
>>30934911autismal
any more vids?
>>30933071this one is good
Is there any hope for a poorfag who has a RTX 2070 8gb GPU or is it over for me?
>>30933851You magnificent pervertCan you do a walking Megumin candid butt shot like with the angel girl in the beach from before?
Using 10Eros_v1.4. Trouble with gens not lip-syncing to their own audio. This isn't uploaded audio that I'm asking it to sync to, to be clear. It's audio and video genned together.So sometimes it eventually catches up and the character starts speaking the audio. This time it just never got it together.Anybody overcome this issue?
New slime penetration lora for LTX pulled me right back into doing animu stuff kekIt's neat. Real good internal cumshots obviously don't work all that well.>>30935326That usually only happens to me when the mouth is busy with something else, or the prompt doesn't reinforce that her mouth clearly articulates the spoken words.
>>30935419Ah, so basically really hammer home the idea that it's the woman speaking. Most of my prompts are formatted with dialog as:>she says in a [whatevr] voice: blah blah blah.And that's been good enough.
why are there like million different ai threads?
>>30935652To piss you off.
>>30935652Ask chatgpt
>>30935670If you mean AI "threads" as in separate chat conversations, it's mostly because different apps and AI systems organize conversations differently.A few common reasons:Separate conversations preserve context. Each thread has its own history, so you can have one for coding, another for travel planning, another for recipes, etc., without them getting mixed together.Different AI models or assistants. Some platforms offer multiple models (fast, reasoning-focused, image generation, research, etc.), each suited to different tasks.Different purposes. You might see threads for customer support, brainstorming, writing, coding, tutoring, and more because companies build specialized assistants on top of the same underlying AI technology.Collaboration and organization. Teams often keep separate threads for different projects so conversations stay searchable and relevant.If you're referring to ChatGPT specifically, every new chat is essentially a new thread. Starting a new one gives you a fresh context, while staying in the same thread lets me use earlier messages in that conversation to answer follow-up questions.If you meant something else by "AI threads" (for example, AI agents, models, or something on a specific website), let me know where you're seeing them.
>>30935659that is pretty rude
>>30935677I really need to make a "THIS IS. .4CHAN" 300 parody ai gen
>>30935703you can make wicked rick astley transitions, boomer
>>30935765better than the pool bimbo
>>30935834Pfft...Gen X muthafucka...cause we don't give a fuuuuck
Still fighting that Mash gen. New waifu in the stable, in the meantime.
>>30935882gen sigma here
>>30935892any more?
>>30936111Tomorrow. Promise.
>>30935625I like this
>>30936115many such cases
>>30936149kinda gay ngl
>>30936200>tattoowhore
>>30924437Lost
>>30922677>>30922697ive used it on LTX 2.3 and WAN 2.2, it absolutely works but youll need gguf versions and as a pointer if you are doing I2V keep your source images at a lower resolution (512x512) and output the same and you should get some solid results.
>>30936149Is this what Griffith took from us?
