[a / b / c / d / e / f / g / gif / h / hr / k / m / o / p / s / t / u / v / vg / vm / vmg / vr / vrpg / vst / w / wg] [i / ic] [r9k / s4s / vip] [cm / hm / lgbt / y] [3 / aco / adv / an / bant / biz / cgl / ck / co / diy / fa / fit / gd / hc / his / int / jp / lit / mlp / mu / n / news / out / po / pol / pw / qst / sci / soc / sp / tg / toy / trv / tv / vp / vt / wsg / wsr / x / xs] [Settings] [Search] [Mobile] [Home]
Board
Settings Mobile Home
/wsg/ - Worksafe GIF

Name
Options
Comment
Verification
4chan Pass users can bypass this verification. [Learn More] [Login]
File
  • Please read the Rules and FAQ before posting.
  • Supported file types are: GIF, WEBM, MP4

08/21/20New boards added: /vrpg/, /vmg/, /vst/ and /vm/
05/04/17New trial board added: /bant/ - International/Random
10/04/16New board for 4chan Pass users: /vip/ - Very Important Posts
[Hide] [Show All]


[Advertise on 4chan]


Post any AI-generated content, preferably something you generated yourself.
Discussion of local stuff also welcome.

Keep coomer shit on /gif/, there are coomer threads there already for AI stuff.

Previous Thread - >>6208133
>>
>>
>>
isn;t this the new one?

>>6230814
>>
>>
>>6230914
That appears to be a "webms like this" thread, which is not necessarily an AI thread. I checked the catalogue for AI stuff.
>>
>>
>>
File: whybutreallythoughwhy.mp4 (3.99 MB, 1472x800)
3.99 MB
3.99 MB MP4
>>6230923
I think seedance is the best model out there in terms most realistic look among the online services. Grok is a close second, but grok does have better jiggle physics.
>>
File: yeah.mp4 (3.99 MB, 1272x720)
3.99 MB
3.99 MB MP4
>>
File: icanfeelthesun.mp4 (4 MB, 736x416)
4 MB
4 MB MP4
>>
>>6230913
The fuck does he need a camera for? He can see through things.
>>
File: vin00vin.mp4 (2.73 MB, 1168x784)
2.73 MB
2.73 MB MP4
>>
File: sj_00041_compressed.mp4 (5.89 MB, 864x480)
5.89 MB
5.89 MB MP4
1/3

stay tuned tomorrow for the riveting and incoherent conclusion
>>
File: E9ZrtpYQ6KSvHsXe.mp4 (1.14 MB, 576x1024)
1.14 MB
1.14 MB MP4
>>
Anyone got the one in which George Floyd attempts to rob Jerry Seinfeld before Kramer rescues him?
>>
>>6231114
>>
>>6231022
Very dreamlike.
>>
>>6231007
What are they saying? We need more in this style, is there a source?
>>
What happened with Gossip Goblin's Pomegranate movie? He deleted it from the internet with no explanation.
>>
File: sj_00042_compressed.mp4 (5.92 MB, 864x480)
5.92 MB
5.92 MB MP4
>>6231022
2/3
>>
File: sj_00043_compressed.mp4 (5.84 MB, 864x480)
5.84 MB
5.84 MB MP4
>>6231186
3/3
>>
>>
File: h4OzSgm-m1y6Zxzy.mp4 (3.42 MB, 784x1168)
3.42 MB
3.42 MB MP4
>>6231133
vin00vin
>>
File: rViuAIegn0VC0Kjq.mp4 (2.09 MB, 1280x720)
2.09 MB
2.09 MB MP4
>>
How do I make long video? Stitching 10- 15 seconds of video is a pain in the ass, and the quality degrades with every new clip
>>
>>6231366
You gotta make do with the time limits, there's no optimal way around them. Good luck with the video editing!
>>
>>6231366
think of it as each gen is either a new scene of a change in camera. SO H3 you can easily gen 20 secs if you are static to a one camera and then the next gen instead of trying to continues that same one static type camera you move to another camera using either a similar. The obvious problem are continuity errors anything in the background could be different in a new gen and can happen even in the same gen if you say are switching from one camera back to the original view unless you put an end image that has everything you want from the original one - you can even use the same source image. As for t2v gens then you can use an end image. But really the only way to avoid continuity errors is to avoid trying to keep the same background unless you want to go the kind of greenscreen approach where the background is used as a reference image and people etc are added. If you want a good free video editor than use Shotgun https://www.shotcut.org/
>>
>>6231491
counting or not counting nerf violence
>>
File: DOUB0081.webm (5.85 MB, 1760x880)
5.85 MB
5.85 MB WEBM
>>
File: DOUB0082.webm (5.71 MB, 1760x880)
5.71 MB
5.71 MB WEBM
>>
File: DOUB0083.webm (5.69 MB, 1760x880)
5.69 MB
5.69 MB WEBM
>>
>>6231617
why do are the on teams
>>
>>6231617
Im surprised that more AI slop isn't like this. Just random cool shit.
>>
>>
>>
>>
File: bunwife.mp4 (2.95 MB, 1280x720)
2.95 MB
2.95 MB MP4
>>6231321
A lot of these gens make the audio waaay to clean, like bad ADR. A little ambient noise & reverb go a long way.
>>
File: DOUB0090.webm (5.82 MB, 1280x720)
5.82 MB
5.82 MB WEBM
>>6231620
are you having a stroke?

>>6231650
im more surprised at the boring ai slop here
>>
>>6231003
for true 4k quality
>>
>>6231675
My youtube algorithm is starting to be filled with stuff like this.

It might be a while before I can really generate deep rabbit holes myself. Entire shows and seasons. I imagine that is around 5 years down the line.
>>
>>6231688
you can do it now but it takes some work since you can only make a few scenes at a time.
>>
>>6231688
>>6231695
Anon is right

>>6231614
This is from some Chinese series called monster wars, it has 6 5minute episodes right now. It’s pretty damn good actually. I came across a lot of other really good series with refreshing plots. Ai is really awesome, only problem is that it takes a whole day to make a 5minute clip that’s watchable
>>
File: theessentials.mp4 (3.99 MB, 640x480)
3.99 MB
3.99 MB MP4
>>6231366
That's kinda the only way though. Longest I've seen where it's a single solid gen was 30s with wan3. No extends no stitching.
>>
File: white.mp4 (3.99 MB, 960x960)
3.99 MB
3.99 MB MP4
>>
>>6231722
lmfao
>>
>>
https://desuarchive.org/wsg/thread/6221745/#q6222215
https://desuarchive.org/wsg/thread/6221745/#q6222520
https://desuarchive.org/wsg/thread/6221745/#q6222541
https://desuarchive.org/wsg/thread/6221745/#q6225375
https://desuarchive.org/wsg/thread/6221745/#q6226195
https://desuarchive.org/wsg/thread/6221745/#q6226654
https://desuarchive.org/wsg/thread/6221745/#q6226657
https://desuarchive.org/wsg/thread/6221745/#q6227816
https://desuarchive.org/wsg/thread/6221745/#q6227826

4chan deleted the thread before I could save the clips. :(

Someone please repost any of these.
>>
>>
>>6231861
cool, prompt? i never tested japanese shows. what the right triggers?
>>
>>6231866
Generated the first frame with ChatGPT and had it form the prompt for me (based on some vague guidelines) to keep it accurate. Model is MiniMax H3. I had codex do some post-editing using Shotcut and Audacity.

Prompt is literally too long to paste, but you get the idea from below:

Animate this image into a 15-second authentic late-1990s / early-2000s Japanese TV variety-show segment.

Preserve the exact people, clothing, studio set, composition, Japanese TV graphics, 4:3 framing, and soft analog broadcast appearance from the source image. Peter Griffin is a proportionally realistic live-action human version of Peter Griffin, not cartoonish, grotesque, uncanny, or exaggerated. Do not change anyone's identity or appearance.

The scene should feel like a genuine spontaneous Japanese variety-show conversation, with constant natural activity rather than everyone posing for a photograph. The Japanese hosts and panelists talk over each other slightly, laugh, react, nod, gesture with their hands, lean toward Peter, look between Peter and each other, and behave like experienced Japanese television personalities. Peter is relaxed, funny and animated, gesturing while speaking fluent conversational Japanese.

0–5 seconds:
The panel is already in the middle of lively banter. Peter says something amusing in Japanese while smiling and gesturing casually. The Japanese panelists immediately react with laughter. The woman beside him laughs with her hand near her mouth, one male host leans forward laughing, and another host enthusiastically responds to Peter. There is natural conversational overlap and studio audience laughter.

Suggested natural Japanese chatter:
Peter: 「いや〜、日本のテレビってすごいね!」
Host: 「ピーターさん、楽しんでますね〜!」
Peter, laughing: 「最高ですよ!」

5–9 seconds:
As the laughter settles slightly, one Japanese male host excitedly asks Peter to perform his famous laugh.

etc...
>>
>>6231724
lol, that looks like 90s Clarence Thomas.
>>
>>6231871
thx,dude
>>
File: M86lweOyYevt8SEL.mp4 (2.04 MB, 1280x720)
2.04 MB
2.04 MB MP4
>>
File: never-forgetti.webm (5.35 MB, 1280x720)
5.35 MB
5.35 MB WEBM
I didn't have time to remake this one for the big 25 this year. Oh well.
>>
File: MiniMax_H3_00715_.mp4 (5.08 MB, 1184x896)
5.08 MB
5.08 MB MP4
>>
>>6231954
where was you when spaghetti never forgetti
>>
File: m2-res_720pz_4chanha.mp4 (2.54 MB, 722x720)
2.54 MB
2.54 MB MP4
>>
>>6231994
lol, I recognize the JAV scene this sampled.
>>
>>6231133
>What are they saying?
"This swimsuit is getting tighter"
"Mine too, I've had to buy so many"
>>
>>
>>
>>
File: EjxI6Pqj9mtHQSVg.mp4 (1.81 MB, 1280x720)
1.81 MB
1.81 MB MP4
>>
>>6232268
>>6232269
>>6232270
no celebs even with this 200b model?
>>
>>6232463
Jesus christ, man. Break yourself free from this celebrity worship bullshit.
>>
>>6232402
millenial corn dog slop
>>
>>6232467
are you blind? these threads are packed with copyrighted content. besides, my tastes aren't going to destroy this already corrupt world. it's already screwed. might as well have some fun, man
>>
>>6232491
Brother, what does copyright have to do with anything?
>>
File: sj_00044_compressed.mp4 (5.86 MB, 864x480)
5.86 MB
5.86 MB MP4
1/3
>>
>>6232515
no 2/3 or 3/3. computer crashed when rendering 3/3 and apparently I rendered 1/3 twice instead of 2/3. unfortunate
>>
>>6232402
Love how Omori doesn't want to be there.
>>
>>6230908
>>
File: sj_00046_compressed.mp4 (5.81 MB, 736x416)
5.81 MB
5.81 MB MP4
1/4
>>
File: sj_00047_compressed.mp4 (5.9 MB, 736x416)
5.9 MB
5.9 MB MP4
>>6232808
2/4
>>
File: sj_00048_compressed.mp4 (5.86 MB, 736x416)
5.86 MB
5.86 MB MP4
>>6232811
3/4
>>
File: sj_00049_compressed.mp4 (5.81 MB, 736x416)
5.81 MB
5.81 MB MP4
>>6232813
4/4

was supposed to be matrix anime style, but I need to rework the visual defs. still lots of prompt errors and hallucinations
>>
File: Feeling Baras's Anger.mp4 (3.01 MB, 1280x720)
3.01 MB
3.01 MB MP4
>>6230908
this seems like the "proper" place to ask, does anyone have those (early) renditions of YWNBARW by Joe Biden or the General from Rome Total War?
I'd be glad even with a vocaroo of them, especially the general one
been looking for close to a week and I feel like pic related
>>
>>6232814
this one was kinda kino
>>
File: Slaanesh-Lisa_00002_.mp4 (1.09 MB, 384x528)
1.09 MB
1.09 MB MP4
I made this based on a individual images a generated awhile ago. I thought I cloned her voice, but i messed it up. When i redid it with her voice cloned the performance wasn't as good. So I prefer this one.
https://files.catbox.moe/g6q7je.mp4
>>
>>
>>
>>
File: Slaanesh-Lisa_00026_.mp4 (880 KB, 384x528)
880 KB
880 KB MP4
Some select scenes WITH proper voice cloning. I think I needed a more energetic voice sample. The default voice had more room to emote
>>
File: Tng Hologram Mystery01.mp4 (5.81 MB, 304x304)
5.81 MB
5.81 MB MP4
>>
>>6232869
i watched the whole thing but dont understand the intended plot at all
>>
Took from r**dit.
>>
>>6232875
>plot
the reason is to be honest I genned and made up the "plot" as I went along based on the gens. So even from the first gen in the holodeck the only thing I had in mid was that he was going to create an street in England in the 1950's. So from then on it it just put together. As it was getting too long (in order to even post) it was kind of rushed but I had the idea of making it so another one could be made to explain basically what happened. As of yet I have no tried or bothered but may well make another 5 min video to explain it all. A better solution would of course be to get a storyline and stick to it, make each gen stick to a devised plot. I actually do think the holodeck has great potential because it can be anything, time travel is always a given premise because the holodeck "creates" any time period. it does take more effort to try to link together more gens, keep continuity as much as possible. Hopefully models in the future will make it easier, or maybe at least it could scan a video and use the info from it to continue it. We are not at levels were the model can create extended videos without lot's of editing and multiple gens. the secret would be using more closeups to get rid of the possibility of continuity errors in the background or simply going back to the same source image for the next gen, end frames of gens could be used if it's a static camera. But as soon as cameras are changed and you try to go back to before there will be continuity errors unless you use an end image that is as per the first image used.
>>
>>6230908
hey debo :)
>>
File: 6_6vgJSq7cZYgQox.mp4 (1.47 MB, 972x718)
1.47 MB
1.47 MB MP4
>>
>>6232887
>1:02
TWO SHELDOR? aaaaa
>>
>>6231668
yknow making them talk like people and being fucking annoying kinda ruins the appeal
>>
File: 2ZTue1buSnm8rtZe.mp4 (2.05 MB, 1280x720)
2.05 MB
2.05 MB MP4
>>6233073
I disagree, seeing the fantasy bunny wife in a normal-ass relationship makes it even more appealing to me. Perhaps it's just the novelty, idk, but by comparison all the happy sexy waifus like pic related just make me roll my eyes.
>>
File: 33zW2hYA9IejDaTP.mp4 (2.54 MB, 1280x720)
2.54 MB
2.54 MB MP4
>>
File: george floyd seinfeld.mp4 (2.39 MB, 1280x736)
2.39 MB
2.39 MB MP4
>>6231114
>>
These 10 clips were from the thread before the last.

https://desuarchive.org/wsg/thread/6221745/

Someone please repost any of these 10 clips, thank you.
>>
Also please repost these 2 vids, curious to see them.

https://desuarchive.org/wsg/thread/6221745/#q6228367
https://desuarchive.org/wsg/thread/6221745/#q6228370
>>
>>6233365
Bummer.
We need a buuro.
>>
>>6232877
I want it to be real.
>>
File: drum.webm (3.84 MB, 704x1280)
3.84 MB
3.84 MB WEBM
When's Samuel coming out with Sora 3
>>
>>6233431
they dont have the compute available. they just canceled the $200 subscription cuz they don't have enough compute
>>
File: 1KDau0a1L7z5UOX7.mp4 (1.76 MB, 1280x720)
1.76 MB
1.76 MB MP4
>>
>>6233433
>they canceled their highest revenue flow because they lack capital
>>
>>6233439
just because its the most expensive subscription doesn't mean its the highest revenue flow. if anything, the revenue-vs-compute ratio was likely the worst of all their products (minus free users), hence why they scrapped it. a $200 sub could easily burn thousands upon thousands of compute
>>
>>6233457
Cap em. Don't take away the tier.
>>
>>6233462
I'm assuming they thought capping the usage would create worse sentiment than just dumping the tier altogether. or maybe they weren't thinking at all, which would be normal for them
>>
>>
>>
>>
>>
1/4
>>
>>6233539
2/4
>>
>>6233540
3/4
>>
File: sj_00060_compressed.mp4 (5.81 MB, 736x416)
5.81 MB
5.81 MB MP4
>>6233541
4/4
>>
>>6230908
>>6230913
>>6233502
>>6233503

Do you have the one where batman says "the dark kyke is here!"?
>>
>>6231194
what string did you use for the retro psx visuals?
>>
>>6233085
can you please share the prompt for this? i want to try with some of my stuff o.o phenominal
>>
File: file.mp4 (1.87 MB, 864x480)
1.87 MB
1.87 MB MP4
Got a neat thing made for genning and wanted to share as well as a genned /vidya from it sharing the prompt in the .txt file because its hooge!
( prompt https://files.catbox.moe/rrttpr.txt / https://litter.catbox.moe/pd6xqtyx90iac6e8.txt prompt ) ( my custom node for doing this stuff with built in prompt bible and custom logic + 4 character reference https://files.catbox.moe/bfb9wv.zip / https://litter.catbox.moe/ic1jir47qrh224em.zip my custom node for doing this stuff with built in prompt bible and custom logic + 4 character reference ) ( my .json workflow drop-in to load it all up https://files.catbox.moe/fk1fra.json / https://litter.catbox.moe/5vlbph2i5mgb5l5w.json my .json workflow drop-in to load it all up)
>>
>>6233590
not mine
https://x.com/rarepinkieman



[Advertise on 4chan]

Delete Post: [File Only] Style:
[Disable Mobile View / Use Desktop Site]

[Enable Mobile View / Use Mobile Site]

All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.