|
Small Strange Bird posted:A dumb joke from my childhood finally becomes fake-real: the justly forgotten yet unaccountably long-running cop show, Hodgson & Tooler! I love this, the only thing your background is missing is a reference to a final reveal that the whole show was a dream by an autistic boy like St Elsewhere.
|
# ? Oct 23, 2023 23:52 |
|
|
# ? May 30, 2024 10:48 |
|
code:
code:
Good Sphere fucked around with this message at 00:32 on Oct 24, 2023 |
# ? Oct 24, 2023 00:08 |
|
Good Sphere posted:
Unsure if this looks fun or terrifying. I know I'd probably die in it, but that doesn't sway me either way. Some weird environments:
|
# ? Oct 24, 2023 00:34 |
|
I like typing in gibberish to see what the AI comes up with. Apparently, this is what a flibbered sklipapoot dancing with a ruppened chost looks like. 80s Courtroom drama entitled "Not Without My Legbones" with a crusty Hibbityjibs and Eustice Carol dressed as a smork. A new hit action movie starring Hendall Fishersnitch and a crumpled up Cookychoo as they take on the army of the Blorkborks 80s kids cartoon entitled Muscleman Man and the Defenders of the pile of garbage at the end of the streetcorner that no one will clean up
|
# ? Oct 24, 2023 01:23 |
|
Small Strange Bird posted:A dumb joke from my childhood finally becomes fake-real: the justly forgotten yet unaccountably long-running cop show, Hodgson & Tooler! Don't forget the PC game that came out way too late:
|
# ? Oct 24, 2023 02:55 |
|
WhiteHowler posted:Every so often I'll feed a random song lyric into an AI image generator and see what pops out. 'be quick or be dead' gave me a speed-skating Eddie And then whatever the hell this is:
|
# ? Oct 24, 2023 02:59 |
|
|
# ? Oct 24, 2023 03:47 |
|
Anyone know anything about making a model of your own voice? Is there something like SD I can run myself? I have a progressive disability and it's starting to effect my speech. I have a youtube project I'm working on with a friend but I have to solve this first.
|
# ? Oct 24, 2023 04:16 |
|
Spoke Lee posted:Anyone know anything about making a model of your own voice? Is there something like SD I can run myself? I have a progressive disability and it's starting to effect my speech. I have a youtube project I'm working on with a friend but I have to solve this first. If you want open source, you could experiment with TortoiseTTS, but if you want to save time I'd try with elevenlabs first to see if voice cloning fits your purpose.
|
# ? Oct 24, 2023 04:42 |
|
|
# ? Oct 24, 2023 05:37 |
|
Gashroom posted:https://elevenlabs.io/voice-cloning is probably one of the better ones... not free, but it's cheap to try out with the starter tier. Thank you so much.
|
# ? Oct 24, 2023 05:43 |
|
Spoke Lee posted:Anyone know anything about making a model of your own voice? Is there something like SD I can run myself? I have a progressive disability and it's starting to effect my speech. I have a youtube project I'm working on with a friend but I have to solve this first. Voice cloning: https://github.com/serp-ai/bark-with-voice-clone Singing: https://github.com/svc-develop-team/so-vits-svc Elevenlabs is still the gold standard in quality but the above links can be run free and locally like SD can. I still haven't messed with audio gen as I'm too busy trying to merge chatbots and SD models together lol.
|
# ? Oct 24, 2023 05:50 |
|
Are there any live models you can use with your own clone? Like a voice changer.
|
# ? Oct 24, 2023 06:51 |
|
|
# ? Oct 24, 2023 07:11 |
|
Good Sphere posted:
I was trying to recreate a scene from an old game and Bing wasn't having it because there was a wounded person in it. So I changed the prompt to 'an actor pretending to be injured covered in fake blood', and through it went!
|
# ? Oct 24, 2023 08:03 |
|
Spoke Lee posted:Anyone know anything about making a model of your own voice? Is there something like SD I can run myself? I have a progressive disability and it's starting to effect my speech. I have a youtube project I'm working on with a friend but I have to solve this first. I've mentioned the free open source Pinokio app before with its voice manipulation plugins, though it's probably dissimilar to how kohya_ss trains on top of existing SD checkpoints, so you'll have to read the docs. Regardless, you can get them from the app's Discovery tab and see if they help with your project. KwegiboHB posted:
Speaking of advanced natural language prompting (for animating in Stable Diffusion), I ended up cramming a huge list of D&D wildcards into ComfyUI's ImpactWildcardProcessor node to try and worldbuild a bootleg superverse, and looks like it works p alright as a random character generator for fantasy settings: • Svelte Classic action villain, Mind control aura, wearing Indigo Nano-fiber Camouflage Suit, full body, from front, doing Fireballs • Solid Collapsing male action hero, Space-time continuum aura, wearing Phthalo Turquoise Plasma-lace Energy Gloves, full body, from front, doing Lightning Bolts • Buoyant Vaporized action villain, Reality Fragmentation aura, wearing Dull Carbon-fiber Tactical Armor, full body, from front, doing Creating Illusions • Balanced Air Pressure male action hero, sprites and elves aura, wearing Tan Carbon-fiber Tactical Armor, full body, from front, doing Healing Waters • Statuesque Gripping male action villain, death aura, wearing Alizarin Crimson Cybernetic Neural Implant, full body, from front, doing Weather Manipulation • Well-defined 3D Graphics action hero, Quantum Thunder aura, wearing Malachite Green Cybernetic Implant, full body, from front, doing metal magic
|
# ? Oct 24, 2023 08:53 |
|
KakerMix posted:Don't forget the PC game that came out way too late: Maybe I should do Hodgson & Tooler: The Next Generation!
|
# ? Oct 24, 2023 11:01 |
|
https://twitter.com/DiffusionPics/status/1715441914848383432
|
# ? Oct 24, 2023 11:58 |
|
code:
code:
Good Sphere fucked around with this message at 12:53 on Oct 24, 2023 |
# ? Oct 24, 2023 12:01 |
|
Good Sphere posted:
Thanks. I hate it. also it is pretty cool Bored fucked around with this message at 13:10 on Oct 24, 2023 |
# ? Oct 24, 2023 12:52 |
|
Oops. Sorry, heavier than it looked! ... This isn't going to end well. A large Minotaur in a china shop. He is looking at a broken item on the floor, the shopkeeper is appalled
|
# ? Oct 24, 2023 12:56 |
|
code:
code:
|
# ? Oct 24, 2023 13:11 |
PaleFigure posted:Oops. Sorry, heavier than it looked! Made me Good work! An apple a day keeps the doctor away. So here are some Doctors upset about apples. Or judging me for my life choices.
|
|
# ? Oct 24, 2023 13:17 |
|
Spoke Lee posted:Are there any live models you can use with your own clone? Like a voice changer. Not as far as I know. Eleven Labs keeps teasing that, but it's not live.
|
# ? Oct 24, 2023 13:53 |
|
Trying to make some christmas cards
|
# ? Oct 24, 2023 16:57 |
|
There's something wrong with these christmas cards
|
# ? Oct 24, 2023 17:12 |
|
I was making a picture of a cat using tinfoil and a blowtorch to cook various things and well your post got me thinking.. What is Santa and Krampus e: oh maybe you want a cat too
|
# ? Oct 24, 2023 17:12 |
|
LazyDivey posted:I like typing in gibberish to see what the AI comes up with. Apparently, this is what a flibbered sklipapoot dancing with a ruppened chost looks like. That's really fun Hile Joilt and palps awulty. The pealt pollop grew boiler in warm vuilts, Groalt a chully Vast. Warm walt and Builty wealt poap.. HURH UH... jeal groilt w'i'h hollop Chealp! Chealp-! Moalt gugs cholps; chags; pkoik (BRELL.)
|
# ? Oct 24, 2023 17:34 |
|
Spoke Lee posted:Are there any live models you can use with your own clone? Like a voice changer. The gold standard right now for self hosted and self trained models is RVC2. It's near real time but can go further than older voice changers by changing speaking patterns/pitch and mapping accents more realistically. It's what a lot of the "AI" voice-overs are using right now. https://github.com/IAHispano/Applio-RVC-Fork is, as far as I know, the most user friendly self run one around right now. Example of it in use: https://www.youtube.com/watch?v=HamZ8rSsyhI&t=90s
|
# ? Oct 24, 2023 18:12 |
|
Spoke Lee posted:Are there any live models you can use with your own clone? Like a voice changer. It takes a minute or two to turn text into a 1024x1024 image. Changing voice into voice by passing it through a model , even at 32kbps, is an exponentially larger data rate if you want output voice to come out within a sub-second of the input. Tldr: not yet. Stick with voice changer apps now. Objective Action posted:The gold standard right now for self hosted and self trained models is RVC2. It's near real time but can go further than older voice changers by changing speaking patterns/pitch and mapping accents more realistically. It's what a lot of the "AI" voice-overs are using right now. EVIL Gibson fucked around with this message at 18:17 on Oct 24, 2023 |
# ? Oct 24, 2023 18:13 |
|
You actually can get it down to near real time but you need some technical knowhow right now and the only pre-built solution I know of is mostly in Japanese. It's not going to be as good as the version where you upload recorded segments because you can't tell it to try again or do pre-analysis for pitch recovery but you can absolutely run the same RVC model in this mode and get down to ~25ms delay with a CUDA card or ~100ms on a good CPU.
|
# ? Oct 24, 2023 18:23 |
|
Objective Action posted:You actually can get it down to near real time but you need some technical knowhow right now and the only pre-built solution I know of is mostly in Japanese. The model would be crazy since to be better you would need to pull more frequencies from the input voice. To make it better, pull more frequencies and there is no stopping since a waveform is near frequencies. I've heard those Japanese voice changers and I am pretty sure the sing song voices is not what the person was asking for unless they really want to sound like a worse Microsoft Sam lol. There ARE a million of them on Japanese youtube but they all sound very much the same with the only difference being the graphic model of the avatar. I will up your Neco-Arc with cover most people don't recognize where her voice comes in because it is so suited for screaming Chop Suey
|
# ? Oct 24, 2023 18:30 |
|
Am I in jail?
|
# ? Oct 24, 2023 18:33 |
|
EVIL Gibson posted:The model would be crazy since to be better you would need to pull more frequencies from the input voice. To make it better, pull more frequencies and there is no stopping since a waveform is near frequencies. Agreed that its not perfect but the RVC implementation really is way better than the older ones if you use a good model. This video https://www.youtube.com/watch?v=pHhjg2JwdPI&t=373s shows the delay and has both a robotty anime model and an actually decent model in it. Again though the realtime version can only do so much on pitch adjustment and the accent stuff basically gets dropped altogether because the time window just isn't long enough. Not to spend too much time on Neco Arc but Chop Suey was pretty good! I also like Paint It Black, very funny for reasons I refuse to introspect on at all https://www.youtube.com/watch?v=I332pv84RBQ.
|
# ? Oct 24, 2023 18:39 |
|
As far as gibberish goes, i am continuing my exploration of what it does with as little prompting as possible "The" "A" "This" "That" I think this probably gives a bit of insight on what kind of images it is tuned to produce, in terms of style and composition
|
# ? Oct 24, 2023 18:48 |
|
Fine quote:To be, or not to be, that is the question: Whether 'tis nobler in the mind to suffer The slings and arrows of outrageous fortune, Or to take Arms against a Sea of troubles, And by opposing end them: to die, to sleep No more; and by a sleep, to say we end The heart-ache, and the thousand natural shocks That Flesh is heir to? 'Tis a consummation Devoutly to be wished. To die, to sleep, To sleep, perchance to Dream; aye, there's the rub, For in that sleep of death, what dreams m Oh this is interesting. It looks like DALL-E 3 thinks that coward is usually related to lions. quote:Thus conscience does make cowards of us all, Diana Wynne Jones posted:Imagination doesn’t just mean making things up. It means thinking things through, solving them, or hoping to do so, and being just distant enough to be able to laugh at things that are normally painful. Head teachers would call this escapism, but they would be entirely wrong. quote:A world in which there are monsters, and ghosts, and things that want to steal your heart is a world in which there are angels, and dreams and a world in which there is hope. in a comic book style quote:an opossum in the style of video game "O-KAMI" Bored fucked around with this message at 19:54 on Oct 24, 2023 |
# ? Oct 24, 2023 19:03 |
|
what happens if we just send fladjistant diswalight ? can we do that for the first prompt and the other 3 try using it as various parts like maybe imply it's a vehicle in one or a type of food in another but the first one should just be the phrase GPT+DALLE fladjistant diswalight Illustration of a sleek futuristic vehicle named the 'Fladjistant Diswalight', zooming across a neon-lit cityscape. Cartoon drawing of a gourmet dish presented in a high-end restaurant, known as the 'Fladjistant Diswalight Delicacy', garnished with unique ingredients. Watercolor painting of a mystical forest where creatures gather around a glowing plant called the 'Fladjistant Diswalight', revered for its magical properties.
|
# ? Oct 24, 2023 19:09 |
|
|
# ? Oct 24, 2023 19:27 |
|
Someone posted this in the Elon thread and I think it goes here too
|
# ? Oct 24, 2023 19:34 |
|
|
# ? May 30, 2024 10:48 |
|
|
# ? Oct 25, 2023 00:58 |