• úvod
  • témata
  • události
  • tržiště
  • diskuze
  • nástěnka
  • přihlásit
    registrace
    ztracené heslo?
    KERRAYAI Art - Tvorba za pomocí umělé inteligence: Midjourney, Nanobanana, ComfyUI, Suno
    MICHIGAN
    MICHIGAN --- ---
    ARAON: tohle budou muset rozhodnout soudy. v podstate se jen alibisticky vyvlikli z odpovednosti. navic neni pravda, ze to uplne odmitli - ten copyright ji nechali, jenom na celkovy dilo jako kompilaci.
    ta zenska v podstate fotila sama sebe a pres remix tusim genrovala ty obrazky. i kdyby pouzivala jmena autoru (coz nejsou nic jineho nez vektory), tak vsem musi byt jasny, ze u soudu by ten zakaz nemel sanci projit. v mezich soucasnych zakonu na tom neni nic spatne.
    nesmime zapominat, ze copyright maji i fotografie, a ze zamer a kompozice autora, i vlastni upravy, jsou taky soucasti copyrightu.
    ARAON
    ARAON --- ---
    "We conclude that Ms. Kashtanova is the author of the Work’s text as well as the selection, coordination, and arrangement of the Work’s written and visual elements. That authorship is protected by copyright. However, as discussed below, the images in the Work that were generated by the Midjourney technology are not the product of human authorship."
    https://twitter.com/MAiJiNTHEARTIST/status/1628505937580552193
    SLL_QUY
    SLL_QUY --- ---
    Tutorial: Creating an AI Chimera Using Stable Diffusion
    https://www.youtube.com/watch?v=wj6xnIXK_BI
    KERRAY
    KERRAY --- ---
    #nástroj #novinky

    Discord
    https://discord.com/channels/662267976984297473/1037743153471553618/1075517842180558888

    Midjourney OFFICE HOURS NOTES - February 15, 2023

    This is the summary of what they're working on and what they're thinking about that David shared at the top of the call.

    Working on v5 – a lot of issues still in v4, apologies, to be honest a lot of the issues that remain in v4 are mysteries as to why they're there, but v5 isn't doing them so we're focusing on that.

    Sharper, higher resolution, better job with realistic imagery, does hands significantly better. Might launch with no upscaler but default resolution matching v4 upscaler and then add upscaler to 2k, 4k, 8k? Final upscaler target resolution for v5 TBD

    Tiling seems to be working, may release to v4 but otherwise v5

    Experimenting more with inpainting/outpainting and realtime drawing thingies.

    Outpainting is working pretty well so that's probably going into v5 as "zoom out" but inpainting isn't working well yet, more time needed due to the way MJ differs from other AI image generation systems.

    Cool real-time applications of drawing, will probably demo at some point but it'll take a while to ship.

    New rating systems coming soon. Will do a ratings system test on v4 and will use it as a "v4 rating contest" to see which the top images are across all of v4, possible prizes involved.

    Then will do ranking on v5 images. At some point will deploy the new ranking system so that it's linked to the daily theme channel and use that to identify best images and prizes may come along there too.

    Eventually will have a free tier for the ranking, but that's not coming for a while still. .
    KERRAY
    KERRAY --- ---
    KERRAY
    KERRAY --- ---
    #obrázek #text
    ChatGPT dostal za úkol vymyslet zadání, další AI to pak spočítala

    Reddit - Dive into anything
    https://www.reddit.com/r/aiArt/comments/11793sh/comment/j9andjg/
    KERRAY
    KERRAY --- ---
    #obrázek
    https://twitter.com/DSzymborski/status/1627038833455038469
    HITOMI
    HITOMI --- ---
    KERRAY: v tom videu, co jsem postovala nedavno HITOMI: je to prave hezky vysvetleny :)
    KERRAY
    KERRAY --- ---
    a vůbec, ControlNet pro Stable Diffusion je dost cool #nástroj, ke kterému vzniká spousta cool doplňků, a kdo má grafiku se 6 a víc GB VRAM, tak vůbec není těžký to rozjet lokálně...

    https://www.reddit.com/r/StableDiffusion/comments/1169ikl/new_extension_for_a1111control_net_openpose_editor/

    ARAON
    ARAON --- ---
    https://twitter.com/bilawalsidhu/status/1627495393771630594
    DRAGON
    DRAGON --- ---
    QWWERTY: to je hodně dobrý
    QWWERTY
    QWWERTY --- ---
    Futurama as a dark 80s movie! - Album on Imgur
    https://imgur.com/gallery/FHF1e9p

    QWWERTY
    QWWERTY --- ---
    KERRAY
    KERRAY --- ---
    #hry #design #nástroj
    https://twitter.com/rowancheung/status/1626988760637243393
    SPLNYX
    SPLNYX --- ---
    Tohle mi teď vyplivl deepdreamgenerator.com.

    QWWERTY
    QWWERTY --- ---
    HITOMI
    HITOMI --- ---
    Tohle je pro SD docela velký game-changer :)
    #video #nástroj

    NEXT-GEN NEW IMG2IMG In Stable Diffusion! This Is TRULY INCREDIBLE!
    https://youtu.be/OxFcIv8Gq8o
    CRAZY_IVAN
    CRAZY_IVAN --- ---
    KERRAY: sorry, neznal jsem to auditko, díky :)
    KERRAY
    KERRAY --- ---
    spíš bych to řešil v [Artificial Intelligence AI], ale je to teda dost masakr, fakt jsem zvědav, k čemu to povede
    #text #psycho
    [KERRAY @ Artificial Intelligence AI]
    CRAZY_IVAN
    CRAZY_IVAN --- ---
    UNTOY
    UNTOY --- ---
    QWWERTY: https://twitter.com/bristowbailey/status/1625165718340640769
    Criminals will start wearing extra prosthetic fingers to make surveillance footage look like it's AI generated and thus inadmissible as evidence.
    MICHIGAN
    MICHIGAN --- ---
    SLL_QUY: mno, jenze ten glaze vezme zdrojovej obrazek, projede ho skrz model natrenovanej na posun stylu a vyslednej obrazek pak aplikuje na zdrojevej art v pomeru plus minus 95:5.
    prijde mi pomerne vtipny, ze lidi s odporem k AI budou pouzivat AI model pro ochranu svych vlastnich obrazku pred AI :)

    jinak uspech cele tehle metody bude taky dost zaviset na kvalite toho jejich predtrenovanyho modelu. pokud to testujou na sto obrazich ctyr ruznych stylu, tak to rozhodne neznamena, ze to bude perfektne fungovat i na vsechny ostatni styly. spis bych se teda dost divil.
    navic 5% odchylka muze lehce vyustit v dost nepekny artefakty.
    SLL_QUY
    SLL_QUY --- ---
    MICHIGAN: Nejsem odborník ale asi bych úplně netvrdil že to nepůjde, tady o tom třeba mluví dost zevrubně a různé formy data poisioningu vypadají jako dost reálná věc.

    "So for example, the canonical, like adversarial example type is you have an image, you add really small perturbations, changes to the image, it can be so settled at to humanize, it's hard to is even imperceptible, imperceptible to human eyes. But for the uh for the machine learning system then the one without the perturbation, the machining system can give the wrong, can give the correct classification for example, but for the perturbed division, the machine learning system will give a completely wrong classification And in a targeted attack, the machining system can even give the the wrong answer"

    Dawn Song: Adversarial Machine Learning and Computer Security | Lex Fridman Podcast #95
    https://youtu.be/HhY95m-WD_E?t=1045
    Kliknutím sem můžete změnit nastavení reklam