Log in

Lisalotte Tscherteu

  • Tutorial 9.1. Maieutics Maxxing

    By

    Emma Breitenecker

    ,

    Florian Krispler

    ,

    Ylvie Moschitz

    ,

    Lisalotte Tscherteu

    and

    Abrar Aitounejjar

    We finalized the script:

    Made a timeline & a shotlist:

    Trial and Error with 10 and more different AI models.

    Categories of competences we need from an AI:

    • Video generation
    • with audio
    • more than 5s video content
    • real-looking, camera style
    • Character consistency over several clips
    • Voice consistency over several clips
    • First frame/last frame, add to former scenes
    • not super expensive

    We tried:

    • Sora 2 (Open AI)
    • MindVideo (Pro subsciption)
    • Gemini (Google)
    • Flow (Google)
    • Comfy UI
    • OpenArt
    • Dzine
    • Digen AI
    • Several Hugging face models
    • Chat GPT
    • Adobe suite

    SOME of the bad results:

    And some good ones:

    Sound mood board:

    A lot of complicated workflows:

    We were at the coding lab

    Current expenses for subscriptions:

    • MindVideo: 26€
    • OpenArt: 7€
    • Dzine: 14€

    We still need a 54€ subscription for when we run out of free credits in Flow.

    Questions:

    • Does the film need to contain our names/DI etc.?
    • How much of the finals presentation is about the process?

  • feedback december 5

    By

    Lisalotte Tscherteu

    ,

    Florian Krispler

    ,

    Ylvie Moschitz

    ,

    Emma Breitenecker

    and

    Abrar Aitounejjar

    Storyline

    premise: the internet is dead -> AI doesn’t work

    Due to the overflow of false information AI is not able to tell the truth from fake news. Also everything one can find on the internet there is no way of verifying if it is true or not.

    show this state with: news, AI reading it

    a. the gov decides every AI will be m.AI (maybe beta testers)

    b. people develop this and there is no choice but to use it because everything else is trash (no alternative)

    c. the AI itself notices that users are more and more unhappy with its answer so it stops answering and starts questioning. eg.: FLO

    -> in comes m.AI (maiuticsAI)

    our chracters use it but are rather frustrated – they then try to turn back to the internet but are even more frustrated.

    -> second try with m.AI

    this time our characters really try and see the benefits and get a kick out of their learing expirience with m.AI.

    -> resoltuion ?

    a. the POV of the single screens merge – they stop talking to their devices and turn to humans for answers.

    What kind of AI could help bring people together?

    What kind of AI could make people see the value of learning and understanding?

    What kind of AI could foster critical thinking?

    What alternatives will we adapt to, in case of the death of the internet?

    Visuals

    our characters will be shown from the devices cameras perspective.

    Characters

    Hippies that revert to also only talking in questions

    Strongly religious people that cannot stand the AI questioning their belief/god

    Children Why? why? why?

    Seniors not affine with any kind of technology

    Single-parents

    People falling into an existential crisis (maybe even suicidal)

    Rebels

    impatient/unwilling people that just want answers

    Positive responders super efficient, super creative

    >> next steps

    storyboard

    write some scenes -> frustration scenes

    WHO are our characters?

    find refined aesthetic references