Hello!

    If this is your first stab at Machine Learning art, I truly welcome you and hope this platform will accelerate your journey!

    Are you returning from the beaten lands of Jupyter Notebooks? I hope you will find this space a haven of calm, joy and experimentation.

    No idea what I'm talking about? Read on!

    What is this?

    In short, Pyttipanna is an interface for Pytti 5. It allows you to structure, narrate and experiment with prompts for video creation. Pytti is a framework for creating and rendering videos using Machine Learning models.

    You don't need to know Machine Learning to use Pytti. The author @sportsracer48 has done an amazing job at making the tool extremely accessible. You do however need a lot of patience and self control if you want to create anything more than simple videos. Google Colab, which Pytti runs on, is not made for complicated interfaces.

    That's where Pyttipanna comes in: On its own Pyttipanna can't create videos, but together with Pytti it enables timeline-like creation of Machine Learning videos.

    Getting started

    1. Pick your device.

      While Pyttipanna is somewhat optimized for Mobile screens, it is definitely easier to work with on desktop or tablets. Google Colab is also quite unpredictable on mobile and in general you want to render your project on a device that can be kept awake for hours.

      Secondly: Everything you do in Pyttipanna is automatically stored in your browser so you can safely close the tool at any time! Hurray! But that also suggests thinking twice about which device and browser you want to keep your projects on as you (for now) can't move them.

    2. Pyttipanna is free. Pytti isn't.

      First thing you want to do is subscribe to Sportsracer48 on Patreon for $5 per month and grab the link to the Google Colab for Pytti 5. You can cancel your subscription after you have the link, should you want to. (But you'll miss out on when Pytti 6 comes out)

    3. Now you are ready to create using Pyttipanna:

      Create a project ━ Add scenes ━ Add prompts that describe the scenes ━ Tweak the camera movements using the graphs. Etc.

    4. Congrats, you're ready to render!

      Press on "Copy to Pytti" ━ Open your Colab link for Pytti ━ Scroll down to "2.2 Load settings (optional)" ━ Paste in the string you copied from Pyttipanna ━ In the main menu, click on "Runtime > Run All" ━ When prompted, log in with your Google account ━ Wait. (Keep your pc awake and don't close your browser!)

    5. Tadaa! Your frames for your videos are slowly being render out into your Google Drive in the folder "pytti_test > Images out". Download and combine them to a video using a tool of your preference.

    If you have any comments or have experienced any bugs, please don't hesitate on reaching out to me :) You can find me on Twitter @_staus

    What now?

    Experiment! If you want more info on the input fields, you can for now read about them in the top of the Pytti colab. At some point I'll add info to each field here in Pyttipanna, but for now Pyttis intro is your friend.

    Tip: You might want it to render faster. Consider subscribing for Colab Pro. It's $10 per month and totally worth it if you render a lot! Before clicking "Runtime > Run All" you just need to go to "Runtime > Change runtime type" and select "High Ram" for Runtime shape to get a faster GPU.

    Also consider joining the VQLIPSE discord. We're an energetic group of creaters exploring Pytti and much more. It's a great place to ask for help and don't worry... NFT talk is kept to a minimum. You can also contact me in there @staus!

    What does Alpha mean?

    Pyttipanna is still in an alpha stage, so anything can happen. Therefore I'll not take responsible if you loose your work. Luckily, whenever you do a render using Pytti, all your settings are saved as a text file next to your frames in Google Drive. That should give you some comfort :)