FUQ (Frequently Unasked Questions)

What is SPAILER?

Spailer is an image generator that uses artificial intelligence to create frames from invented, never filmed movies. Spailer tracks and searches cinematographic data of real movies in open repositories (synopsis, quotes, plots, slogans, genres,…) that it stores to later mix and use in the generation of images through diffusion models such as MidJourney and Stable Diffusion. The database includes abandoned movies, interrupted projects or upcoming releases. No “initial images” nor “influence images” has been used, only text prompts.

The Spailer software is programmed to mix texts, quotes and descriptions from different films in order to create images of impossible movies; the information in the database is constantly updated and includes not only data from real films data but also from abandoned movies, interrupted projects or upcoming premieres.

Spailer are photographs never taken from movies never filmed.

What´s the license for the images generated by Spailer?

All images have a Creative Commons License:  Attribution-NonCommercial-ShareAlike 3.0 Unported (CC BY-NC-SA 3.0) You are free to:

  • Adapt: remix, transform, and build upon the material
  • The licensor cannot revoke these freedoms as long as you follow the license terms.
 
Under the following terms:
 
  • ShareAlike: If you remix, transform, or build upon the material, you must distribute your contributions under the same license as the original.
  • No additional restrictions: You may not apply legal terms or technological measures that legally restrict others from doing anything the license permits.

Is Spailer using copyrighted material?

All data introduced in Disco Diffusion and MidJourney are descriptions, quotes, taglines and keywords about movies in text format, no copyrighted images has been entered as an “init image” or “inspirational image”. Spailer uses MidJourney and Stable Diffusion as diffusion models. Stable Diffusion has been trained on pairs of images and captions taken from LAION-5B dataset. You can consult the licenses and dataset here. For MidJourney, you can check their page with the info about Copyright.

How to access the private area?

The private area is now on a Closed Beta. All visitors can see the images at a base resolution of (at least) 512 pixels in height but, from the moment you become a member, you could enter the private area where you´ll have access to:

  • High resolution images for printing.
  • Interact with the system by voting for some of the words / concepts to be included in the commands for the image generator.
  • Generate a unique variation of the original images just for you.

How to help?

By becoming a Patreon you will help to expand the computational power necessary for the generation of images. (+ Info in the coming weeks)

Are the images edited in any way?

No, all published images have no edition / postprocessing. If  in the future any image is edited, it will have an “edited” tag.

What´s the hardware used to generate the images?

  • GPUs / TPUs (Google Colab PRO):
    • Tesla T4 16GB GPU
    • NVIDIA A100 SXM4 40 GB

Is there censured words?

Yes an No. No with the modified version of Stable Diffussion that runs on a local computer and in Google Colab PRO. And yes when using Midjourney as there are censured words / phrases when using the Discord server. You can check out a (incomplete) list of censured words in this Google sheet: https://docs.google.com/spreadsheets/d/13r7AB6WOPhrt3sZIbwi54NOIRoMVC0pTHKaSfE_oJA4/edit#gid=377237055.

How is Spailer programmed?

The Spailer system is completely programmed in Python and a bunch of libraries. Spailer includes diffusion models for image generation: a customized version of Stable Diffusion and MidJourney.

How is the aspect ratio?

In a next phase, aspect ratio will change per photo. Nowadays the aspect ratio is “fixed”.

Can I access the database?

Not at this time, it will be possible in the future with a more stable system.

Who runs “Spailer”?

Spailer is an art project by Rubén Ares.

Sorry, the comment form is closed at this time.