I Ate Mushrooms

Introduction

I Ate Mushrooms is a image and noise generator based on a pretrained BIGGAN (https://github.com/huggingface/pytorch-pretrained-BigGAN). It can generate images and videos from 1000 different objects and living beings as well as interpolations and abstract variations of those images. To start, add the plugin to a solid layer.

There are several modes, „Generated“, „Neural Transition“, “Frame Interpolation” and “Text-to-Image”.

Generated

Ingredients

In this section you can define the content you want to display. Choose a number of classes and assign different classes and their respective influences on the image. This way you can mix different objects.

You can find a overview over the different classes here: https://www.youtube.com/watch?v=YY6LrQSxIbc

Noise

The noise that influences the generated image. “Generated” uses a precoded noise function and a seed, which is also loopable (every 360°)! “Custom Layer” lets you choose another layer which then feeds the image ai with noise.

Resolution

The output resolution of the model. The smaller the faster.

Variety (former Creativity), Content Multiplicator (former Empty Stomach), Content Addition (former Bright Lights), Noise Multiplicator (former Funky Music), Noise Addition (former Overtired)

Some random sliders that affect your trip through the ai’s space alot.

Display

Scaling

How the ai’s output is scaled on the effects layer.

Interpolation

Choose “Nearest” for a blocky appeareance, and “Bilinear” “and Bicubic” for a smooth image interpolation.

Use Superresolution

An ai based upsampling of the image. This together with model resolution “Low” can give good and fast results.

Text-to-Image

Use masks names to drive the image content and masks opacity to control its respective influence. Keyframing masks opacities creates interpolation. You can download a sample project here.

Noise

The noise that influences the generated image. “Generated” uses a precoded noise function and a seed, which is also loopable (every 360°)! “Custom Layer” lets you choose another layer which then feeds the image ai with noise.

Variety (former Creativity), Content Multiplicator (former Empty Stomach), Content Addition (former Bright Lights), Noise Multiplicator (former Funky Music), Noise Addition (former Overtired)

Some random sliders that affect your trip through the ai’s space alot.

Scaling

How the ai’s output is scaled on the effects layer.

Interpolation

Choose “Nearest” for a blocky appeareance, and “Bilinear” “and Bicubic” for a smooth image interpolation.

Use Superresolution

An ai based upsampling of the image. This together with model resolution “Low” can give good and fast results.

Automatic keying

Create a ai based key for the predominant object in the scene.

Add text input

Adds another mask which can be used to control the images content.

Backend & Performance

Hardware Acceleration (on GPU version of the plugin only)

Run calculations on the GPU. This will give massive speedups compared to CPU mode.

Lower Precision

Compute with reduced precision if possible. This can save up to half of the memory and give you some speedups.

Optimize for low VRAM

If you have not that much graphics card memory enabling this will try to optimize memory consumption.

Samples (not available for all settings)

The number of ai samples to calculate. This will improve the models accuracy.

Computation Tiles (not available for all settings)

Split the computation into several tiles. This can help if you run out of memory.