Help the Neural Models to better identify Structures/Textures

Hi :slight_smile:
In some Pictures the Neural Models don’t recognize the correct Stuctures in a Picture (for example some Grass-Structures, or water and so on)…

Would it be possible to make a new Tool in GigaPixel to “help” the neural Network to better identify the Content in the Picture to choose the right Textures/Structures? :wink:

Something like to draw an Area within the GigaPixel App and say “this is Water”, “this is Grass”, this is Wood" etc :slight_smile:
What do you think?

A.I. is just a Ad Word.

Machine Learning as it is also called is only a comparison between a and b.

If the water is to be specifically improved, you need the right water for every conceivable situation, at every time of day and light angle.

The same applies to everything else.

Up to now, it has worked well below 2x, but above that, the model has to guess from experience what could be at the new pixel.

So all in all, quite a lot of work.