Unlike other applications, Toon.Pics uses training data created with 3D graphics programs specifically for this purpose.
We rely on applications from Reallusion.com,
such as IClone and Character Creator, which we have been working with for several years.
We are constantly working on new data, to improve the training process by using possibly less data to gain optimized results.
AI typically requires a large amount of data to produce a suitable result -
we try to distinguish between the amount of data and the quality of the data needed.
The training software is based on neural network applications, based of some kind of trial and error. Starting with weights and vectors - make a prediction
and compare it to a "desired" output - adjusting weights and vectors, as long as the predictions difer from the desired output.
These applications are written in python and run on a GPU powered cloudserver with Keras, PyTorch and TensorFlow frameworks.
Trained data then is transferred to the Toon.Pics webserver, to meet the needs of the intended applications. AI is special software - and there is no
software, that is able to satisfy all the expected needs - what depends on the expectations too.
demo images used are from www.pexels.com - "The best free stock photos, royalty free images & videos shared by creators"
that is how they describe themselves
here you find a list of creators whose photos can be seen as samples on toon.pics - no specific order:
@inspiredimages | @alexgtacar-745150/ | @thgusstavo/ |
@bess-hamiti-83687 | @shvetsa | @mikebirdy |
@hikaique | @brett-sayles | @lysenkow |
@eberhardgross | @jaime-reimer-1376930 | @gilberto-reyes-259461/ |
@minan1398 | @juanlaurio | @nastyasensei-66707/ |
@kevinbidwell | @italo-melo-881954 | @enric-cruz-lopez |
@suzyhazelwood | @brettjordan | @pixabay |
@toni | @elenarubtso/ | @incrediblerafa |
@chloekalaartist | @myca/ | @flavio-augusto-918711/ |
@david-florin-1326024/ | @divinetechygirl/ | @francesco-ungaro/ |