Prototype I | Deep-Picture Converter

Prototype No. I - finished on 27th of October, 2021

The following snapshots and example outputs are a based on the first Deep-Picture converter, that Nilesh from India has developed on 27th of October, 2021. The converter can deal with pictures up to 700 Megapixels (depends on the size of the mother-picture and the adjustments made) with an random access memory of 8GB needed. The raster-based mother-pictures are transformed into both an svg-file, consisting out of diffently sized tiny squares and a png-file, composited in the same way. Untill now, only the linear pixel growth is possible, that means the pixels are growing from one row to the next by an absolute value (e.g. ''2'')

The interface

This early version is already easy to deal with and pretty attractive. The user has got the opportunity to choose a picture, determine the size and location of the highest resoluted area and adjust the pixel growth. Also it is already possible to choose between a random colourization for a fast conversion and a most frequent colourization, to receive the best result.

How the center is selected

To select wich area should be kept in original high resolution, the user has to select a center by actively spotting with the computer-mouse. A red square occurs, which can be saved through the ''Save''-button.

The application in action

After a picture is loaded and all preadjustments have been made, the ''Convert''-button can be pressed, to start the transformation. At first, the net of the new pixel-structure is calculated, which is shown live in a window that automatically opens up when running the application. A loading bar shows the progress of net creation, another the colour generation. After conversion the in- and output-file are opening up automatically.

The output

After conversion three files occure in a special output folder - a png-file, showing the structure of the net, a svg-file, showing the converted file by saving each square seperately and a png-result, to easily deal with the converted result.

And for the example with the stadium...

Here you can see the converted, above mentioned output-file. Of course one could have choosen a center with a little more sense.