I did a search within the forums regarding this claim. Is there any literature regarding how this is achieved? I heard through a recent Twitch broadcast a mention that is write instructions rather than pixel data or something along those lines?
Would just love to read any specifics on this as I am embarking on a project that would benefit from this.
Thanks!
Hi,
I will check to see if we have specific documentation. The savings can vary depending on how the substances are created. If the substance is pure procedural the savings will be very high, but if there are embedded bitmaps, the bitmap size will effect the overall size of the published substance.
With the Atlantis demo which showed a 97% reduction provided a comparison of the size of the sbsar files versus the storage size of the maps at a given resolution.
The reduction is achieved in that the procedural substances do not contain pixel data. They only contain information the substance engines uses to generate the pixel data when called to do so. Nothing is rasterized until the Substance Engine is requested to generate textures from the substance file.
I will check to see how the information in stored in the sbsar file. The main thing to note is that using completely procedural substances will be very lightweight. When you embed bitmaps into your substance, the bitmap will then begin to raise the size as it will then be storing actual pixel data.
Cheers,
Wes