Nexture takes care of the tedious task of adding the final level of detail of textures. By synthesizing fine details onto a large-scale image, it helps texture artists reach desired quality, faster.
Displacement map synthesized with pattern s6 (scaling factor: 0.4)
Nexture combines a state-of-the-art artificial neural network with a custom image synthesis algorithm, to transfer details from a reference pattern bank. This dataset was captured on actual human skin by us at Cronobo. Thanks to this unique approach, generated images maintain the structure of the user's image while having a lot more visual details. This results in a more crisp and more believable 3D render.
While most IA-based image synthesis algorithms are limited to low resolution images, Nexture on the other hand is capable to synthesize 8k resolution textures, thanks to a extremely well calibrated tiling system.
IA-based image synthesis (such as style transfer) is known to be either slow (many hours to produce a low-definition image), or low-quality (fast style transfer). Nexture on the other hand achieves VFX-ready quality in record time (3.5 hours for an 8k map). And thanks to tiling, first results are available usually in a few minutes.
Artists can apply multiple details to the same texture and delimitate where each detail should be applied. The synthesized image is naturally seamless without performing any form of pixel-interpolation.
Artists have a few expressive settings to control the output, and Nexture integrates a preview mode that provides immediate feedback of these settings.
It is also possible to provide new reference pattern banks, and immediately start using them for synthesis. This opens up many new workflows for VFX houses.
3D Renders / 4K Displacement courtesy of Nicolas Collings
To quickly get up to speed, Nexture features a unique human-skin texture bank, included with the software and specifically tailored for texture transfer. It contains a diversity of 100 patterns commonly found on human skin at unprecedented definition and scale.