FRACTALPIX Upscaler™ -
Ultra High Quality Upscaling of Digital Assets
|
| | | |
FRACTALPIX Upscaler™ Cloud Service - Use Interactive Web Demo
FRACTALPIX Upscaler™ is our in-house distributed system for upscaling video frames from production masters and other HDR images, up to RGBA floating point 32-bits per color channel (128 bits per pixel).
Pixel values are calculated with single-precision floating-point 32-bits per color channel.
High quality upscaling is non-trivial and is a subjective managed process that varies according to characteristics of the source content.
Multiple factors effect image quality requiring more sophisticated solutions than general purpose one click upscalers.
Viewing upscaled images at 800% zoom reveals the effectiveness of the applied algorithm.
For Digital Cinema and other professional use cases we advocate the lossless video workflow advantage to future proof valuable visual assets. High quality upscaling is difficult to achieve if source images are polluted
with compression artifacts. Upscaling results are as important as the original production efforts, working with frames extracted from lossy compression formats is not recommended. Subtle differences in quality matter when generating upscaled masters that should have the same pixel fidelity
as original debayered camera RAW.
FRACTALPIX Upscaler™ utilizes the state-of-art upscaling technology that produces superior results than standard linear scaling methods.
Upscaled 4K to 8K Examples Cubic vs FRACTALPIX Upscaled Drawing
16-bit Medical X-Ray Example
FRACTALPIX Upscaler™ 2x upscaling example from 480x270 source image
FRACTALPIX Upscaler™ example of interpolation and edge preservation
FRACTALPIX Upscaler™ cropped 4K upscaled to 8K, Source:RED .RD3 RedGamma4 Rec. 2020
2x Lanczos3 vs FRACTALPIX Upscaler™ cropped 4K upscaled to 8K, Source:RED .RD3 RedGamma4 Rec. 2020
Lanczos3 vs FRACTALPIX Upscaler™ cropped 4K upscaled to 8K zoomed 800%, Source:RED .RD3 RedGamma4 Rec. 2020
Sharpened Lanczos3 vs FRACTALPIX Upscaler™ cropped 4K upscaled to 8K zoomed 800%, Source:RED .RD3 RedGamma4 Rec. 2020
An example of a poorly implemented upscaling algorithm that destroys pixel fidelity with blocking artifacts
|