How is Fujifilm X-Trans RAW demosaiced, and are there downsides versus Bayer sensors?
Asked 4/18/2012
8 views
2 answers
0
Fujifilm’s X-Pro1 introduced the X-Trans color filter array, which uses a less repetitive 6×6 pattern instead of the standard Bayer arrangement. Fujifilm claims this reduces moiré and false color enough to omit an optical low-pass (AA) filter.
How does RAW conversion for this sensor work in practice? Is it basically the same kind of demosaicing used for Bayer sensors, just adapted to a different pixel pattern, or does it require fundamentally different processing?
Also, aside from possible software support issues for third-party RAW converters, are there any likely drawbacks as well as the claimed benefits?
Originally by Photography Stack Exchange contributor. Source · Licensed CC BY-SA 4.0
Photography Stack Exchange contributor
14y ago
2 Answers
2
This isn't the first camera to deviate from the standard RGB bayer layout, there have been cameras released with cyan as a fourth colour (replacing half of the green filters) as well as a fourth clear filter for better luminance resolution and low light ability. Fuji have also experimented with octagonal sensors and split dynamic range sensors so know a thing or two about nonstandard demosaicing processes!
The layout Fuji have chosen will actually make demosiacing easier, you could probably get away with a linear interpolation if you had to.
More complex demosiacing algorithms all try to make guesses about which values are likely to remain constant between adjacent pixels, giving you an extra sample for free. Exactly the same principals can be applied to the Fuji arrangement. However the implementations will need to be tweaked to take into account the swapped red and blue filters.
However these tweaks are unlikely to be implemented in the popular RAW converters such as Lightroom due to it's niche market so users of the camera will probably be stuck with whatever software Fuji come up with for some time...
Originally by user1375. Source · Licensed CC BY-SA 4.0
user1375
14y ago
0
Generated from our catalog & community — verify before relying on it.
It still uses demosaicing; the core idea is the same as with Bayer sensors: estimate the missing color values at each pixel from neighboring samples. The difference is that the algorithm must be adapted to Fuji’s non-Bayer X-Trans pattern.
So this is not a completely different class of RAW conversion, just a different sampling geometry. In principle, the same kinds of interpolation and edge-aware demosaicing ideas can be used.
A likely advantage of the X-Trans layout is reduced moiré and false color because the color filter pattern is less repetitive, which is why Fuji can omit the AA filter and preserve more sharpness.
The main downside is software complexity and support. Because the pattern is nonstandard, RAW developers must specifically support it, and early or weaker implementations may produce poorer results than mature Bayer workflows.
In short: X-Trans demosaicing is more complicated in implementation, but not fundamentally different in concept. The tradeoff is potentially better detail and fewer moiré artifacts versus a greater dependence on good RAW-converter support.
Recommended products
UniqueBot
AI14y ago
Your Answer
Related Questions
How does Fujifilm X-Trans improve image quality compared with a Bayer APS-C sensor?
Are some digital cameras better suited to black-and-white photography than others?
Can uncompressed Fujifilm RAF files be converted to the newer compressed RAF format?
How much light and effective resolution do Bayer color filter arrays reduce in digital cameras?
Does a Bayer RAW file record one color value per photosite or full RGB per pixel?