Image quality

Key takeaways:

  • The X-T3's image quality is impressive, especially at low ISO
  • The highest ISO setting falls a fraction behind the best of its peers (basically, don't assume a low light improvement, just because it's BSI)

Our test scene is designed to simulate a variety of textures, colors and detail types you'll encounter in the real world. It also has two illumination modes to see the effect of different lighting conditions.

The X-T3's images look great at low ISO. The slightly lower base ISO provides a small but measurable improvement over its predecessor (even when compared at the pixel level, which puts cameras with smaller pixels at a disadvantage).

Looking at the Raw detail, the X-T3 does well, relative to its peers, but with no sign of moiré (though this doesn't mean it's totally free from aliasing). X-Trans' slightly lower frequency and less uniform capture of red and blue information does leave the possibility of occasional glitches in certain colors and patterns. Overall, the X-T3 is very competitive in terms of noise until its very highest ISOs (something we look into more closely, further down this page).

JPEG color lives up to Fujifilm's usual high standard, even in the default 'Provia/Standard' mode. It's comparable with the very likeable results from Canon, but with slightly punchier greens, along with reds and pinks with a fractionally less orange tint.

The default sharpening seems to strike a good balance between emphasizing detail without going over-the-top. Meanwhile noise reduction also appears pretty well balanced between suppressing noise and retaining detail. There are still hints of noise reduction causing desaturation that's especially noticeable in faces, with a loss of distinction between the face and the whites of the eyes.

Changes in Adobe's processing

We were surprised at how significant the differences appear between the X-T2 and X-T3 at high ISOs. Our read noise measurements suggest the difference should be around a third of a stop but our original conversions made it look like more than that.

A discussion with Adobe revealed that they've made some changes to "increase accuracy of shadow (near-zero) values in raw files," which can give deeper blacks and more contrast in dark regions of the image. Even re-processing the X-T2 files using this revised processing still doesn't fully account for the slight differences we're seeing at very high ISOs.

Capture One

We show Adobe Camera Raw since it is probably the most widely-used Raw processing software, either as a Photoshop plugin or as the underpinnings of Lightroom. However, given how many readers were reporting better results when using other software and in the light of Phase One's announcement of Capture One Express Fujifilm, we thought we'd see how the two cameras compare.

X-T3 | ISO 25,600 | Capture One Fujifilm X-T2 | ISO 25,600 | Capture One Fujifilm

With noise reduction minimized, the X-T3 still exhibits fractionally more noise than the X-T2. Not quite as dramatically as in ACR.

Even upon being downscaled to the same resolution as the X-T2, the X-T3 still shows greater noise at its highest ISO setting.

X-T3 | ISO 51,200 | Capture One Fujifilm | Resized to 24MP
X-T2 | ISO 51,200 | Capture One Fujifilm

The most likely explanation we can think of for this discrepancy is a very slight increase in read noise as a result of running all the electronic components faster, to allow the faster sensor readout. At the highest ISOs even a tiny increase in read noise would be amplified many, many times.

If this is the case then the slightly reduced high ISO performance is the price you pay for the camera's increased speed. However the difference is extremely subtle and only really has an impact at ISO 51,200. Overall we'd say the X-T3's image quality is at least as good as the X-T2's: just don't expect a low-light improvement just because it's a BSI sensor.