The spectre of sample variation has reared its ugly head yet again!
I am disappointed to say that the first copy of the Fuji 100-400mm OIS zoom that I was eagerly awaiting, has proved to be an unacceptable example and I have already returned it to the retailer.
While I am unable to do scientific tests on lenses I can complete enough tests to identify fairly obvious optical issues such as uneven performance around the frame that can indicate decentering or tilting of elements.
I put this lens through my usual new lens process of shooting real-world images and some more careful test images to check for issues. When I did the first shots of the moon I posted about previously, I did notice what appeared to be more softness on the left half of the frame and some slight "double-imaging" of the contrasty left edge of the Moon. I hoped this was down to the VR or some atmospheric effects but the more I shot the lens, the more I felt there was an issue on the left side and a very significant issue in the upper left corner.
The optical issues manifested at max zoom, which was critical to me as I am likely to use the lens at or around 400mm much of the time and will use the teleconverter too. The latter only seemed to aggravate the lens issue, as one might expect. I therefore did not test formally at all focal lengths as the lens was never going to be a keeper if it didn't perform at 400mm. I did however take a few test chart shots at 200-300mm and the issues were still present.
To test, I placed the lens on a tripod, VR off, levelled in pitch and roll with a spirit level and aligned the camera as carefully as I could on an A2 test chart mounted on foam board. To ensure alignment issues did not affect the tests, I broke down and repeated the setup a number of times to average out errors and even rotated the camera 180 degrees. I also shot with 2 bodies to minimise the possibility of lens mount issues. I used remote release with both manual and electronic shutter.
The weak left and left upper corner were very obvious even at normal screen viewing sizes and no pixel peeping was needed. Worse still, the issues remained a stop down from maximum at f8. On a relatively slow lens like this it really has to achieve even performance across the frame a stop down. A disappointing result and not consistent with comments/reviews I have read concerning the performance that can be expected from this lens.
The following are some of the images. I know they are small due to the blog format and this may mask the issues somewhat, but the lack of resolution, smearing, loss of contrast and bleeding between black and white should hopefully be visible, even here. Imagine how it looks at 100%! A £1400 zoom lens simply has to perform better than this else it is not fit for purpose.
The tilted or non-central chart positions exist because I was more concerned about camera/lens alignment than exact centering on the chart.
Here the 1951 chart and star at upper left is weaker as is the left side overall. This is at f8!
On this one note the low contrast of the small star at upper left in particular.
Here the small star at upper left is weaker and there is smearing/contrast loss on the left side of the large star.
I have now had reason to send back 3 Fujinon lenses for QC issues (my first copies of the 23 f1.4 & 18-55), which is a worrying hit rate, even though I do have most of the XF range. This is a shame when so much design effort has gone into the theoretical performance of these lenses. Let's hope that things improve and the assembly quality starts to match the design quality! I know these big zoom lenses must be a QC nightmare with huge numbers of elements, OIS system, zoom and focus groups and extending barrels, but it shouldn't be a lottery on this scale.
I await a second copy and just hope!