It’s no secret that I’m a fan of the recently launched Sony FS5. It represents an great combination of capabilities and ergonomics, at a competitive price. In my original review I tested it mainly in 1080 HD as I believe this is the resolution most documentary and news users will opt for, but I also noted the relatively weak internally recorded 4K performance – especially when compared to the sister FS7.
Later testing in 4K revealed that the FS5 didn’t have the same stellar low light performance as the full frame a7S II. This is of course to be expected as the FS5 Super35 sensor has much smaller photosites than the a7S II. It gets very noisy when you get over 3200 ISO.
Along with that noise are some rather strange encoding issues that I didn’t spot at first. Paul Antico of Need Creative podcast began to see issues in his newly purchased FS5 and brought them to my attention. There is very obvious macro-block artifacting when there is any motion in the scene. Changing between common picture profiles did not eliminate the problem. It appears that the camera’s internal 100 Mbps 4K encoding just isn’t able to cope when dealing with noise and movement at very high ISO. To demonstrate the issue clearly Paul shot the video at the start of this article. Please download the full 4K version from Vimeo to really see what is going on.
Just to confirm that the issue wasn’t limited to just one camera I also asked Newsshooter contributor Slavik Boyechko to run similar tests on his FS5. The results were the same and you can see some grabs below. The lower image is a blow-up of the first and shows the macro-blocking well.
Paul’s tests show that this is not a problem with the FS5 sensor. When recording in HD internally, or 4K externally to a recorder like the Atomos Shogun or Convergent Design Odyssey 7Q, the FS5 has a nice clean image that rivals the FS7.
The strange thing is that according to Paul even the a7R II, that has a similar XAVC implementation at the same bitrate and colour depth to the FS5, seems to do a better job with fewer artifacts. The noise levels may vary but I was not expecting to see obvious encoding issues of this kind present in FS5 footage, it isn’t seen on other current Sony large sensor cameras.
How big of an issue is this? I don’t really think that too many factual or news shooters will be using the camera in 4K just yet. HD performance is very good and I would have no hesitation in using it. Higher end 4K productions using the FS5 on brushless gimbals or Steadicams are likely to be recording externally to a recorder and so would be unaffected. As a b-cam for lit interviews in 4K it should also be acceptable as the light levels are usually high enough to the keep the ISO low.
If you shoot in low light and want a new 4K camera that is where I might look at something other than the FS5. The combination of noise and compression mean that it really needs an external recorder to work well, and at that point you are ruining the small camera ergonomics and adding to the cost. I think it is far easier to step up to an FS7 instead. Another workable option may be to use an a7S II for low light work and the FS5 in good light.
We have brought this issue to Sony’s attention and they are looking into it. Hopefully they will be able to improve the FS5 performance via a firmware fix, but until they do don’t plan on using a FS5 without a recorder for your next 4K nighttime shoot.
Many thanks to Paul and Slavik for all the FS5 testing.
