Advertisement

Sony BURANO 8K Lab Test: Rolling Shutter, Dynamic Range, and Exposure Latitude

Sony BURANO 8K Lab Test: Rolling Shutter, Dynamic Range, and Exposure Latitude

The empire strikes back! Looking at the specs and innovative features of this new camera from Sony, we eagerly anticipated the arrival of this latest addition to the CineAlta line at our CineD headquarters. Now the time has come, and we torture-tested the BURANO in our standard CineD lab test. Curious to hear the results? Then read on …

A lot has been written on CineD about the new BURANO by my colleague Nino here or here, but let me quote my colleague Omri in his article here: “With a price tag of $25,000, it probably won’t find its way to each and every content creator, but will appeal to some owner-operators, studios, and rental houses. The Sony BURANO fills a unique niche. The camera incorporates a class-leading spec list with a single-operator-aimed design.”

Now, we wanted to run it through its paces in our standard CineD lab test, but we had to wait for a unit with final production firmware version 1.0, but now all has been sorted and we can finally test it! A big “thank you” again to my dear colleague Florian Milz who helped with every aspect of this lab test!

The Sony BURANO in the CineD lab
The Sony BURANO in the CineD lab. Image credit: CineD

So let’s start with rolling shutter.

Rolling Shutter of the Sony BURANO 8K

Let’s start first with full-frame 8.6K X-OCN LT mode in 25 frames-per-second:

Rolling shutter in full-frame 8.6K X-OCN LT mode
Rolling shutter in full-frame 8.6K X-OCN LT mode. Image credit: CineD

We get a rolling shutter of 18.9ms (less is better) – ouch! That is on the high side. The Sony VENICE 2 (lab test here) has less than 3ms rolling shutter. Compared to other offerings in the full-frame arena, even Sony’s very own Alpha cameras like the A1 (lab test here) with 16.6ms, or the leader in the consumer full-frame space, the Sony a7S III, with 8.7ms (lab test here) fare much better. Not to mention recent cameras like the Sony A9 III (lab test here) or the RED V-Raptor X (lab test here), both of which feature a global shutter sensor, thus completely eliminating any image skew issues …

In 6K 25fps full-frame FFc (~1.07 crop) mode, rolling shutter improves slightly to 16.9ms. At 6K 59.94p FFc read-out mode changes again, and the rolling shutter reduces further to 14.8ms – otherwise we couldn’t get 60fps (with 16.9ms):

Rolling shutter in 6K FFc mode at 59.94fps
Rolling shutter in 6K FFc mode at 59.94fps. Image credit: CineD

Please refer to the lab database for additional results in all the other modes.

Dynamic Range of the Sony BURANO 8K

Please head over here if you don’t know how we test dynamic range. We developed the X-OCN files in DaVinci Resolve 18.6.6 using the following settings in the camera raw tab (ISO3200 test shown here):

Sony RAW settings in DaVinci Resolve 18.6.6
Sony RAW settings in DaVinci Resolve 18.6.6. Image credit: CineD

Let’s start with Sony RAW, X-OCN LT 8.6K at ISO800. The waveform shooting the Xyla21 chart shows about 12 stops, if not 13.

Waveform plot of the Xyla21 chart in 8.6K X-OCN LT mode
Waveform plot of the Xyla21 chart in 8.6K X-OCN LT mode. Image credit: CineD

Just to make you aware, and because there were many questions on the RED Raptor X lab test, here is the RGB waveform for 8.6K at ISO800, X-OCN LT. As you can see, there is no highlight recovery in action (as is the case for RED cameras with the IPP2 image pipeline), the second patch from the left already has all RGB color channels, whereas the first patch is clipped:

RGB Waveform plot of the Xyla21 chart at ISO800, X-OCN LT
RGB Waveform plot of the Xyla21 chart at ISO800, X-OCN LT. Image credit: CineD

The patches toward the shadows look quite noisy. Let’s have a look at IMATEST to reveal noise levels:

BURANO IMATEST analysis for 8.6K X-OCN LT at ISO 800
BURANO IMATEST analysis for 8.6K X-OCN LT at ISO 800. Image credit: CineD

We get 12.2 stops at a signal-to-noise ratio (SNR) = 1, and 10.8 stops at SNR = 2. At first glance, these are rather mediocre results. BUT: looking at the “Noise spectrum” graph on the lower right side, you can see signal amplitudes are retained even towards very fine detail, hinting at very little, if any, internal noise reduction at play. Also, the middle graph shows about 3 more stops toward the shadows (above the blue “12.2” line).

This is very promising. If the image pipeline sensor – codec can retain this fine detail and noise, then we will have a good chance to recover dark stops in our latitude test later.

Traditionally, Sony has always applied a varying degree of internal noise reduction to the compressed XAVC codecs. That holds true as well for the BURANO, looking at the full-frame 8K DCI XAVC H-I HQ mode (SG3.cine / SLog3) at ISO800 (now 8192×4320 resolution):

waveform plot for FF 8K DCI XAVC H-I HQ
waveform plot for FF 8K DCI XAVC H-I HQ. Image credit: CineD
BURANO IMATEST analysis for 8K DCI XAVC H-I HQ at ISO 800
BURANO IMATEST analysis for 8K DCI XAVC H-I HQ at ISO 800. Image credit: CineD

Now, we are getting 13 stops at SNR = 1 and 11.9 stops at SNR = 2. Better, right? Just a little side note, DaVinci Resolve did not recognize the data levels for 8K XAVC in “auto” mode. We had to set them to “full” manually.

Okay, so let’s take a look at the 6K FFc full-frame crop mode, which looks a bit like the sweet spot of this camera to me. In 6K FFc X-OCN LT mode, which has a resolution of 6052×3192 the dynamic range reads 13.5 / 12.4 stops for SNR 1 / 2 (for both 25 and 59.94 fps) in X-OCN LT mode:

BURANO IMATEST analysis for 6K FFC X-OCN LT at ISO 800
BURANO IMATEST analysis for 6K FFC X-OCN LT at ISO 800. Image credit: CineD

For XAVC 6K FFc the camera downsamples to 4K DCI in 10bit XAVC H-I HQ mode. This mode is very interesting for a lot of documentary shooters, who deliver compressed XAVC files in 4K straight out of camera, hence I included both the waveform and the IMATEST results for this mode below. We get 14.1 / 13.1 stops at SNR = 1 / 2. Obviously, the combination of downsampling and some internal noise reduction leads to these very good results:

Waveform plot of the Xyla21 chart at ISO800, 6K FFc XAVC H-I HQ. Image credit: CineD
BURANO IMATEST analysis for 6K FFc 4K DCI XAVC H-I HQ at ISO 800
BURANO IMATEST analysis for 6K FFc 4K DCI XAVC H-I HQ at ISO 800. Image credit: CineD

The Sony BURANO features a dual native ISO sensor, with the second native ISO at ISO3200.

Here are the dynamic range results for Sony RAW, X-OCN LT 8.6K at ISO3200, starting with the waveform plot:

Waveform plot of the Xyla21 chart at ISO3200, using 8.6K X-OCN LT mode
Waveform plot of the Xyla21 chart at ISO3200, using 8.6K X-OCN LT mode. Image credit: CineD

IMATEST reads the following:

BURANO IMATEST analysis for ISO3200, using 8.6K X-OCN LT mode
BURANO IMATEST analysis for ISO3200, using 8.6K X-OCN LT mode. Image credit: CineD

Dynamic range drops significantly to 11.7 stops at SNR = 1 and 9.35 stops at SNR = 2 (vs. 12.2 / 10.8 at ISO800) – usually dual native ISO sensors show a more consistent performance across the two ISOs. I think it is again related to the fact that there is minimal to no internal noise processing. The Sony VENICE 2 behaved differently here. The SNR = 1 result was the same for both ISOs and the SNR = 2 result was only 0.5 stops worse for X-OCN XT.

In 8K XAVC H-I HQ, the difference between ISO800 and 3200 is less:

BURANO IMATEST analysis for 8K XAVC H-I HQ at ISO 3200
BURANO IMATEST analysis for 8K XAVC H-I HQ at ISO 3200. Image credit: CineD

Now we get 13 stops at SNR = 1 and 11.2 at SNR = 2 (vs. 13 / 11.9 at ISO800).

Please head over to the database for all the other results in all the different modes.

Exposure Latitude of the Sony BURANO

As stated in earlier articles, latitude is the capability of a camera to retain details and colors when over- or underexposed and pushed back to a base exposure. This test is very revealing, as it pushes the complete image pipeline of any camera to its absolute limits – not just in the highlights but mostly in the shadows.

Our studio base exposure is (arbitrarily) chosen as having an (ungraded) luma value of around 60% on the forehead of our subject on the waveform monitor – in this case, my colleague Johnnie:

Base exposure of our standard CineD studio scene
Base exposure of our standard CineD studio scene. Image credit: CineD

Again, we developed the full-frame 8.6K X-OCN LT (Sony RAW) files to S-Gamut3.Cine / S-Log3 at ISO800 in DaVinci Resolve. The images were brought into the Rec709 space using an input color space transform (CST) to DaVinci wide gamut/intermediate, then adjusted to base exposure and finally bringing it to Rec709 by another color space transform node at the end. The image is a bit on the pinkish side.

Node tree in DaVinci Resolve 18.6.6
Node tree in DaVinci Resolve 18.6.6. Image credit: CineD

Any post-noise reduction was always made on the first node. Side note: I also tried the ACES workflow which has the advantage that exposure can be adjusted by entering the stop value in the HDR tab. There was a slight contrast difference to the CST workflow, but the results were the same. Also, here there is a slight pink touch to the image.

At 4 stops above base exposure, we are at the cusp of starting to clip the red channel on Johnnie’s forehead, but the image brought back to base looks totally fine:

4 stops overexposed, brought back to base exposure
4 stops overexposed, brought back to base exposure. Image credit: CineD
RGB waveform of the 4 stops over, brought back (and developed) image
RGB waveform of the 4 stops over, brought back (and developed) image. Image credit: CineD

Now, we start to close down the iris of our trusted CP.2 85mm T1.5 lens by one-stop increments, and from T8 onwards we halved the shutter angle.

Doing this until we reach 4 stops of underexposure, we get the following image, brought back to base exposure:

4 stops underexposed, brought back to base
4 stops underexposed, brought back to base. Image credit: CineD

We are at 8 stops of exposure latitude. This is usually the limit for most of the full-frame consumer cameras that we have tested so far. No noise reduction has been applied, and the image looks totally fine although some noise is showing up already. But the noise is very finely dispersed, which looks very nice and organic to me.

The recently tested Sony A9 III was the first Sony camera to reach a solid 9 stops of exposure latitude (VENICE 2 reached between 8 and 9 stops). Also, the recently tested RED V-RAPTOR X reached a solid 9 stops.

So, let’s move to 9 stops on the Sony BURANO:

5 stops underexposed, brought back to base
5 stops underexposed, brought back to base. Image credit: CineD

Well, this image definitely shows noise, but it is again very finely dispersed. It can be easily cleaned up by using noise reduction in DVR:

5 stops underexposed, brought back to base, using noise reduction
5 stops underexposed, brought back to base, using noise reduction. Image credit: CineD

This looks totally fine. For the Sony VENICE 2, it was already game over at 5 stops under in X-OCN XT!

Now let’s move to 6 stops of underexposure:

6 stops underexposed, brought back to base
6 stops underexposed, brought back to base. Image credit: CineD

Heavy noise starts to kick in. Let’s see if we can clean this up:

6 stops underexposed, brought back to base using noise reduction
6 stops underexposed, brought back to base using noise reduction. Image credit: CineD
Temporal and spatial noise reduction settings in DVR
Temporal and spatial noise reduction settings in DVR. Image credit: CineD

We are at 10 stops of exposure latitude! The only full-frame camera so far that was capable of pulling this off was the ARRI ALEXA Mini LF that we tested here. Noise cannot be completely eliminated as now there are also larger blotches of chroma noise that appear (look at Jonnie’s shirt). Johnnies shadow side of the face becomes greenish. But there are no horizontal or vertical lines (like we have seen on the VENICE 2 test), and only minor color shifts toward green, so I would call this result borderline, but sort of OK. If you compare it to the ARRI Mini LF result, the BURANO definitely looks a tad worse.

At 7 stops of underexposure, pulled back it is definitely “game over”. Just for your reference, here is the image:

7 stops of underexposure, pushed back.
7 stops of underexposure, pushed back. Image credit: CineD

All in all, this leads me to conclude that the Sony BURANO comes in at a very close second place behind the ARRI ALEXA Mini LF. Just for reference, our current benchmark for dynamic range and latitude is the ALEXA 35 (test here) with 12 stops of exposure latitude.

Side note: I also tested the full-frame 8K 10bit 4:2: XAVC H-I HQ codec for latitude. And clearly, it is worse than X-OCN LT. So no need to mention it any further here.

Summary

The Sony BURANO has one Achilles heel, which is the rather high rolling shutter of 18.9ms in 8.6K full-frame mode. But, for most use cases (especially as a solo operator or documentary shooter), the internal sensor stabilization will come in very handy to mitigate this issue.

Other than that, the BURANO exhibits the best results we have ever had in the lab for a Sony camera. It shows close to 10 stops of exposure latitude – the real-world test in our CineD studio! And yes, it is definitely better than the Sony VENICE 2, which fared worse in the latitude section.

If you are looking for the best dynamic range results out of the box and not so heavy files (e.g. for documentary work) the 6K FFc XAVC mode (which has a crop of around 1.07) which downsamples to 4K DCI 10bit XAVC 4:2:2 exhibits superb results at 14.1 / 13.1 stops at SNR = 1 / 2, coming in at 6th place in our dynamic range ranking, right behind the VENICE 2 (with its downsampled 4K mode).

To conclude – yes, the Sony empire definitely struck back!

What are your experiences with the new Sony BURANO? Have you shot with it already? Let us know in the comments below!

45 Comments

Subscribe
Notify of

Filter:
all
Sort by:
latest
Filter:
all
Sort by:
latest

Take part in the CineD community experience