Sony F35 - 2000ISO test

Heck, lets shoot Slog, and rate the camera at 2000ISO. We'll get an extra 2stops of DR in the highlights, and a much more filmic tone curve (tons of highlights, less shadows), yeah baby!
Noisy footage? Of course! But Neat video can do magic and now that it can be integrated inside Davinci resolve, it's a super simple process. 

So if the F35 already had a very pleasing highlights rolloff, imagine how good it's going to be now with an extra 2 stops (or 3 if you're crazy) in the highlights. 
I shot these tests on 10bit 422, I see no banding issues of course, but I'm sure 12bit 444 would be the best for this workflow, as most information gets concentrated in the lower part of the histogram.

Shooting Joyride circuit for Pennzoil

Alexa mini. This is the smallest handheld setup we were able to build.

Alexa mini. This is the smallest handheld setup we were able to build.

Car mounted shots were shot on the Sony A7r2 for fear of damaging the Alexas at high speed. ;D

Car mounted shots were shot on the Sony A7r2 for fear of damaging the Alexas at high speed. ;D

For obvious reasons most of the film was shot from the russian arm. Away from the cold of the night. 

For obvious reasons most of the film was shot from the russian arm. Away from the cold of the night. 

Samsung NX1 - firmware update 1.30

Samsung just released a firmware update that boosts bit rate up to 80Mbits on 1080p modes among other improvements. Time to test.

UHD downsampled to 1080p shows a clear improvement in image quality vs just recording at 1080p. So yep, it's the way to go unless you need slow motion.

 

Testing Gamma modes

Does the DR gamma mode get extra dynamic range? Yep, but only on the shadows. Highlights remain the same.

Samsung NX1 - real life footage

Some real life footage captured with the Samsung NX1 - UHD internal recording. No grading. 

My notes:
- small "turist" camera that delivers pretty good images
- rolling shutter is decent
- AF is impresive
 

SAMSUNG NX1 - 4K mirrorless camera test

Samsung NX1 vs Sony A7s
by Bruce Allen and Macgregor

Alright, lets compare the NX1 to the king of the 4K mirrorless, the Sony A7s. But please remember, that the A7s requires an external recorder if you want 4K, that makes the combo about x4 times the price of the NX1 (currently $1199). So not a very fair comparison. We used the Odyssey 7Q+ on this test.

1. NX1 UHD internal recording. Camera matches lightmeter reading.
2. NX1 UHD external recording to Odyssey on prores HQ. Camera matches lightmeter reading. Shift in colors/levels vs internal recording.
3. A7s UHD cine4 gamma. External recording to Odyssey/prores HQ. Camera matches lightmeter reading.
4. A7s UHD Slog gamma. External recording to Odyssey/prores HQ. Camera metering.
5. A7s UHD Slog gamma. External recording to Odyssey/prores HQ. Lightmeter metering (which produces under-exposed results).

No apparent change in detail between internal and external recording with the NX1. However there is a shift in levels and colors.

The NX1 seems to produce slightly sharper images at 4K. But it introduces very visible sharpening artifacts. The A7s doesn’t seem to resolve the same amount of detail. Of course this difference is totally invisible when watching the footage at 1080p and even in 4K the differences are subtle.

Levels shifts:
The levels shift appears to be in the conversion from h.265 to ProRes: most h.265 converters convert the NX1’s files to a 0-235 ProRes file (well, the 10 bit equivalent thereof). The Odyssey saves out a 0-255 ProRes file, so it appears brighter.

Color shifts:
The color shift also appears to be in conversion from h.265 to ProRes: if you decode the h.265 file assuming it has Rec.709 color primaries, they don’t match the Odyssey. However, if you decode the h.265 file assuming it has Rec.601 color primaries (old school standard def NTSC colors), the colors do match the Odyssey. Also, if you load the ProRes file and perform a Rec.601 to Rec.709 conversion, they match the h.265 colors. Since the h.265 colors seem better, it seems likely that there is an error in the HDMI output.

 

Luma and chroma sharpening and blurring issues:
It is important to separately examine luma and chroma on the NX1 as they each have their own issues.

The luma is over-sharpened, yet the chroma has a blurry, heavy-handed noise-reduction algorithm applied, even at ISOs such as 100 and 200. There seems to be no way to turn this off.

This can clearly be seen in the Macbeth chart: the colors of each chip spread for several pixels both horizontally and vertically into the gray dividers. The A7s does not suffer from this to the same degree at all.

This is disappointing as the NX1’s 4K image is supposedly derived from the full-res sensor. The downsampling of that Bayer data from approx 6K to 4K should give it the best color detail on the market, possibly approaching true 4:2:2 color detail. Instead the color detail seems worse than 4:2:0. It’s definitely worse than the A7s.

Unfortunately, the smudging of the color takes place quite early in the pipeline because it is evident in the Odyssey recordings as well. The h.265 compression has little effect on color detail - it just adds some extra blocky compression artefacts.

Hopefully a future firmware update can give us an option to turn off these algorithms that do more harm than good, and perhaps devote those CPU cycles to other more beneficial tasks.

Samsung NX1 vs Sony A7s
4K image quality test - studio version 2

After noticing how the NX1 still applies huge amounts of sharpening I lowered this setting to the minimum level (-10) and I raised the  A7s detail to 0 to see if I could get a sharper images out of the A7s in 4K.

Results:
At detail 0, the A7s shows sharpening artifacts.
he NX1 still shows sharpening artifacts but it seems to resolve more detail.     

 

Samsung NX1 - dynamic range

I can see about 10 stops of dynamic range. 
With the same testing procedure, the Sony A7s delivers 12 stops.

What about just shooting internally on 1080p?

Samsung seems to win in terms of resolved detail. However I think I prefer the beter DR of the Sony. However, remember that the sony costs twice as much as the Samsung, and the UHD downsampled to 1080p on the Samsung blows the Sony away and gets you 444 color sampling. So it's a hard call.

Slow motion on the NX1:
Testing if the quality drops when shooting 120fps vs 24p. Nop, detail and noise seem to be the same. 

No visible loss in quality when switching between 24, 60 or 120 fps.

 

Some conclusions:
- Great image quality
- Easy to use
- internal 4K recording
- great price
- s35mm equivalent sensor size
- excellent EVF
- Great 120fps slow motion mode (given the price of the camera)
- EXCELLENT Autofocus performance in video mode

The NX1 seems to have a great sensor but it is hurt by its own processing. Right now, Samsung could seriously improve the quality of the NX1’s video just by:

- creating a true Dynamic Range or Log mode that doesn’t clip the shadows and highlights
- blurring the chroma less (or allowing us to control it or turn it off entirely)
- giving us an easy way to turn sharpening off (0 is clearly heavily sharpened)
- increasing the bit rate in the h.265 encoder or finding some way to get it to stop the heavy blocking artifacts
- disable the noise reduction for real?

These could be accomplished by firmware updates. Given how much the camera has improved in the firmware updates so far, we’re hopeful.

Additionally, there are several functional firmware updates that could really help:
- allow us to hit a button to zoom in to check focus even if using Samsung’s own autofocus lenses.
- allow us to zoom in to check focus while we are recording
- get rid of the Video recording standby mode, it's annoying and confusing
- let us assign video recording to the shutter button, as the video recording button is hard to reach and press.

Oh, did we mention the camera has touchscreen? It's pretty awesome!

 

 

 

 

Arri Amira 2K vs Sony A7s+shogun UHD

Workflow:
Amira - 50mm Sony 4K PL cinealta lens - 2K prores4444 - Davinci resolve - Alexa to 709 LUT - exported as 1080p
A7s in Slog/cinema - nikon 70-200 f4 - shot at UHD on the Shogun - proresHQ - Davinci resolve - data levels - Slog F35 macgregor LUT - exported as 1080p

Note: Have in mind that the A7s was on a magic arm attached to the Amira (which acts as a spring), that explains the jerky bumps (rolling shutter artifacts apart).