HBO CAS 2023

[Kemalettin Sert]

I was at the screening yesterday and it was an eye opener for sure.

Alexa 35 is the new benchmark for dynamic range and i loved how V-Raptor looked cleaner than Venice 2 at high iso.

I would like to hear your opinions as well.

[David Mullen ASC]

It was pretty interesting.

First question I had, but was answered by Jim Mathers as moderator in the Q&A, was what sort of lights were being used for the skin tone test because there was a lot of variation but mostly in saturation, which seemed odd. I figured the issue was probably less to do with the cameras than with the lights. It turned out, as I suspected, they used LED lighting (Creamsource Vortex RGB panels.) Oddly enough, 35mm 5219 did not perform well under those, the faces looked a bit desaturated, and conversely, the reds were oversaturated on the Sony Venice 2. The panel said that up until now, they always used tungsten for their skin tone test, but since LEDs were now the most commonly-used lights on set, they decided to test with those because if there was an issue, then that might spark a useful debate.

You would have expected 35mm film to be the gold standard when it came to skin tones but in this case, it wasn't.

In the mixed light test, film and the Alexa Mini were the only two cameras to pick up the green spike in the kitchen fluorescents, but then the Alexa 35 really picked up the green spike in the mercury vapor street lighting but oddly enough, rendered the kitchen fluorescents as neutral as did the other digital cameras.

Resolution-wise, in 4K projection, the 3K sensor of the Alexa Mini did not seem any softer than the 4.5K, 6K, and 8K sensors of the other cameras; it was surprising how competitive the Alexa still is, even with a decade-plus sensor design.

The Red V-Raptor was sort of the surprise hit of the test though it maybe was a bit less saturated than the others. It was so clean in the shadows and worked so well in the underexposure tests, but still had some clipping issues in highlights, that it seemed clear to me that it should be rated a stop faster than recommended. It's more like a 1600 or 2000 ASA camera.

The Sony Venice 2 looked great but was in some ways a bit uneven, oversaturated in the reds sometimes. But in the latitude test which had rooms of different exposures in one frame, the dimmed tungsten lampshades in the dark bedroom lost some saturation in the Venice 2 version for some odd reason, the other cameras rendered them as orange but the Venice 2 rendered it as a bit grey-ish orange.

The Alexa 35 blew everyone else out of the water when it came to dynamic range and only film was close to the same amount of highlight latitude. In both the test with the hot sparkling fireworks of the safe explosives and the hot curtain sheer, only film and the Alexa 35 held detail, with the Alexa Mini following.

In the 3200 ASA moonlight test, the Sony Venice 2 and Red V-Raptor did the best, 500T film (with a 2-stop push) did the worst. At first, the Alexa 35 was also one of the best but in the second shot, with a lot of movement, there was some weird compression artifacts from their internal noise reduction software. I don't know if the level of noise reduction can be dialed down in that mode. 2nd to last was the Blackmagic 12K camera, which had a lot of fixed pattern noise.

[Kemalettin Sert]

 

12 minutes ago, David Mullen ASC said:

It was pretty interesting.

First question I had, but was answered by Jim Mathers as moderator in the Q&A, was what sort of lights were being used for the skin tone test because there was a lot of variation but mostly in saturation, which seemed odd. I figured the issue was probably less to do with the cameras than with the lights. It turned out, as I suspected, they used LED lighting (Creamsource Vortex RGB panels.) Oddly enough, 35mm 5219 did not perform well under those, the faces looked a bit desaturated, and conversely, the reds were oversaturated on the Sony Venice 2. The panel said that up until now, they always used tungsten for their skin tone test, but since LEDs were now the most commonly-used lights on set, they decided to test with those because if there was an issue, then that might spark a useful debate.

You would have expected 35mm film to be the gold standard when it came to skin tones but in this case, it wasn't.

In the mixed light test, film and the Alexa Mini were the only two cameras to pick up the green spike in the kitchen fluorescents, but then the Alexa 35 really picked up the green spike in the mercury vapor street lighting but oddly enough, rendered the kitchen fluorescents as neutral as did the other digital cameras.

Resolution-wise, in 4K projection, the 3K sensor of the Alexa Mini did not seem any softer than the 4.5K, 6K, and 8K sensors of the other cameras; it was surprising how competitive the Alexa still is, even with a decade-plus sensor design.

The Red V-Raptor was sort of the surprise hit of the test though it maybe was a bit less saturated than the others. It was so clean in the shadows and worked so well in the underexposure tests, but still had some clipping issues in highlights, that it seemed clear to me that it should be rated a stop faster than recommended. It's more like a 1600 or 2000 ASA camera.

The Sony Venice 2 looked great but was in some ways a bit uneven, oversaturated in the reds sometimes. But in the latitude test which had rooms of different exposures in one frame, the dimmed tungsten lampshades in the dark bedroom lost some saturation in the Venice 2 version for some odd reason, the other cameras rendered them as orange but the Venice 2 rendered it as a bit grey-ish orange.

The Alexa 35 blew everyone else out of the water when it came to dynamic range and only film was close to the same amount of highlight latitude. In both the test with the hot sparkling fireworks of the safe explosives and the hot curtain sheer, only film and the Alexa 35 held detail, with the Alexa Mini following.

In the 3200 ASA moonlight test, the Sony Venice 2 and Red V-Raptor did the best, 500T film (with a 2-stop push) did the worst. At first, the Alexa 35 was also one of the best but in the second shot, with a lot of movement, there was some weird compression artifacts from their internal noise reduction software. I don't know if the level of noise reduction can be dialed down in that mode. 2nd to last was the Blackmagic 12K camera, which had a lot of fixed pattern noise.

Thank you for the insight Mr Mullen.

Yes i was totally shocked when i saw 2 stop pushed 500T in that moonlight test.It was completely unusable.

For Alexa 35 they said it only used in ES mode and i found ES mode creates a weird noise pattern and i talked to few other colorists they prefer non ES mode on Alexa 35 and use Resolve s denoise plugin.

Im  sure Arri will fix that in the future updates.

Edited July 30, 2023 by Kemalettin Sert

[Diego Obledo]

Being a roadshow it may not be available but wanted to ask. Is there any video of this by any chance? Thank you all 

[David Mullen ASC]

No, the test was for internal use at HBO with limited public screenings where they can make sure the 4K presentation is accurate.

[Diego Obledo]

Understood.Thank you so much!

[victor huey]

they just screened it at Discovery Warners in NYC during NAB

[Karim D. Ghantous]

On 7/31/2023 at 8:50 AM, Kemalettin Sert said:

when i saw 2 stop pushed 500T in that moonlight test.It was completely unusable.

I have seen both 7219 and 5219 pushed and they looked great. Even 5298 (discontinued 20 years ago) looks great when pushed by two stops. Perhaps there was exposure compensation used in post?

[David Mullen ASC]

32 minutes ago, Karim D. Ghantous said:

I have seen both 7219 and 5219 pushed and they looked great. Even 5298 (discontinued 20 years ago) looks great when pushed by two stops. Perhaps there was exposure compensation used in post?

It was unusable just because 500T pushed to ISO 2000 was still grossly underexposed under moonlight.