Nikon Z9 N-RAW Exposure (DR) Test
Фильм және анимация
This test looks at the over and under-exposure range for the two versions of Nikon's N-RAW files produced by the Z9. As you see, I chose a contrasty scene with a lot of gray tones that challenges the DR of a camera.
METHODOLOGY: I designed this test to show how far off a "proper" exposure the Z9 can be-in both SDR mode and LOG-and still bring a shot back to a "normal" rendering. To ensure consistency, all of the test shots were exposed based on the reading of my Sekonic light meter in incident mode. (At the end of the video there's a sequence comparing that exposure method with Z9 Matrix metering, Z9 zebras, and false color from an Osee external monitor.) They were then processed to bring them back to normal exposure.
PROCESSING: All the shots, both SDR and LOG, were developed in Resolve to N-LOG gamma and rec.2020 color space, using DaVinci Managed Color. The contrast of the normal exposure shots were adjusted using a group node, after the clips' grade so that basic contrast adjustment applied to all the shots. Then each of the over and under shots were graded as close to matching the normal exposure as I could make them.
TAKEAWAYS:
1. SDR has about one stop of overexposure lattitude. By +3 stops, a lot of detail is being lost above the mid-tones, and I would consider the shot unusable. +4 and +5 stops are totally unusable. And while the +2 stop shot could be used in a pinch, it's definitely losing detail in the hilights.
2. LOG has a LOT of overexposure headroom, in stark contrast to behavior of the SDR files. Even at 5 stops over, a shot can be salvaged with acceptable hilight detail (in this situation). There is lost dynamic range. That shows up as increased contrast-seen especially in the bright stone chimney and the evergreen shrub to the right of frame. There is very little color shifting as overexposure increases. All good news. But don’t think this means you can overexpose without consequence. The increased contrast (reduced DR) will be very unkind to skin tones and other subtle shades.
3. SDR has a LOT of underexposure latitude. At -2 and -3 stops, the graded shot looks surprisingly good, and even at -4 stops, the shot could be usable in a pinch. Importantly, that means shadows can be pulled up by 3 stops in a properly exposed shot with very little penalty, and by 4 stops in a pinch. But that’s the limit. DR falls off a cliff between -4 and -5 stops.
4. LOG files don’t handle underexposure well at all. There is about 1 stop of underexposure latitude. Even at -2 stops, color is shifting to green and noise is becoming apparent. At -3 stops, the green shift in the lower-mids is really quite bad, and at -4 and -5 stops, the shots are masses of noise.
Пікірлер: 13
Finally, someone who knows what the hell they're doing. Thank you for this
So Nlog is better with overexposure, SDR is better with underexposure.
@ZuZusGramp
Жыл бұрын
Yes. That’s the takeaway I had from this.
That was very interesting. Thanks for posting it. Tomorrow, I'm going to experiment with N-raw to see how N-raw SDR set at iso lower than 800 does when grading it as HDR in Resolve. It certainly held up well underexposing at low iso grading to SDR. I often grade HDR rec 2020, so curious to see whether raw really requires 800 or higher when grading HDR..
@ZuZusGramp
2 жыл бұрын
I'd be interested to see that test. In quick tests using Resolve's HDR grading controls, I saw no real difference in results, though their increased granularity helped on some cases. But I haven't experimented at all with HDR output, as I have no clients asking for that yet.
very useful. so as for NRAW+SDR and NRAW+NLOG, which one do you suggest while we shooting a sunset or sunrise such high light contrast sence? which one has more dynamic range we can edit in Davinci to get a better quality video? or they are actually same for the result? just when we shoot in NRAW+SDR, we make sure the correct exposure of the sky part in shooting sunrset? when we shoot in NRAW+NLOG,we make sure the correct exposure of the land(dark) part ? sry about my English,hope i make my question clear to you
Thanks 🙏 could you show how you can recover when over and under exposed in post ? I wonder how much you can bring back😊
Great test, thank you! I've found that the image on the back LCD looks overexposed, when in fact it is well exposed. I need to keep reminding myself to judge exposure by the waveform and ignore the over exposed image on the back LCD. I wish we could import custom LUTs into the Z9.
@ZuZusGramp
Жыл бұрын
Yeah, Nikon is making some odd choices in monitoring its N-raw formats. Both the SDR and LOG that you see are overlays on the N-raw image data and not very reliable representations of what you’re shooting.
I’m confused by what you mean when you say processed as log vs SDR. If it’s shot in N Raw isn’t the log vs SDR setting simply metadata that doesn’t carry over into the final look?
@ZuZusGramp
Жыл бұрын
When you take an Nraw file into Resolve, Resolve reads the metadata in the file and develops the raw file using the appropriate gamma for the metadata-so Nlog for Nraw LOG and Rec709 for Nlog SDR. You don’t have to do that. You can tell Resolve develop the raw file using any gamma you like in the Camera Raw settings. In this case, I told it to develop all the files to Nlog so the signal path would be identical.
Do you by any chance have a polarizer on this? The blue sky seems to be unusually dark.
@ZuZusGramp
Жыл бұрын
No. I am using Firecrest Ultra IRND filters, but no polarizer. The sky is as dark as it is because the exposure is accounting for the glare off the pavement and the light paint on the house. Also, it was an especially clear, dry day, so there was about zero haze.