ISO and light-metering options revisited

by **Admin** 2021-05-19, 01:36

ISO and light-metering options revisited

[size]
Catching up on this thread...

Quote:
[/size]

Originally Posted by aiolosimport
[wingsfromheaven] Thanks for your suggestion to set ISO 100/Spot metering for better video.
Did test flight and may be due to hazy weather condition, to my poor eyes, all Auto setting seems clearer view.

[size]
Thank you so much for the test! But wow, I was obviously unclear about ISO 100 and separate exposure metering. I'll try to provide some background details but the 'key takeaways' are implied by the headings in bold.

ISO 100 background:

We try to keep the ISO as low as possible in order to minimize the impact of both noise and noise-reduction (NR) on the video image. Even the best NR algorithms are somewhat destructive to the image, but the improved aesthetics generally outweigh any loss of detail. But NR algorithms vary greatly in effectiveness and in their destruction of image definition, even within a specific NR technique based on select parameters. We don't have any details about the particular noise-reduction employed by the X8-2018 camera, so we're left to speculate based on circumstantial evidence and the results.

Circumstantial evidence of NR limits:

X8-2018 camera hardware is very similar to that of the Mavic Pro/Platinum (MPP). What's interesting here is that some discerning MPP pilots learned to disable the in-camera NR using a hidden feature of the 'sharpness' setting, in order to pursue better results through post-production. We can only presume that this became necessary due to the inability to run more advanced NR on the MPP's A9 SoC, while the existing NR proved too destructive. Our A12s SoC is very similar, and given Fimi's relative inability to manipulate the stock camera firmware (X8-2018 camera) our only control over NR comes down to 'hiding' the noise by minimizing the ISO. But this objective of minimizing noise and NR effects has its limits in low-light as presented below.

Daylight conditions = Auto ISO:

The results suggest that daylight conditions lead to the 'clearest' image with the highest variable bitrate with the X8-2018 camera. Bitrates typically stay above 90Mbit/s with 4K@30fps in daylight to include high-contrast scenes. That's a good thing and it theoretically provides a 50% improvement in encoded 'image definition' (detail) and/or 'movement tracking' relative to MPP results. We can only assume that Auto ISO drives the ISO to 100 in good daylight, so manual ISO 100 shouldn't be necessary in these lighting conditions. Besides, you get manual Ev adjustment through the right-shoulder dial in Auto ISO mode, which is generally preferable over spot-metering in daylight conditions.

Low-light conditions = Auto ISO:

Let's face it, this is not a 'low-light video' camera given the 1/2" sensor and Auto shutter. Shutter speeds should remain sufficient at ISO 200/400 for slow/cinematic moments during Civil Twilight. But as it gets darker, particularly on moonless nights, we may have little choice but to preserve shutter speed using Auto ISO. Unfortunately, the resulting high-ISO drives a very aggressive NR algorithm in the X8-2018 camera, which severely softens the image. But generally speaking, this 'softening' is preferable to the color-distortion you'll encounter in post-production as you try to raise the exposure-curve more than ~1 Ev. . If you do find yourself increasing the exposure in post-production, F-Log holds-up better to extreme exposure corrections than General or Vivid. F-Log simply reveals random colored dots (noise) in post beyond +1 Ev, while the same adjustments tend to solarize the image with General/Vivid camera output.

High-contrast lighting conditions = manual ISO 100:

During 'high-contrast' lighting conditions such as Golden Hour, it might be best to select ISO 100 'manually' since the lighting should be sufficient for reasonable shutter-speeds. When we select manual ISO the right-shoulder dial switches from manual Ev adjustment to manual ISO adjustment. This mode continues to offer both full-frame metering and spot-metering, but spot-metering becomes the only manual influence over the exposure. And in this particular lighting, spot-metering can be key depending on the subject matter and planned drone flight path, as we'll see...

Spot metering with ISO 100:

'Ev' is a combination of both ISO and shutter speed due to the fixed aperture. So with 'manual ISO' selected we are influencing the 'shutter-speed' exclusively through light metering (full-frame auto-metering or spot-metering) in video mode.

Spot metering is to be used sparingly and only when the selected spot is 'less' likely to encounter significant changes in the scene illumination during planned drone maneuvers. The case-in-point under discussion when I mentioned spot-metering and ISO 100 was a high-contrast scene (with low sun-angle) in an orbit maneuver around a subject positioned [presumably] below the horizon relative to the drone altitude. In other words, it was a scenario in which full-frame exposure metering would invariably change throughout the orbit maneuver, perhaps changing the ISO and corresponding NR effects in the process. So the combination of spot-metering and ISO 100 promised the best results in that particular case.

Auto ISO with full-frame metering and Ev adjustment:

Your tests involved very different scenario due to the low-contrast scene illumination, good daylight (albeit thin overcast) and the particular drone maneuver. In those tests you suggest that the image looks more 'clear' with Auto settings than with manual ISO 100/spot-metering. I would suggest that we're simply seeing an aesthetic difference due to a difference in exposure. To my eyes the 'clarity' (detail and sharpness) appears to be the same in both cases, judging by the parking lot stripes in the lower righthand corner at 4:15. The exposure difference is a matter of your initial spot-meter placement. That particular spot drove the exposure higher than the automatic full-frame metering. ISO doesn't come into play in those tests because we can be [virtually] certain that ISO remained at 100 in both cases (Auto and Manual ISO) based on the scene lighting conditions. So the only variable is exposure metering through automatic shutter-speed control. Full-frame exposure metering was more appropriate for the reverse dolly maneuver you performed because spot-metering simply risks dragging the 'spot' across variable scene illumination including darker buildings and a river. In fact, the spot-metered clip begins brighter than the full-Auto scene at 4:15, and then ends-up darker than the full-Auto scene by the end at 4:33. That's not the desired outcome. So based on that particularly maneuver, spot-metering just isn't appropriate in this case.

Conclusion:

So with the X8-2018 camera, most scenarios are best shot in Auto ISO perhaps using manual Ev adjustments to influence the exposure, while some select scenarios are best shot in Manual ISO while relying on either full-frame or spot-metering to best maintain the desired exposure. Perhaps the bottom-line is that if you're having any trouble judging the exposure due to sun-glare on your FPV device, then Auto ISO is the way to go. During hi-contrast, golden-hour lighting, manual ISO might be necessary to ensure ISO 100, but you'll have to weigh the potential benefits of low ISO with the potential pitfalls of losing manual Ev with full-frame metering or relying on spot-metering to influence the exposure. If it's getting so dark that Auto ISO immediately looks better than manual ISO 100/200/400 then by all means transition back to Auto ISO unless you really know what you're doing.[/size]