p.8 #1 · Easy ETTR technique using the Zebra pattern
dclark wrote:
I agree that ETTR should be practiced to get the best possible SNR even if it is not possible to get to base ISO, but if base ISO cannot be achieved due to the limitations on light and other shooting requirements (min shutter speed and f/#) then the strategy to get the best SNR is changed. To get the best possible SNR we want as many photoelectrons as possible. That means as much light on the sensor as possible without exceeding the photoelectron capacity of any pixels (i.e. any pixels for which we can't accept blown highlights). ETTR is the method of achieving that. If we can get enough light on the sensor to get to saturating pixels it is necessary to be at base ISO to keep from saturating the ADC before the pixels are saturated. If the light level is low, minimum shutter speed is high, and DOF requires a large f/#, it may not be possible to get enough light on the sensor to get to the maximum capacity of the pixels. The usual ETTR/bracketing strategy is not possible, since it is not possible to increase the number of photoelectrons in the shadows.
What is the ETTR/bracketing strategy when base ISO cannot be achieved? The only remaining parameter that can be varied is the ISO. How should ISO be set? The best practice is the usual ETTR method, increase ISO without saturating the ADC. With an ideal "ISO-less" camera this does not yield any real improvement. Since the ideal ISO-less camera has not been produced, there is some improvement, although with the best cameras it is very small. I believe it is not wise to push too hard since the gain is small and the risk of blown pixels is unrecoverable. Is it useful to bracket by increasing the ISO to get better SNR in shadows? Again the gain is very small for the best cameras but there is some improvement. Usually I don't bother.
There is an exception. It is useful to study the input referenced noise curve for your camera (http://www.photonstophotos.net/index.htm). That tells you a bit about how close your camera is to ISO-less, but more importantly it tells you if the camera has more than one analog gain (actually a change in capacitance at the sense node). For example, the A7R2 and A7R3 change gain at ISO 640. So with those cameras it is helpful to get to 640. For example, if your base frame is at ISO 400, getting a second frame at ISO 640 will get a frame with improved SNR in the shadows. Capturing a third frame at higher ISO is probably a waste of time.
Thank you, this answers many of my questions. So there is marginal improvement, but not much. I'm guessing less than 0.3 stops? Is it worth it to get closer to the clipping point still in order to avoid color shifts? Or should I just stick to ISO 640?
p.8 #2 · Easy ETTR technique using the Zebra pattern
phototiimo wrote:
Thank you, this answers many of my questions. So there is marginal improvement, but not much. I'm guessing less than 0.3 stops? Is it worth it to get closer to the clipping point still in order to avoid color shifts? Or should I just stick to ISO 640?
The only time I have noticed color shifts is if I push too far and clip one of the colors. I am pretty cautious and keep the ISO low enough to be sure I don't clip. You may be referring to color shifts when pulling up shadows. If so, what you suggest may help color even though the the noise is about the same. Interesting idea. I need to experiment with that.
p.8 #3 · Easy ETTR technique using the Zebra pattern
dclark wrote:
The only time I have noticed color shifts is if I push too far and clip one of the colors. I am pretty cautious and keep the ISO low enough to be sure I don't clip. You may be referring to color shifts when pulling up shadows. If so, what you suggest may help color even though the the noise is about the same. Interesting idea. I need to experiment with that.
Dave
I was thinking both in terms of pulling down highlights and bringing up shadows. Let us know what you find! I'd be very interested to hear the results.
p.8 #4 · Easy ETTR technique using the Zebra pattern
Dumb question- do I need to set the exposure and then add the compensation for every bracket or can I just leave the comp wheel set to +2 and expose so that the zebras just disappear?
p.8 #5 · Easy ETTR technique using the Zebra pattern
mfenske wrote:
Dumb question- do I need to set the exposure and then add the compensation for every bracket or can I just leave the comp wheel set to +2 and expose so that the zebras just disappear?
Actually you don't need to add exposure compensation. Just set a custom level "lower limit" 107+ or 108+ (instead of 100+) and the zebras will exactly follow the RAW exposure. This if using the default camera settings, if you use picture profiles the required limit changes. Note that zebras only show the green channel.
p.8 #6 · Easy ETTR technique using the Zebra pattern
cgarcia wrote:
Actually you don't need to add exposure compensation. Just set a custom level "lower limit" 107+ or 108+ (instead of 100+) and the zebras will exactly follow the RAW exposure. This if using the default camera settings, if you use picture profiles the required limit changes. Note that zebras only show the green channel.
Cool! I'll give this a go with the bracketing tonight. Thanks!
p.8 #9 · Easy ETTR technique using the Zebra pattern
saxguy wrote:
Fred, have you compared this technique to just using Highlight Metering and adding +2 EV?
I have not played with this on the A7R IV yet but I think that using Zebra is still more accurate. As soon as you see any 'blinking' is fine adding +2EV but when using Highlight Metering sometimes +2 is too much. With the latter, I would go +1.3EV when shooting a high contrast scenes to be safe. Will update the thread when I test this further.
p.8 #10 · Easy ETTR technique using the Zebra pattern
Fred Miranda wrote:
I have not played with this on the A7R IV yet but I think that using Zebra is still more accurate. As soon as you see any 'blinking' is fine adding +2EV but when using Highlight Metering sometimes +2 is too much. With the latter, I would go +1.3EV when shooting a high contrast scenes to be safe. Will update the thread when I test this further.
Fred did you notice that you van set zebras to any custom level, like 170%, on the r4 so you can find a setting that really is equivalent to RAW overexposure...
p.8 #11 · Easy ETTR technique using the Zebra pattern
DavidBM wrote:
Fred did you notice that you van set zebras to any custom level, like 170%, on the r4 so you can find a setting that really is equivalent to RAW overexposure...
p.8 #14 · Easy ETTR technique using the Zebra pattern
DavidBM wrote:
I think the best bet is to try a bunch of settings, examine them in a RAW histogram program, and see what level corresponds to raw clipping...
My suspicion is that there is a little more room in most cases.
p.8 #15 · Easy ETTR technique using the Zebra pattern
cgarcia wrote:
Actually you don't need to add exposure compensation. Just set a custom level "lower limit" 107+ or 108+ (instead of 100+) and the zebras will exactly follow the RAW exposure. This if using the default camera settings, if you use picture profiles the required limit changes. Note that zebras only show the green channel.
109+ still shows zebras to early for me with RAw, including green channel. Adding negative contrast affects when the zebras show, but even -3 is too early. It's likely possible with PP but I don't want to grade my videos.
p.8 #16 · Easy ETTR technique using the Zebra pattern
nobody23 wrote:
@Fred@
With the 3rd Gen of Sony cameras we got 'Highlight' metering.
Renders your post #1 obsolete or what do you think?
I did a quick comparison at home and the highlight metering mode gives a far different exposure suggestion than what F M is suggesting. For example, all at f2.8, the camera's multi-metering mode suggests a shutter speed of 1/15s. The highlight metering mode suggests 1/25s. In both these cases, in the built-in camera histogram, the light distribution is about one full quarter away from the right wall!!! So I try FM method of +2ec after a 100 Zebra setting and the shutter speed is 1/4s and the built-in camera histogram approaches the right wall but doesn't quite reach it.
Can we trust the in-camera histogram? I am planning to do this comparison outside on a scenery and then, bring the actual photos under examination on the computer.
p.8 #17 · Easy ETTR technique using the Zebra pattern
raminolta wrote:
I did a quick comparison at home and the highlight metering mode gives a far different exposure suggestion than what F M is suggesting. For example, all at f2.8, the camera's multi-metering mode suggests a shutter speed of 1/15s. The highlight metering mode suggests 1/25s. In both these cases, in the built-in camera histogram, the light distribution is about one full quarter away from the right wall!!! So I try FM method of +2ec after a 100 Zebra setting and the shutter speed is 1/4s and the built-in camera histogram approaches the right wall but doesn't quite reach it.
Can we trust the in-camera histogram? I am planning to do this comparison outside on a scenery and then, bring the actual photos under examination on the computer.
p.8 #18 · Easy ETTR technique using the Zebra pattern
I have a very basic and perhaps naive question. When using zebras, I assume I'm going to disregard things that would normally show overexposure even when underexposing most of the image (i.e. reflections as points of light on curved polished objects, very bright pinpoint lights, etc.) when getting my initial baseline? In other words, get to the point where other than these areas, zebras just disappear - and then add back in +2 EV?
p.8 #19 · Easy ETTR technique using the Zebra pattern
Fred Miranda wrote:
In your example above. If one would add 2EV stops to the highlight metering reading (1/25s), the camera would be 1/6s.
I see you mean even when using the highlight metering mode, I would need to overcompensate by two stops. I didn't know. When I tried 'highlight metering mode' and noticed it underexposes, I just gave up on it and switched back to multi-metering mode. If we can universally over-expose by two stops, doesn't that mean Sony deliberately set up their metering modes to underexpose? If the answer is yes, why do they do that?
p.8 #20 · Easy ETTR technique using the Zebra pattern
raminolta wrote:
I see you mean even when using the highlight metering mode, I would need to overcompensate by two stops. I didn't know. When I tried 'highlight metering mode' and noticed it underexposes, I just gave up on it and switched back to multi-metering mode. If we can universally over-expose by two stops, doesn't that mean Sony deliberately set up their metering modes to underexpose? If the answer is yes, why do they do that?
Whenever possible, I still use zebras instead of highlight metering. The former has been more reliable to me when shooting landscapes and even under contrast lighting.
For general photography, adding 2EV to the highlight metering exposure may cause a slight highlight data loss under high contrast lighting so perhaps compensating 1 1/3 stop may be a safer bet.
