|
Forum: General Photography
09-04-2018, 07:24 PM
|
| |
Yes it is subjective. But there really are three distinct variables at play that affect three different dimensions of image quality:
[1/1000, f/8, ISO 100, Push 0]: more motion blur, less DoF, less noise
[1/1000, f/11, ISO 100, Push 1]: more motion blur, more DoF, more noise
[1/2000, f/8, ISO 100, Push 1]: less motion blur, less DoF, more noise
It comes down to trade-offs in motion blur, DoF, and noise as modulated by shutter speed, aperture, and either ISO gain or post-processing push.
|
|
Forum: General Photography
09-04-2018, 07:07 PM
|
| |
Actually, the exposure triangle still exists but the ISO axis could also be labelled the "post processing push" axis.
For example, a tiny sub-triangle of the full exposure triangle for EV=16 which might have these three possible settings that all yield the same image brightness levels:
[1/1000, f/8, ISO 100], [1/1000, f/11, ISO 200]
[1/2000, f/8, ISO 200]
Using ISO invariance with base ISO of 100 and pushing in post could re-define this triangle as:
[1/1000, f/8, ISO 100, Push 0], [1/1000, f/11, ISO 100, Push 1]
[1/2000, f/8, ISO 100, Push 1]
There's still a triangle because there are still the trade-offs of higher and lower shutter speeds with wider and narrower apertures and higher or lower ISO or push-in-post processing.
|
|
Forum: General Photography
09-04-2018, 04:26 PM
|
| |
The horse on the beach image illustrates ETTR more so than the merits of ISO invariance. If spot-metering the brightest part of the sky in TAv mode had given ISO 1600 as the "correct" exposure, exposing at ISO 100 would not have helped. But spot metering the dark horse would have resulted in a blown sky, center-weighted might have blown the sky, and who knows whether magical matrix metering might have done the right thing.
That said, exposing at base ISO is more likely to protect highlights than letting ISO float with TAv or other auto-ISO modes (unless the metered reading gives base ISO as the correct exposure).
Exposing at base ISO instead of metered ISO may help preserve highlights in some cases but not all. It's up the photographer to evaluate the scene and decide the appropriate amount of headroom for highlights. Exposing at base ISO does make chimping harder (with a risk of botching the image from not being able to easily review it) and there's always the risk that the amount of pushing might be too high and the photographer should have compromised on shutter speed or aperture to reduce the metered ISO or required post-process push.
|
|
Forum: General Photography
09-04-2018, 11:18 AM
|
| |
Yes, and the part of the curve where DR drops exactly 1 stop for each stop of higher ISO is the ISO-invariant part of the curve. And if the camera's entire curve loses 1 stop DR per stop of ISO, then the camera is known as ISO invariant.
The flat part of the curve or the part of the curve that falls less then 1 stop DR per stop ISO is the non-invariant part where an under-exposed-base-ISO-pushed image has worse final image DR than a properly-exposed-boosted-ISO image.
|
|
Forum: General Photography
09-04-2018, 10:10 AM
|
| |
It seems that the true definition of ISO invariance has been lost in all this An ISO-invariant camera is one that produces identical DR whether the ISO is boosted N stops in the camera or the shot is taken at base ISO and boosted N stops in post. It says nothing about whether boosted ISO or pushed base ISO image is better. In fact it says they are the same. And in both the boosted-ISO and pushed base ISO cases, you've lost N stops of DR.
Whether a photographer chooses to use their ISO-invariant camera at base ISO with pushing or at a boosted "correct exposure" ISO depends on balance the pros and cons of the two strategies:
Shooting at base ISO with pushing in post has the following pros & cons:
* PRO: highlights are more likely (but not guaranteed) to be preserved
* CON: the chimped image will be dark
* CON: the SOOC JPG will be dark
* CON: you need to check the amount of metered underexposure to know how much DR the sensor will offer in the post-processed image
Shooting at metered ISO has the following pros & cons:
* PRO: you can chimp the image
* PRO: the SOOC JPG is usable
* PRO: you know how much DR you'll get based on the metered ISO
* CON: you are more likely to blow the highlights
ISO invariance has nothing to do with superior DR. In both the boosted ISO image and the pushed base ISO image, you've lost the same N stops of DR.
In contrast, there are some cameras that are non-invariant where the ISO 3200 shot is somewhat cleaner than the ISO 100 shot pushed 5 stops in post. For those non-invariant cameras, it really does pay to more carefully pick the ISO (with the usual provisos about blowing the highlights). Whether a camera is invariant or not arises from the various types of noise in the pixel, analog read circuits, and ADC systems.
Even with invariance, the photographer still has to trade-off shutter speed, aperture, and either ISO-boosted or pushed DR. If picking the preferred high shutter speed (to stop motion) and the preferred narrow aperture (to get depth-of-field) implies needing to push 10 stops in post or use ISO 102400, then chances are the photographer will need to use a slower shutter or wider aperture to get a more acceptable balance between motion blur, out-of-focus blur, and image noise.
|
|
Forum: General Photography
09-03-2018, 07:19 AM
|
| |
Actually, the reason film handles over exposure is because of the statistical nature of the binary grains of silver halide. That is, each grain either develops to black if by statistical chance it was hit with photons or develops clear if it was not.
Imagine a piece of film with 100 grains of silver halide in it.
Now expose the film with just the faintest amount of light needed to get one grain in 100 to develop to black. Now there's 1 black grain , 99 clear grains.
Next, double the exposure so that 1 of the remaining 99 unexposed grains develops to black. Now there are 2 black grains , 98 clear grains.
Next, double the exposure a second time so that another 2 of the remaining 98 unexposed grains develops to black. Now there are 4 black grains, 96 clear grains.
Next, double the exposure a third time so that another 4 of the remaining 96 unexposed grains develops to black. Now there are 8 black grains, 92 clear grains.
Imagine increasing the exposure until about 50 of the 100 grains are developing to black and 50 remaining clear grains.
Next, what happens if you double the exposure again? What are the chances that the 50 remaining unexposed grains become black? The added light only ensures that each unexposed grain has another 50% chance of turning black so the result is that doubling the exposure only brings the black level from 50% black to 75% black. There are still 25 unexposed grains.
What happens if you double the exposure again to 4X the light needed to create a 50% black negative? Each of the remaining 25 unexposed grains has a 75% chance of turning black. Thus another doubling of the exposure only brings the black level from 75% black to about 94% black.
The point is that it takes exponentially more and more light to ensure that the few remaining unexposed grains become black. That implies that even the brightest highlights have some gradation in tonality.
In contrast, a silicon sensor is almost perfectly linear. Doubling the exposure when the sensor is already at 50% exposure level results in a 100% exposure level and saturation.
Note: I've greatly simplified the math, physics, and the chemistry for this. But the point remains that film has a nonlinear response to light which gives it extra latitude for over-exposure. Silicon sensors are linear but saturate which makes them unforgiving to over-exposure.
|
|
Forum: General Photography
08-30-2018, 05:49 AM
|
| |
The two biggest differences are:
1) the well depth which is the maximum number of electrons that the pixel can store. The more electrons the pixel can hold, the LOWER the base ISO.
2) the quantum efficiency which is the average number of electrons generated per incoming photon. The more efficient the sensor is at collecting photons and converting them to electrons, the HIGHER the base ISO.
|
|
Forum: General Photography
08-29-2018, 09:08 AM
|
| |
For diffusing water and ghosting crowds, interval composite (set to "average") can be a decent solution.
|
|
Forum: General Photography
08-29-2018, 08:17 AM
|
| |
The easiest way to have a really low base ISO is to lower the quantum efficiency of the sensor -- if the photons don't get converted to electrons, the sensitivity is really low. Of course, then the low-light performance suffers.
If you want a sensor with modern quantum efficiency (for good low-light performance) but don't want it to saturate in high light levels, then you need to increase the electrical capacitance of the pixels. That implies building a high surface-area capacitor into each tiny surface-area silicon pixel. For example, one could imagine fabricating tiny tantalum or super-capacitor elements on top of each pixel (to store al the extra electrons generated by longer exposures in bright light) and then flipping the sensor chip for back-side illumination.
P.S. Film has really low quantum efficiency -- about 1/25th that of silicon sensors -- which is why it has such low base ISO and such humongous grain for high ISO films.
|
