Tony Northrup has posted a video discussing the exposure term ISO and has deemed it “totally fake”. It's an interesting take on the use of this legacy term in digital photography.
Tony summarizes:
- It's just a post-processing instruction; generally, it doesn't physically change anything about the picture taking process, unlike shutter speed and aperture.
- While the term “ISO” refers to a real standard, camera manufacturers don't comply with that standard, like, at all. Many cameras are more than a full stop off the proper ISO. For that reason, light meters simply won't work as you expect them to. Using the same settings on multiple cameras will yield very different results.
- High ISO could be eliminated completely; it's just an arbitrary limit to a multiplication problem.
- Low ISO could be eliminated by simply using image averaging, allowing us to get long exposures in-camera without using an ND filter. This would also reduce noise in good conditions.
The truth is unless your sensor is iso invariant, which few are and no Canons are, then the iso setting is not irrelevant so the premise is immediately and obviously faulty.
I liked the subject of the video, it's an interesting take. I'm more interested in the reactions I've read about it.
tried make 5 stop underexposed 100 ISO pic as 1200 ISO pic . they were from different planet
Most Canon cameras are ISO invariant above 6400. You should always expose to the right until you hit that point. Once you hit your ISO invariance point, maximize as much light as you can for the situation (aperture, shutter, external light)while keeping the ISO at that invariant number(unless if there's different noise patterns at different numbers. Some cameras have minor differences at different ISO settings, even if they're ISO invariant).
Some Sony sensors used in Nikon and Sony have two different levels of ISO invariance. One below 640 ISO and one above. ISO and ETTR still matters to a certain extent.
It's also worth pointing out that an ISO invariant sensor doesn't automatically mean it's better than one that depends on ISO. The Canon 6D is a great example for astrophotography, it still hangs in with today's crops of cameras in certain scenarios.
I'm not sure it would yield the same "graphical" results of a long exposure. Anyway, that's just a single use case for low ISO.
That suggestion works similar to Google's Night Sight, which is extremely impressive. Like you said, it's still a tool that has limitations and isn't a one stop solution. I'm also wondering if the wording is meant to say high ISO could be eliminated instead of low ISO. The Night Site I'm using normally shoots at base ISO 50.
Edit:
It appears that Google's Night Site is probably a modification and early adaption of Sony Semiconductor's tools they're offering with some of their newer sensors. It says right in the marketing material that it's using a few frames at 60 fps. The shutter at the most would then be 1/60th.
https://www.sony-semicon.co.jp/products_en/IS/cmos/index.html
I remember that the ISO standard about the negatives specified a particular optical density of the film that shall be achieved during exposure (and then standard processing). Which actually led to the situation where low-contrast film was being slightly underexposed at their standard ISO.
So, ISO, in a sense, was already "fake" in the film era.
IMHO it's on both sides - low ISO + ND filter to extend the exposure, or high ISO for low-light capture. As image averaging started IIRC in astrophotography, it's probably more common to think about the latter than the former - where you may want smooth water/clouds and "remove" moving elements - it could work, albeit I'm not sure if they would be enough alike.
What makes me dubious is you would be "prisoner" of algorithms upon which one may have little control - and if artifacts appear, it may be difficult to get rid of them. How a long exposure/ND filter work is much more deterministic. Sure, if you perform the averaging/merging yourself you may have far more control, but it's also a bigger effort, and would increase the storage needs.
2., The higher the ISO (and the smaller the sensor) the more the noise therefore it DOES effect your photos,
3., ISO also effects shutter speed that also effects your pictures since at shoter shutter speeds you have to worry less about blurry pictures.
4., ISO speed is part of the exposure triangle – see above – so regardless the value it will affect your picture.
Most Canon sensors are isoinvariant above iso640, not 6400. Maybe you made a typo.
Right now, a exposure at ISO100, f/8, 1/125 sec still produces a good exposure. If you boost or reduce the exposure, it will be incorrect. You may be able to correct that error by digital manipulation of the image file, but the ISO value to get a correct exposure in camera still works.
The issue is how to rate the ISO of a digital sensor when it contains lots of noise processing circuitry that distorts the conventional rating method of looking at the output of the sensor.
So, ISO still works, but, as in film, changes that reduce inherent noise allow you to boost the apparent ISO, it was done in film by longer processing times, it can be done in digital by changing exposures and digitally manipulating the resulting files, and eventually, a new rating will come out.
Yes, and film was also marketed with higher ISO ratings, but was the same exact film but with longer processing times. A similar bunch of comments were made.
The better Canon sensors are isoinvariant above about iso160-200, which covers much of their range: see
http://www.photonstophotos.net/Char... X Mark II,Canon EOS 5D Mark IV,Canon EOS 80D
for 5DIV (=EOS R), 80D, 1dXII
Personally I feel I owe him a lot.
I think he is a great communicator and pitches things at just the right level - at least for me - I'm currently doing an honours degree in maths and physics and rarely feel that Tony is being patronising or completely dumbed down ... but equally he is pitching for a wide audience and definitely not for camera technicians.
Like all good teachers he has to "pitch" the level of what he is teaching, and inevitably that involves making some simplifications - which are essentially inaccuracies.
Keep up the good work Tony !