ISO invariance, ETTL, and Uni-WB

A forum to ask questions, post setups, and generally discuss anything having to do with photomacrography and photomicroscopy.

Moderators: rjlittlefield, ChrisR, Chris S., Pau

RobertOToole
Posts: 2627
Joined: Thu Jan 17, 2013 9:34 pm
Location: United States
Contact:

ISO invariance, ETTL, and Uni-WB

Post by RobertOToole »

Admin note: This thread has been split from another thread, Need help with strobes, as it is branching into a new set of topics.

--Chris S.
----------------------------

Chris S. wrote:.......
3) Turn your flash units on, set them to manual, and set up whatever triggers you use. Adjust flash intensity by taking pictures and examining the histogram (apply the ETTR method).....--Chris S.
Good advice Chris and agree with everything, but it should be added that exposure method really depends on the camera sensor.

ETTR is no longer needed with new cameras.

The ETTR method will improve shadow noise with older Nikons, most Canon, and Nikon D5, or D4.

Underexposing or ETTL with most, not all, new cameras, Nikon D500, D850, D810, D7200, Sony A7R, A7RII, A7RIII, Sony A6300 is the better method. The sensors on these cameras are ISO-Invariant so its safe to underexpose 0.3 to 0.7 all the time with any increase in shadow noise or DR penalty.

When I give tech presentations I show two slides taken of the same scene with the same shutter speed and aperture with D500 at ISO 400 and ISO 1600. One of these has to be darkened in Photoshop to match the other but the noise is identical.

All the best,

Robert
Last edited by RobertOToole on Fri Jan 05, 2018 8:48 am, edited 1 time in total.

mawyatt
Posts: 2497
Joined: Thu Aug 22, 2013 6:54 pm
Location: Clearwater, Florida

Post by mawyatt »

Chris & Robert,

Good advice all around. I generally use ETTL (slightly to the left) now and watch the red channel in the histogram carefully as this tends to be the first to saturate for my lights, setup and subjects.

Best,

Mike

rjlittlefield
Site Admin
Posts: 24009
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

RobertOToole wrote:The sensors on these cameras are ISO-Invariant
The term "ISO-Invariant" was new to me, so I did a search and studied the explanation at https://photographylife.com/iso-invariance-explained .

As I read that article, what "ISO Invariant" really means is that it doesn't matter what ISO value you set, after you've decided how much light to drop onto the sensor. However, it remains important to drop as much light as possible onto the sensor, so as to minimize the impact of photon noise and front-end read noise.

With that understanding, the simple recommendation continues to be "Set the camera on base ISO, then adjust the exposure to be as bright as possible without risk of overflowing any channels at any pixel positions."

Does that sound right?

--Rik

mawyatt
Posts: 2497
Joined: Thu Aug 22, 2013 6:54 pm
Location: Clearwater, Florida

Post by mawyatt »

rjlittlefield wrote:
RobertOToole wrote:The sensors on these cameras are ISO-Invariant
The term "ISO-Invariant" was new to me, so I did a search and studied the explanation at https://photographylife.com/iso-invariance-explained .

As I read that article, what "ISO Invariant" really means is that it doesn't matter what ISO value you set, after you've decided how much light to drop onto the sensor. However, it remains important to drop as much light as possible onto the sensor, so as to minimize the impact of photon noise and front-end read noise.

With that understanding, the simple recommendation continues to be "Set the camera on base ISO, then adjust the exposure to be as bright as possible without risk of overflowing any channels at any pixel positions."

Does that sound right?

--Rik
With that understanding, the simple recommendation continues to be "Set the camera on base ISO, then adjust the exposure to be as bright as possible without risk of overflowing any channels at any pixel positions."

Rik,

Yes, that's generally now what I do, except lean slightly to the left on exposure to keep a little margin from saturation. With modern sensors that have really good dynamic range I see no reason to try and "squeeze" the last bit of level from each pixel.

I used to ETTR as the norm and did run into some red channel saturation. The chip images I normally do have a lot of bright metal areas that can easily saturate pixels, and some have solder balls with very bright reflections. I would just look at the luminance histogram on a typical single shot and it would look OK, then go with that exposure for the stack. Many times I would find pixel saturation, and upon further investigation especially in the red pixels. With very shallow DoF and sliding across the chips surface at an angle the individual exposure can change in the in focus areas from one side to the other and possibly shift to the right enough to saturate a few pixels. Anyway, this was/is my thinking, so I've decided to lean a little to the left.

Best,

Mike

RobertOToole
Posts: 2627
Joined: Thu Jan 17, 2013 9:34 pm
Location: United States
Contact:

Post by RobertOToole »

rjlittlefield wrote:
RobertOToole wrote:The sensors on these cameras are ISO-Invariant
The term "ISO-Invariant" was new to me, so I did a search and studied the explanation at https://photographylife.com/iso-invariance-explained .

As I read that article, what "ISO Invariant" really means is that it doesn't matter what ISO value you set, after you've decided how much light to drop onto the sensor. However, it remains important to drop as much light as possible onto the sensor, so as to minimize the impact of photon noise and front-end read noise.

With that understanding, the simple recommendation continues to be "Set the camera on base ISO, then adjust the exposure to be as bright as possible without risk of overflowing any channels at any pixel positions."



Does that sound right?

--Rik
Yes Rik. That is why I recommend on new cameras to set the EV at -0.3 to 0.7 from ideal all the time just to make sure there is no overexposure.

Correct also that exposure dictates shadow noise levels, not ISO setting!

One of my favortie sources for this info:

http://www.photonstophotos.net/index.htm

The site author's name is Bill Claff.

This graph shows Shadow noise improvement vs ISO setting.

The D850 is ISO-Invariant from 400 and up. The Canon D80 is well, best to look at the graph.

http://www.photonstophotos.net/Charts/P ... kon%20D850

Image

Robert

RobertOToole
Posts: 2627
Joined: Thu Jan 17, 2013 9:34 pm
Location: United States
Contact:

Post by RobertOToole »

mawyatt wrote: I used to ETTR as the norm and did run into some red channel saturation. The chip images I normally do have a lot of bright metal areas that can easily saturate pixels, and some have solder balls with very bright reflections. I would just look at the luminance histogram on a typical single shot and it would look OK, then go with that exposure for the stack. Many times I would find pixel saturation, and upon further investigation especially in the red pixels. With very shallow DoF and sliding across the chips surface at an angle the individual exposure can change in the in focus areas from one side to the other and possibly shift to the right enough to saturate a few pixels. Anyway, this was/is my thinking, so I've decided to lean a little to the left.
Its not the sensor Mike, as far as I know its the camera's processing of the images. The processing engine in modern cameras amplifies, not sure if this is the proper term, the red channel. This means the histogram is wrong.

Some people actually run around with red filters on their cameras to counter the red saturation. Or the other solution, brings up another interesting technique, using Uni-WB.

With this you set a custom WB to level out the RGB channels and tame the red channel by setting your WB on a green uni-WB target image. This forces the cameras processing engine to cut on green channel so the RGB channels even out so the histogram is much more accurate, no more boosted reds.

The down side is green review images, you can set the camera to B&W to help, but the upside is the histogram is accurate and helps with the red channel problem.

Its been a long time since I have used Uni-WB, hope I got that right!

This is a good link for the histogram issue with the red channel:

Using a standard WB and checking the histogram is not an accurate tool to judge overexposure, Uni-WB is one way to solve the issue.

https://www.rawdigger.com/howtouse/beware-histogram

Robert

Post Reply Previous topicNext topic