[ZombieArmy]

If you are shooting your camera at iso 100 or iso 1600 and neither you are overexposing your image you will capture the same DR.

This is wrong I'm fairly sure. Shooting at a higher ISO will limit the amount of detail recovery in shadows no matter what. If you capture the same scene at 100 and 1600 with no need for any recovery the scene itself was low on the dynamic range to begin with.

[Ian Stuart Forsyth]

Shooting at a higher ISO will limit the amount of detail recovery in shadows no matter what.

If you are using the same exposure for both the iso 100 and iso 1600 shot they will have the same amount of shadow recovery, what you will lose is how much highlight you can store, but some cameras what are not iso invariant may well increase the DR you are able to capture for that exposure.

[ZombieArmy]

If you are using the same exposure for both the iso 100 and iso 1600 shot they will have the same amount of shadow recovery, what you will lose is how much highlight you can store, but some cameras what are not iso invariant may well increase the DR you are able to capture for that exposure.

I can test this easily tomorrow. Highlight recovery is terrible at any ISO on digital as far as I'm aware.

[Ian Stuart Forsyth]

With most Digital this can be between 1 and 2 stops depending on the camera

K7 nearly 1 stop

[Jonathan Mac]

Using the lowest possible ISO is generally your best bet but I believe that some (newer) sensors are dual-gain in which they essentially have two native ISOs. I think they are usually the base ISO (100-ish) and 640.

I remember that it is often better to use 640 as the second native ISO than use, say 500, which is boosted from 100. I don't recall the science behind it but it certainly sounds like a good idea.

[GUB]

If you are using the same exposure for both the iso 100 and iso 1600 shot they will have the same amount of shadow recovery, what you will lose is how much highlight you can store, but some cameras what are not iso invariant may well increase the DR you are able to capture for that exposure.

I feel photons to photos has it right Ian. About one stop of DR lost for every stop of iso.
I repeated that experiment I presented earlier with the camera set to iso 6400 and starting with what I felt to be a 100.100.100 exposure, stopping down from there.
I created a similar swatch to the first one above but no matter how I manipulated it I couldn't pull the dark end up and maintain the tonal range in the bright end. I considered there was a maximum DR of 6 stops.

[Ian Stuart Forsyth]

I feel photons to photos has it right Ian. About one stop of DR lost for every stop of iso.

Where d you think that DR is lost is it in the darker end of what the sensor records or is it in the highlight range of what can be stored
Image comparison: Digital Photography Review

---------- Post added 12-12-2021 at 03:44 AM ----------

stopping down from there.

Well if you are stopping down you are decreasing the exposure

[BarryE]

DR (dynamic range) is often confused with noise level, and that is due to the definition that has been given by camera sensor test labs (DXO etc) for marketing purpose. In practice, I get more useful dynamic range if I under-expose as much as it is required until highlights aren't clipped. Also practically, 8 stops of exposure latitude is more than enough because human eyes are much less sensitive to details in zone I & II of the zone system (near black and deep shadows). What matter in a photograph is that the key subject elements of an image receive more light than other parts of the photograph so that to lead the eyes on what's important. Noise and details in shadows play a much smaller role in making photographs an impact, that's why I say that the lab definition of dynamic range is mostly relevant to sell cameras. Typically, if your subject matter is in shadows and what's non-subject is in the light, no camera in the world will be able to correct for that bad configuration. OTOH, if the subject is lit properly and the rest is in shadows, 6 stops DR is enough.

Yes, in some (many) circumstances. However, take a winter woodland scene on a bright day, not even sunny. This is the environment I shoot in for a good portion of my photography. The more DR I can have the better. Yes, bracketing helps, but it's tricky handling tree branches as they obscure a bright sky.

Another challenge with DR (on K-1 which was is good) is when I shot real estate. Agents love sunny days, their photographers don't. Sure portrait (studio), still life and general shots a more limited DR is fine, but for me more is definitely better.

Of course what you say about subject getting the light etc is key, even in woodland photos where the subject is not always obvious so my friend is Dodge and Burn to lead the eye and control the feel of the photo.

[Ian Stuart Forsyth]

With newer sensors that have DCG if you are not clipping any of your highlights at the iso setting where it kicks in then you are better of using it


Here we can see that we have an increase of DR even when for the higher iso where DCG being used, it is even capturing the same DR as the iso 500 but with a exposure that is nearly half of what the camera can record at iso 800

[GUB]

Where d you think that DR is lost is it in the darker end of what the sensor records or is it in the highlight range of what can be stored
Image comparison: Digital Photography Review

---------- Post added 12-12-2021 at 03:44 AM ----------



Well if you are stopping down you are decreasing the exposure

Well the information was in there in the dark end to create a tonal range.And in the original swatch (which I have lost) because it started at 100.100.100 the bright end steps were there. But it just didn't seem possible to get it all lined up in the one range. This was the result showing the dark end pulled up.
If you look back at the original test at say column 12 in the adjusted row (bottom) you can see (roughly) that tonally it corresponds to about row 6 on the top row. So it has been pulled up about 6 stops and (courtesy of iso invariance) represents an iso 6400 image.
Now think about this with the corresponding column for the iso 6400 test. That same column ends up having to be iso 200,000 - ie maxed out. I suspect that has something to do with it.
As usual I am just trying to take a practical approach to things - but it is surprising how close I ended up to photons photos result.
EDIT Note with this image there is 8 steps on the left that are clipped out so invisible.

[GUB]

Well if you are stopping down you are decreasing the exposure

Yes and so long as you do this with shutter or aperture the sensor just thinks it is a darker subject. So I am presenting the sensor with a 14 stop tonal range.

[Cerebum]

Using the lowest possible ISO is generally your best bet but I believe that some (newer) sensors are dual-gain in which they essentially have two native ISOs. I think they are usually the base ISO (100-ish) and 640.

I remember that it is often better to use 640 as the second native ISO than use, say 500, which is boosted from 100. I don't recall the science behind it but it certainly sounds like a good idea.

I was going to ask about this. When it comes to charts and interdependent figures I have all the technical acumen of a dead moose, but looking at the first graph, it suggests the KP accelerator chip, when it kicks in at iso800, boosts the performance back up to near iso100 levels. Does this mean on the KP we should skip isos 400-640 and go to 800 where everything is lovely again? Or have I missed several important steps, again

[GUB]

Yes that is the theory - think of iso 800 as a sweet spot. I am currently using the K-1 original that doesn't have this step. But when I upgrade (?) I figure where I currently utilise iso invariance and stay at 100 I will make a choice to step to iso 800 if (and only if ) I figure the image is going to require that much boosting in editing anyway and has a challenging DR.

[Jonathan Mac]

I was going to ask about this. When it comes to charts and interdependent figures I have all the technical acumen of a dead moose, but looking at the first graph, it suggests the KP accelerator chip, when it kicks in at iso800, boosts the performance back up to near iso100 levels. Does this mean on the KP we should skip isos 400-640 and go to 800 where everything is lovely again? Or have I missed several important steps, again

I think that's theoretically the best way in terms of dynamic range, but if the accelerator chip in the KP does the same as that in the K-1 II then you may be risking detail retention. I haven't used either so haven't looked into it but some were of the opinion that the irreversible noise handling from the accelerator chip is heavy-handed.

[UncleVanya]

I think that's theoretically the best way in terms of dynamic range, but if the accelerator chip in the KP does the same as that in the K-1 II then you may be risking detail retention. I haven't used either so haven't looked into it but some were of the opinion that the irreversible noise handling from the accelerator chip is heavy-handed.

The reality seems to be far from the fear. My images don’t seem to suffer. My other camera is a Sony A7RIII and the KP does A creditable job keeping up. (And 24mp is higher density than 42mp cropped (18mp).