How much DR is enough?

[ahsanford]

*Perhaps whatever was done with the D810 and D850 sensors have a bit of 1, as evidenced by the ISO 64 setting, which allows more focal plane exposure than 100; other cameras fake the settings less than 100.

Can someone explain to me why Nikon can do this and others cannot? Why Can't Canon offer a 'real' ISO 50 or 64?

- A

 

[BillB]

Wider DR is always better. But whether you need it is a different matter. I recall seeing a beach sunset shot of a wedding couple with the setting sun behind them and the photographer had brought up the shadows on the beach - and due to the limitations of the medium (and that is the real problem) it looked flat and....pants.


Reading this thread, my mind wandered off the main question, and the first was that until a sensor has a dynamic range of 20+ stops, the question is not so much 'how much dynamic range' as comparing (for example) Canon to Sony how many situations can I shoot with a Sony that would not be possible with a Canon, and as far as I can tell in things like sunsets, or shooting to the light, you need to bracket with both bodies.


Also, wider dynamic range is reached by dropping the noise floor on the sensor which then means I can use higher ISO for high shutter speed in ever-darker conditions (great for sport and wildlife) so I see a more useful spin-off.

Not sure that works. The high Dr magic numbers are at base ISO. At higher ISO's it's a different ball game.

 

[stevelee]

OK, let’s say your super sensor has a 20 bit depth per pixel per color. Then what do you do with that? Photoshop has very limited 32-bit support, but maybe that will change by the time we have 20+ bit sensors. Monitors now are up to 10 or 12 bits, maybe? Even with 15 or more colors of ink printers will still be limited by the reflectance of the paper. We will always be mapping down throughout the process.

 

[Orangutan]

OK, let’s say your super sensor has a 20 bit depth per pixel per color. Then what do you do with that? Photoshop has very limited 32-bit support, but maybe that will change by the time we have 20+ bit sensors. Monitors now are up to 10 or 12 bits, maybe? Even with 15 or more colors of ink printers will still be limited by the reflectance of the paper.

We will always be mapping down throughout the process.

Why is that a problem?

 

[3kramd5]

OK, let’s say your super sensor has a 20 bit depth per pixel per color. Then what do you do with that? Photoshop has very limited 32-bit support, but maybe that will change by the time we have 20+ bit sensors. Monitors now are up to 10 or 12 bits, maybe? Even with 15 or more colors of ink printers will still be limited by the reflectance of the paper.

We will always be mapping down throughout the process.

Why is that a problem?

It isn't. That's... photography. It's a process of capturing light and transposing it to some permanent medium. Whether we're talking converting charge to brightness and displaying on an OLED, or using platinotype, we're mapping one thing to another.

 

[Larsskv]

How much you need depends on what you shoot, but one must admit that more is better.

I own the 1DXII and 5DIV, and recently added a 6D for manual focus shooting. So far I haven’t wanted or needed more DR in any of my 6D shots. When you push the files, the 1DXII and 5DIV is better, but I rarely need the advantage they offer.

The biggest advantage the 1DXII and 5DIV offers over the 6D is higher DR at higher ISO. This is much more useful to me than the advantage at ISO 100.

 

[Aglet]

Still looks too flat when you're done?... You can fix that.

The problem is that it looks unnatural. "Photoshopped".

I don't mean to sound trite here but..
If you adjusted it ,and it doesn't look right...
Then maybe you're not done adjusting it yet.

You can do more with more data than you can with less data.

You need to tweak it properly to get the result you want. Sometimes you have to mangle a tone (transfer) curve adjustment to do that and then add the right amount of local contrast enhancement to alleviate that "flatness" that can occur when you compress a wide capture range into a smaller display range.

Using appropriate software helps and editing experience will speed up the process as you learn how to achieve the look you're after.

 

[BillB]

Re: How much DR is enough?or

How much you need depends on what you shoot, but one must admit that more is better.

I own the 1DXII and 5DIV, and recently added a 6D for manual focus shooting. So far I haven’t wanted or needed more DR in any of my 6D shots. When you push the files, the 1DXII and 5DIV is better, but I rarely need the advantage they offer.

The biggest advantage the 1DXII and 5DIV offers over the 6D is higher DR at higher ISO. This is much more useful to me than the advantage at ISO 100.

So, is more DR always better or is it always at least as good?

 

[Orangutan]

Re: How much DR is enough?or

How much you need depends on what you shoot, but one must admit that more is better.

I own the 1DXII and 5DIV, and recently added a 6D for manual focus shooting. So far I haven’t wanted or needed more DR in any of my 6D shots. When you push the files, the 1DXII and 5DIV is better, but I rarely need the advantage they offer.

The biggest advantage the 1DXII and 5DIV offers over the 6D is higher DR at higher ISO. This is much more useful to me than the advantage at ISO 100.

So, is more DR always better or is it always at least as good?

Pros:

• More data to work with (you can't recover data that don't exist)

Cons:

• More data to work with (even if you didn't need it, you still have to process it out)
• Tiny amount more storage space required (assuming standard mapping of brightness to binary values)

Conclusion: if you wouldn't complain about "too much DR" when you work in your studio under controlled lighting, then it's always better.

 

[Kit.]

Still looks too flat when you're done?... You can fix that.

The problem is that it looks unnatural. "Photoshopped".

I don't mean to sound trite here but..
If you adjusted it ,and it doesn't look right...
Then maybe you're not done adjusting it yet.

Or maybe it's technically impossible for your particular scene to do tone compression that looks "natural", and you need to return to your scene and reshoot it in a different light.

Using appropriate software helps and editing experience will speed up the process as you learn how to achieve the look you're after.

The "look I am after" is usually the original look. Uncompressed.

As a compromise, it could be compressed in the areas that are the obligatory "background" part of the composition, but otherwise are not interesting (like blue sky), and the compression doesn't produce visible artifacts.

Of course, it's not the only property of the "look I am after".

Not "decades", but more than a century, and not from "cameras", but from scenes. They taught us that good scenes need good lighting.

Decades, yes; more than ten of them. And yes, a good scene needs good light. What has that to do with what a camera can record?

The topic.

DR is theoretically limited by photon noise (Heisenberg's uncertainty). If you are saying that "the more the better", so there is no DR that is "enough", then, unfortunately, the photography in this world is not for you.

I am saying that sensor's DR is not the only limiting factor in the tonal range of the scene that needs to be represented, so there is some (scene-dependent and use-dependent) "enough" for DR.

I mentioned cameras because we are conditioned to evaluating photographs taken with devices which can not record the entire range of tones in a scene. I can look at the wall by a window and see both foreground shadow detail and background highlight detail.

Actually, no, not at the same time.

I've just checked: the ugly LCD of my old 5D2 represents about the same level of dark foreground detail as I see through the OVF when looking at the background highlights. If I point my eyes on dark foreground shadows, I lose the background highlight detail.

If you do photo murals, a higher (linear) dynamic range could be desirable, if your output media allows. If you make a picture to be seen as a whole, it is probably an overkill anyway.

However if a the range in a photo looks unrealistic, it’s due to technique.

But it could be as well due to the photographer's technique, not to the darkroom technician's technique.

The photographer is the person whose work ends with pressing the shutter button.
The darkroom technician is the person whose work then just starts.
They can be the same person, but not necessarily so.

 

[Keith_Reeder]

Re: How much DR is enough?or

So, is more DR always better or is it always at least as good?

Nope. "as much as you need" is better.

Why worry about "more" if you aren't going to use it?

 

[sdz]

I understand that there are instances when you need more DR, but I have almost never encountered one using my 6D and now my M5. I have tried the Sony FF A7 II - and found that in my landscapes (compared to my 6D) there was no benefit and perhaps a detriment. Compared to my old 300D, I now find that I have to post process more and add contrast to get my photos the way I want, again realizing each person has their own opinion on how a photo should look. I consider contrast to be far more important than a wide DR range in the majority of my shots. As the DR increases, I see so many dull, washed out photos that have a lack of contrast.

Coming from an art and painting background, I may be looking at things a bit differently than the mostly "tech" crowd on this forum. One general rule of painting is that the human eye does not simultaneously see details in both the light areas and the shadows, thus one should not be painting detail in both. Either the light areas have detail and the shadows are essentially dark and featureless or vice versa. (emphasis added) Yes, I understand that you don't want those shadows to be totally black in most cases, but a bit of shadow lifting is easily done with all Canon cameras that I have used. But I don't want anything close to the same level of visible detail throughout the value range of the photo. In my opinion, that is why too much DR - and especially the DR of HDR photos - is often a negative and the end results look fake, washed out, and rather boring. Again, that is just my opinion.

I have found it interesting that the desire for more DR is so overpowering that even the reviewers fail to see the negative side. A number of years ago I remember reading a camera review of a new model. The reviewer complimented the camera for adding a stop or more of DR and then in another part of the review lamented the fact that photos from the newer version seemed to have lacked the "punch" of the previous generation. They failed to make the connection that the lack of punch was directly related to the lesser contrast and the greater DR that was now available on the newer model.

Yes, I understand that with enough post processing one can make use of the greater DR and then add more contrast. But having tried the "better" (more DR) Sony sensor and compared the results with identical shots taken with my older sensor generation 6D, I preferred the look of the 6D shots. So more DR is not anything I care about or am looking forward to. Not in the least.

The implicit standard used when evaluating sensor tech is: The best photo will accurately reproduce 'what exists.'

Hence, the demand for more DR.

This standard is unrealizable for every being but God and thus an illusion in principle. Like a painting, even a photo realistic painting, a photo is a two-dimensional representation of an object or scene that exists in many dimensions. They are always wrong (inaccurate) when compared to the illusory standard which requires perfection. This means that any photo merely selects from the world as given a bit of what exists to represent what cannot be represented. This also means that one can create superb photos using Canon 1, 5 and 6D cameras with their technically inferior sensors.

That said, I believe Canon could stand to increase the effective dynamic range of their better cameras!

 

[3kramd5]

The topic.

DR is theoretically limited by photon noise (Heisenberg's uncertainty).

Agreed. The more light you record, the more shot noise you encounter.

If you are saying that "the more the better", so there is no DR that is "enough", then, unfortunately, the photography in this world is not for you.

I'm saying "the more the better" in the capability of equipment. There is a point at which equipment can no longer improve in one dimension or another, and that's fine.

I am saying that sensor's DR is not the only limiting factor in the tonal range of the scene that needs to be represented, so there is some (scene-dependent and use-dependent) "enough" for DR.

Yes, of course. If shooting a scene which can be entirely recorded with a 5eV system, 5eV is enough, and 15eV is not better for that scene. But it's better that your system be capable of 15eV rather than 5eV, in case you want to image a scene which requires it.

Would it be better photographically to reduce the range necessary by shooting at a different time or filling in your own light? Sure, but that's not always possible.

I mentioned cameras because we are conditioned to evaluating photographs taken with devices which can not record the entire range of tones in a scene. I can look at the wall by a window and see both foreground shadow detail and background highlight detail.

Actually, no, not at the same time.

What do you mean? That my brain is doing some processing? Sure, but it's seamless and fast enough that I perceive them simultaneously.

However if a the range in a photo looks unrealistic, it’s due to technique.

But it could be as well due to the photographer's technique, not to the darkroom technician's technique.

The photographer is the person whose work ends with pressing the shutter button.
The darkroom technician is the person whose work then just starts.
They can be the same person, but not necessarily so.

Yes, that's better stated.

 

[zim]

The thing I really want is less noise at high iso. Is that linked to DR?
I'm quite happy with the malleability of my raw files as is with regard to push/pulling shadows and highlights.

 

[zim]

The topic.

DR is theoretically limited by photon noise (Heisenberg's uncertainty).

Agreed. The more light you record, the more shot noise you encounter.

so I want a camera with less DR?

 

[stevelee]

The thing I really want is less noise at high iso. Is that linked to DR?
I'm quite happy with the malleability of my raw files as is with regard to push/pulling shadows and highlights.

From what I gather here, there seems to be some tradeoff between DR at base ISO, such as 64 or 100, vs. less noise at higher ISOs. A lot of the bad press that the 6D2 has received is because it seems to have lost a stop or maybe part of one in DR at ISO 100 in order to have its fine results at higher ISOs. I don't know that that is true. I rarely shoot anything at ISO 100, and I'm enjoying the quality of pictures I get from over 2000 ISO, so if there is a tradeoff, it's been a good one for me.

 

[3kramd5]

The topic.

DR is theoretically limited by photon noise (Heisenberg's uncertainty).

Agreed. The more light you record, the more shot noise you encounter.

so I want a camera with less DR?

Not at all. What that meant was: You can overcome the noise introduced by the electronics by increasing the well capacity (point beyond which you clip), but you can not overcome the noise inherent to the light.

 

[stevelee]

Depends upon the cat in the box.

 

[Mikehit]

Not sure that works. The high Dr magic numbers are at base ISO. At higher ISO's it's a different ball game.

But with lower noise you can use higher ISo with less noise. Whether those high ISOs have more DR is a different question.

 

[Kit.]

DR is theoretically limited by photon noise (Heisenberg's uncertainty).

Agreed. The more light you record, the more shot noise you encounter.

Actually, that part of the shot noise belongs to the lighter pixels, while people are more interested in darker pixels when they are talking about sensor's DR.

The effect of the larger well depth for a sensor pixel is that it allows you to shoot at lower base ISO, increasing the photoelectron count and thus decreasing the relative shot noise for darker pixels. However, that only works if you can decrease your ISO to the base one. Which is not always the case. When you shoot birds in flight, for example, you usually cannot.

If you are saying that "the more the better", so there is no DR that is "enough", then, unfortunately, the photography in this world is not for you.

I'm saying "the more the better" in the capability of equipment. There is a point at which equipment can no longer improve in one dimension or another, and that's fine.

If you approach the problem as a camera engineer or a gear-obsessed camera enthusiast, it is fine. No one else is going to have any better, no matter how they try.

But if you approach the problem as a photographer, "fine" lies at a different point. You look at the scene and either you are going to press the shutter button or not.

When it comes to DR, you are not going to press the shutter, because of one (or more) of the following reasons:
1. The DR of the scene as would be "enough" for you to capture it is beyond physical limits of the photography;
2. The DR of the scene is wider than the one you would like to be reproduced;
3. The scene is repeatable and someone with higher DR gear may capture it better, making your shot irrelevant.

Only the 3rd case is really about "the more the better" in relation to the subject's question.

I am saying that sensor's DR is not the only limiting factor in the tonal range of the scene that needs to be represented, so there is some (scene-dependent and use-dependent) "enough" for DR.

Yes, of course. If shooting a scene which can be entirely recorded with a 5eV system, 5eV is enough, and 15eV is not better for that scene. But it's better that your system be capable of 15eV rather than 5eV, in case you want to image a scene which requires it.

The scene that requires 15 EV to capture is not always worth reproducing (the case 2 above).

Of course, if your scene only needs 5 EV and 15 megapixel to capture in order to be worth reproducing, it might still be nice to capture more information (both spatial and tonal) from the scene than you currently need, but still 5 EV of your camera is enough to press the shutter.

(and by the way, the binary logarithm of exposure is EV. eV is "electron-volts", a measure of energy)

I mentioned cameras because we are conditioned to evaluating photographs taken with devices which can not record the entire range of tones in a scene. I can look at the wall by a window and see both foreground shadow detail and background highlight detail.

Actually, no, not at the same time.

What do you mean? That my brain is doing some processing?

That the picture reproduced in the same tonal range would look ugly, at least at first glance.

In the real world, the details in the dark foreground and the details in the light background are parts of different scenes, and you don't need to be able to match them together. In the picture, they are parts of the same composition.