Dynamic Range - What's the trade off?

[Manfred M]

Not just the Fuji crowd, either. The Sigma crowd, with their latest Foveon Quattro-sensor cameras, are declaring OOC JPEGs to be as good, if not superior to X3F raw.

Pure heresy, of course.

Especially to a Quattro non-owner. Sigma SD1 Merrill Raw vs. Embedded JPEG:



After no work at all on the embedded image and hours of post-processing in RawTherapee, it should be obvious to all that raw is superior.

Or is my comparison just a tiny bit unfair?

I have never understood how throwing away 6 bits of data (going from 14-bits in a modern raw file to the 8-bit jpeg) can be achieved without losing something. Why not just built 8-bit sensors and increase the data throughput to some ludicrous data rate?

[John 2]

Which Fuji do you have?

You should find the DR Mode in the Q Menu and so long as you have the ISO on Auto or high enough to trigger it the setting is easily switched between options. If you are shooting street (where the Fuji excels) and are taking advantage of the film simulations then the extended DR is a great asset to getting the most from the jpeg engine.

Its dark here now but tomorrow if I get a chance and the weather isn't as miserable as it was today I'll try to show the DR Modes in action.

Robin,
I invested in a Fuji system in the form of an X-Pro1 that I eventually traded in for an XT-1 because the ergonomics were so much better. Also, at the time, their lenses and the sensor performance of the X-Tran II sensor were class leading. Sensor wise, I don't think that is true any more. Other manufacturers seem to have closed the gap. I am not interested in the X-Pro2. Ergonomics again plus I have no use for an OVF. However, I'll keep a close eye on the performance reviews of the X-Pro2 once the initial hype settles down because they should read across to the XT-2.

[Black Pearl]

Similar here, I don't like the X-Pro bodies (I'm weird when it comes to cameras) but love my X-T1. Not interested in the 2 when it arrives as I only got mine in November though the slightly better high ISO's appeal slightly.

The DR settings are on the Q Button or you can asian one of the Function Buttons to access it - if you don't shoot video thats a good button to chose.

[John 2]

...........................The DR settings are on the Q Button or you can asian one of the Function Buttons to access it - if you don't shoot video thats a good button to chose.

Good thought.

[William W]

. . . If you follow the Fuji forums and groups it is this jpeg output and the film simulations offered that make the X System so appealing. . .

Ditto here. I am using the RAF files - AND - and the Post Production workstation - less and less often.

It's just sooooooooooooooooooooo much fun shooting 'film', SOOC.

*

Thanks for the link, Dan K.

WW

[William W]

. . .It might perhaps be helpful to the OP to explain a bit of DR optimising . . . Canon answered with HTP (Highlight Tone Priority), a feature in EOS 20D, which did absolutely nothing, except changing the figures for ISO in the viewfinder from 200 to 2oo . . .

I believe the statement in bold is erroneous.

With Canon EOS DSLR’s, the Highlight Tone Priority is a function enabled and disabled by the Custom Functions Menu {“C.Fn”}

I understand that Highlight Tone Priority as a Custom Function was introduced at:

> the EOS xxD Series at the EOS 40D

> the xxxD Series at the 450D

> the 5D Series at the EOS 5DMkII

Other EOS cameras having Highlight Tone Priority as a Custom Function also include but may not be limited to: 1100D; 7D and 7D MkII; 100D; 6D; 1DsMkIII; 1DMkIV; 1Dx

*

. . . I cannot say how Fuji might have implemented it, and I don't know most other cameras. The ones I have tested thoroughly for the function are Panasonic Lumix G1 and Canon EOS 20D, which both boast of having such a function, but none of them does anything else than altering the ISO figures in the display. Nothing at all, nil!

As previously mentioned, I understand that the Canon EOS 20D does NOT have Highlight Tone Priority as an option.

In any case and to the main point - it is my experience that HTP does NOT do "Nothing at all, nil!"

I have used the HTP function with the: 50D; 450D; 5DMkII and 7D. Most usually on the 50D and 5DMkII, which were/are cameras I own(ed).

I note a credible and apparent difference in A/B tests of the same scene which in turn provided valuable ‘wiggle room’ for later, in Post Production.

I have found HTP especially useful when the output essentially comprises the JPEG SOOC with minor if any PP and the turnaround time is necessarily short.

I think that Canon's HTP is a feature that many Photographers would not find necessarily useful to their outputs.

*

To find out what the Fuji does, just as with any other camera, you'll have to test it under controlled conditions.

I think that A/B testing with the gear one uses is always a good process.

WW

[William W]

. . . The Fuji crowd are a bit of a weird bunch (I'm on the fringe but lean towards the unusual at times) with many of them hell bent on only shooting jpeg and using the Film Simulations along with H-Tone/S-Tone and other adjustments to get results straight out of the camera. . .

HAHAHAHAHAHAHAHA
ROFKML

I recall that you have already admitted (in a previous conversation) that you're hooked. No use fighting it. Then there's that magnificent 35 Prime Lens. . . but that's another topic.

[george013]

I still having trouble how this DR adjustment works.
If the exposure, A or S, doesn't change, then it's all pure PP. Like D-lighting with Nikon.
If the exposure does change, then it's something as Active D-lighting with Nikon.
I must recall that Active D-lighting is a correction on the matrix metering.

George

[pnodrog]

The sensor wells are a set size so in terms of ISO they are not a variable. To change ISO you need a method to scale the charge collected to compensate for the fact that less light(charge) has been captured. The scaling methods can either be by increasing gain for any amplification, lower the A/D converters reference voltage or scale using an algorithm either in camera or PP.

The manufactures will adopt different approaches depending on performance and costs benefits. If speed is not a concern the more they can do using an algorithm (software/firmware) the lower manufacturing costs are likely to be.

However if the exposure is not sufficient to come close to fully charging at least some sensor wells the dynamic range will be reduced and signal to noise ratio increased regardless of how the ISO is derived.

You don't get anything for nothing...

[John 2]

.....................................You don't get anything for nothing...

Hence the original question.

[Inkanyezi]

I believe the statement in bold is erroneous.

With Canon EOS DSLR’s, the Highlight Tone Priority is a function enabled and disabled by the Custom Functions Menu {“C.Fn”}

I understand that Highlight Tone Priority as a Custom Function was introduced at:

> the EOS xxD Series at the EOS 40D

Thanks for pointing out the error.
I went back to the images where this was tried some years ago, and I had mixed up those numbers. Corrected in the post.

[dje]

I've been following this thread with interest but have no experience with or detailed knowledge of the Fuji gear. However I came across this link on the TechRadar site which suggests that in the extended DR mode, two different ISO settings are used on the same capture with the higher ISO used for shadow areas and the base ISO used for highlights. This makes a fair bit of sense to me but it would require different hardware to the norm, perhaps with with two separate ISO amplifiers and ADC's (as was suggested in a classic paper on sensor noise by Emil Martinez in 2008.). Other possibilities might be a non-linear ISO amp mode, or even the dual ISO mode that Magic Lantern has used on Canon cameras where half the pixels are scanned with one ISO setting and the other half with a different ISO setting.

The concept of using two different ISO settings is in some ways similar to the HDR technique but it is done in hardware associated with the sensor rather than in software.

Incidentally TechRadar also suggests that the mode is actually applied to the raw capture (which would be logical if the hardware is as suggested above).

But who would know what Fuji are up to for sure?

Dave

[dem]

DR setting on Fuji cameras changes exposure - it underexposes by one (DR200%) or two (DR400%) stops to preserve highlights. So if you are a raw shooter, AFAIK, there is little benefit of using anything but base DR100% and expose/underexpose as you wish.

To answer the OP, there are three main tradeoffs for increasing DR setting when shooting JPEGs:

1) the higher ISO appearance

2) the lower image contrast

3) the more sparse colour information (not as bad as to cause obvious pasterisation but moving in that direction)

With DR setting Fuji seem to use some global algorithms that squash the whole histogram. If you want to lift the shadows without affecting contrast in mid tones, then set in-camera shadows JPEG setting to -1 or -2.

In terms of curve adjustment, the DR setting moves the black point back by one (DR200%) or two (DR400%) stops and changes the slope of the whole curve accordingly, whereas the in-camera JPEG settings "shadows" and "highlights" bend the curve only at the corresponding end by some preset amount (-2 -1 0 +1 +2 settings).

Shooting raw allows the users to use any S-shaped or whatever-shaped curve they want on the computer. You can also develop raw in camera using the above JPEG presets but they are obviously not as flexible as adjusting curves on the computer screen.

[dem]

However I came across this link on the TechRadar site which suggests that in the extended DR mode, two different ISO settings are used on the same capture with the higher ISO used for shadow areas and the base ISO used for highlights.

Thanks for the link. They do say "the image contrast and tonal depth seem unaffected" that, however, was not my experience. In fact, the X-T1 manual on page 79 says about DR settings:

"Choose lower values to increase contrast when shooting indoors or under overcast skies, higher values to reduce loss of detail in highlights and shadows when photographing high-contrast scenes. Higher values are recommended for scenes that include both sunlight and deep shade, for such high-contrast subjects as sunlight on water, brightly-lit autumn leaves, and portraits taken against a blue sky, and for white objects or people wearing white; note, however, that mottling may appear in pictures taken at higher values."

[William W]

. . . I had mixed up those numbers.

For clarity of meaning: I thought this to be the case.

At first I thought it was a simple typo, but "20D" was repeated so then I thought it was probably long ago when you played with HTP.

The error stood out to me as I still use a 20D regularly and only recently retired my 50D.

WW

[xpatUSA]

I have never understood how throwing away 6 bits of data (going from 14-bits in a modern raw file to the 8-bit jpeg) can be achieved without losing something.

Indeed, it can be even worse than that. Say I opened a 14-bit raw file in a decent editor . . I should be then working in at least 32-bit integer or floating point calculation - with the results temporarily stored in 16-bit linear ROMM space. Meanwhile, if our masterpiece is being reviewed on a 6-bit dithered monitor, that's a whopping 10 bits of data not being shown.

How can people even work under such conditions? ;-)

Fortunately, as a proud NEC P232W owner, I only "lose" 8 bits in the the review image. It's a good thing that my human eyes are so bad that I am not actually inundated with posterization every time I look at the screen.

I still haven't tried the recently acquired PIXMA MG8120 . . almost too scared, now :-)

Why not just build 8-bit sensors and increase the data throughput to some ludicrous data rate?

I did get it, Manfred, but I haven't had the pleasure of a good sarcastic rant lately. Thanks for the opportunity

[george013]

I still have trouble to understand it. To much noise.

DR setting on Fuji cameras changes exposure - it underexposes by one (DR200%) or two (DR400%) stops to preserve highlights. So if you are a raw shooter, AFAIK, there is little benefit of using anything but base DR100% and expose/underexpose as you wish.
....

Reading this it looks nothing more as changing the exposure, A/S, and a selective correction in in-camera PP when creating a JPG. I don't see that much basic difference with ADL with Nikon.
Fuji calls it 200% and 400%, Nikon has more choices: low, normal, high, extra high and auto.

George

[Black Pearl]

I need to correct an earlier post.

The DR expansion modes on the Fuji are available in BOTH file formats - RAW and JPEG

I never checked assuming it was JPEG only but I was fiddling with the camera today and noticed it didn't override the format settings - found this too:

http://www.dpreview.com/reviews/fujifilm-x-pro1/20


As I said before Fuji's idea of raw isn't always what you would expect with quite a lot going on under the hood...

[george013]

I need to correct an earlier post.

The DR expansion modes on the Fuji are available in BOTH file formats - RAW and JPEG

I never checked assuming it was JPEG only but I was fiddling with the camera today and noticed it didn't override the format settings - found this too:

http://www.dpreview.com/reviews/fujifilm-x-pro1/20


As I said before Fuji's idea of raw isn't always what you would expect with quite a lot going on under the hood...

My big, still not answered question is: does the exposure change, A and/or S.

George

[xpatUSA]

My big, still not answered question is: does the exposure change, A and/or S.

George

My answer to the big question is "no".

If I understand correctly, these in-camera dynamic range options simply apply different tone curves. For example, the Nikon D7100:

At its default Active D-Lighting setting of 'ADL Off', the D7100 exhibits a tone curve that differs somewhat from its predecessor the D7000, by displaying a more gentle roll-off from maximum white to highlight detail. In practice, this means that with the D7100 you can gain about .3EV stop more highlight information while maintaining identical midpoint and shadow renderings. The D7100 also exhibits a less contrasty tone curve than recent Nikon DSLRs like the D5200 and D600. Total JPEG dynamic range on the D7100 is roughly 9.5EV, with about 4.5EV in the highlights, again with a relatively gradual roll-off at the top end. Shooting in Raw mode will of course give you more headroom, allowing you to recover some of the brighter tones that might end up being clipped in the D7100's default JPEG mode.

http://www.dpreview.com/reviews/nikon-d7100/14

In DPR's method of measurement, the exposure for the test is set to put the middle part of the test target in the middle of the test image tonal range. If their back-lighting of the test target is constant, then the exposure does not change.

How the aperture A and integration time S are set seems to be a secondary matter - just as there are many combinations of A and S that give the same exposure for any shot. What I am trying to say here is that, if the exposure is correct, only the exposure counts - not the exact settings of A and S.

Would remind everybody that there are several definitions of dynamic range, so saying "DR" without any qualification often leads to misunderstanding. Any disagreement with this particular post should be made with reference to the link posted above and with reference to DPR's test method - not some alternative definition of DR.

Pardon me for restricting the scope for discussion about this post.