Using 'Linear Response' curve rather than 'Film Standard' in C1 ?

[narikin]

I read a few replies here referring to the improved results of using 'Linear Response' in the curve choice, rather than 'Film Standard'

Can someone outline the improvements of this - is it beneficial to all image types, or some more than others? - and run through a brief workflow to set it up as a 'style' and apply it?  I understand you'd need to shoot a test show of the Xrite color checker (can do that) but then what do you do with that shot?

Thanks in advance,

[Bart_van_der_Wolf]

I read a few replies here referring to the improved results of using 'Linear Response' in the curve choice, rather than 'Film Standard'

Can someone outline the improvements of this - is it beneficial to all image types, or some more than others?

Hi,

The Film Standard curve that is applied by usual default, produces an image with similar characteristics to what film does, it has (amongst others) a highlight roll-off somewhat like the shoulder of the film H&D curve does. To achieve that, the highlights get progressively more compressed (lower contrast) as one approaches the brightest highlights of an image.

While that may help to avoid sudden clipping of highlights, e.g. a sun disk in a bright sunset sky turning (hopefully white) with a sharp edge, it also will take the life out of e.g. very bright clouds near that sun. Assuming one has exposed correctly for the highlights, and there is some detail in bright image areas, it is IMHO usually better to not make those highlights which still do have detail, looking dull and having low contrast. To that end, one uses a Linear curve instead, and the images keep looking crisp.

An unfortunate by-product of the way that the Film tonecurve is implemented in Capture One, is that it shows correctly exposed (ETTR) images, without clipping, as being clipped and seemingly overexposed by about one stop for some camera Raws. Correcting that requires reducing the exposure control which darkens the entire image below what the scene is supposed to look like.

- and run through a brief workflow to set it up as a 'style' and apply it?

Once you have the Linear response selected for an image, and other settings you always end up making for a certain type of images (e.g. noise reduction only for High ISO shots), you can save (a selection of) those settings as a User defined Style which you can later recall with a few mouse clicks, or even set one as a default. Just make sure to only include those settings in a Style that you actually want to have changed by selecting that Style when you e.g. switch between Styles.

I understand you'd need to shoot a test show of the Xrite color checker (can do that) but then what do you do with that shot?

No, that's a requirement for Lightroom/ACR, not Capture One.

Cheers,
Bart

[narikin]

Thanks Bart,  very useful info.

So - from this are you saying that 'Linear' is really most/only useful when there is (e.g.) a sky image with lots of highlight detail that's got squashed out by 'Film'?  and by implication, other images that are mostly mid-tones, its of less benefit?

And... you also seem to imply that one should not trust the IQ capture histogram, when ETTR with a sky image (e.g.) as that shows a squashed/compressed highlight, which is not accurate, and will lead to underexposure?  I do see that images that are 'correctly' exposed according to the histogram on back of my IQ, are a full stop darker if I switch to 'Linear Response' when on my desktop in C1.  Unfortunately you can't choose to have a linear response histogram on your back display (I think?) Therefore, your only choice is to guess on the overexposure amount, or work tethered and choose Linear in C1 on your laptop, using that histogram to assess exposure, and ignore what the back histogram says.

 

[Bart_van_der_Wolf]

Thanks Bart,  very useful info.

So - from this are you saying that 'Linear' is really most/only useful when there is (e.g.) a sky image with lots of highlight detail that's got squashed out by 'Film'?  and by implication, other images that are mostly mid-tones, its of less benefit?

Hi,

When there are few highlights, but that must be dark dull images (black cat in a coal mine?), then there is not much difference, although similar things can happen in the deep shadows. I personally dislike the highlight rendering of the film curve so much that I've not studied the shadows in detail.

And... you also seem to imply that one should not trust the IQ capture histogram, when ETTR with a sky image (e.g.) as that shows a squashed/compressed highlight, which is not accurate, and will lead to underexposure?  I do see that images that are 'correctly' exposed according to the histogram on back of my IQ, are a full stop darker if I switch to 'Linear Response' when on my desktop in C1.

That's because the Phase One backs generally underexpose, to protect the highlights (and with a Film curve in mind). To put it differently, their actual ISO is lower than indicated, at least for some models, One would have to check on DxOmark.com if that also applies for a specific model.

Cheers,
Bart

[narikin]

When there are few highlights, but that must be dark dull images (black cat in a coal mine?), then there is not much difference, although similar things can happen in the deep shadows. I personally dislike the highlight rendering of the film curve so much that I've not studied the shadows in detail.

well, not black cats in coal mines (!) but I was just thinking of images of some trees/greenery with no sky, where other than the odd specular detail, 99% of the tonal range is in the middle, especially if made under diffuse daylight.

Not all images have a 'full' tonal range, and it is, imho, wrong to stretch them out and force that, when its not the situation. Flat lighting can be very good sometimes!

[ErikKaffehr]

Hi,

I agree with Bart, by I don't think P1 backs underexpose a lot. What I do is to check my images with RawDigger, and I feel that histogram and blinking highlights are helpful for ETTR exposure. I also think "film curve" is misleading.

I have about a month of experience with a P45+, so I need to admit that I am not really experienced. I also would say I prefer Lightroom so far.

Best regards
Erik

[bjanes]

Hi,

I agree with Bart, by I don't think P1 backs underexpose a lot. What I do is to check my images with RawDigger, and I feel that histogram and blinking highlights are helpful for ETTR exposure. I also think "film curve" is misleading.

I have about a month of experience with a P45+, so I need to admit that I am not really experienced. I also would say I prefer Lightroom so far.

Best regards
Erik

Erik,

The DXO ISO data confirm your impression. The P45+ is an "ISO less" camera in that the camera gain above ISO does not change with increased ISO settings on the camera. The measured ISO with the camera set at 100 is spot on. The ISO handling of the IQ180 is entirely different, and the camera does underexpose since the measured ISO is considerably less than the nominal ISO set on the camera.

My personal opinion is that the camera and software should stick to ISO standards. This also applies to LR/ACR where an exposure offset is used. With the Nikon D3, the offset is +0.5 EV and this complicates evaluation of exposure according to the the LR/ACR histograms. The same applies to automatic highlight protection in LR/ACR, which appears analogous to the C1 film setting. These objections would be overcome by the provision of a switch for raw histograms both on the camera and in the software.

Regards,

Bill

[Vladimirovich]

With the Nikon D3, the offset is +0.5 EV and this complicates evaluation of exposure according to the the LR/ACR histograms.

with ACR/LR nowadays hidden exposure correction is defined by 2 components : what is hardcoded in LR/AC (to find out convert raw to DNG and check the tag) + what is coded in a specific .dcp camera profile selected... so hidden exposure correction = sum of those two numbers... and it changes (might change) based on which .dcp camera profile you select (for Nikons there are > 1 "OEM emulation" profiles from Adobe, in addition to Adobe Standard)

[Bart_van_der_Wolf]

well, not black cats in coal mines (!) but I was just thinking of images of some trees/greenery with no sky, where other than the odd specular detail, 99% of the tonal range is in the middle, especially if made under diffuse daylight.

Not all images have a 'full' tonal range, and it is, imho, wrong to stretch them out and force that, when its not the situation. Flat lighting can be very good sometimes!

Ah, but that depends on whether the output from Capture One should be the final product, or an intermediate for further processing in e.g. Photoshop. Once you process the file with few remaining highlights, all other tones will not be optimally separated for further processing either. On the other hand, if the full tonescale is used, the intermediate tones will be well separated and further processing has a perfect foundation to work from.

A recent plug-in from Topaz Labs, named Clarity, proves once more how useful it is to have a full tone scale without compressed tones. Each and every different level in the full tonescale can be adjusted. The end product can then have the midtones reduced to below average brightness, but they will have an enourmous amount of detail (should that be the goal). Once it's compressed into a single brightness level, there is little possibility left to use or rescue it.

Cheers,
Bart

[bjanes]

with ACR/LR nowadays hidden exposure correction is defined by 2 components : what is hardcoded in LR/AC (to find out convert raw to DNG and check the tag) + what is coded in a specific .dcp camera profile selected... so hidden exposure correction = sum of those two numbers... and it changes (might change) based on which .dcp camera profile you select (for Nikons there are > 1 "OEM emulation" profiles from Adobe, in addition to Adobe Standard)

I am aware of the BaselineExposure tag in the DNG specification, but what is the name of this second tag you mentioned, as listed in the spec?

Regards,

Bill

[Vladimirovich]

I am aware of the BaselineExposure tag in the DNG specification, but what is the name of this second tag you mentioned, as listed in the spec?

1) BaselineExposure = hardcoded for all cameras in ACR/LR code (and recorded in DNG when raw is converted by Adobe software)

2) BaselineExposureOffset = either set in .dcp camera profile or zero if absent

hidden exposure correction = BaselineExposure + BaselineExposureOffset

PS: that is also the way to have expocorrections in ACR/LR "beyond" the UI limits - just create one more profile with proper negative or positive BaselineExposureOffset tag.

[Paul2660]

For all Phase files I still prefer the output from Capture one.  As many have pointed out, the film curve seems to apply almost 1/2  stop of expsoure.  You can see this as the thumbnails are being loaded by C1 as each will load darker and then increase in brightness as it's processed by C1.

With the P45+, this always tended to cause the hightlights to blow out, and often I would use the liner curve.  However with the P65+ / IQ60 there is considerably more headroom on the hightlights and shadows.  Most times you can regain most of the cloud details as Bart mentions, with the highlight tool.  With C1 vr 7 both the highlight and shadow tools are very powerful.

I always found the P45+ to be very unforgiving on highlights, and tended to bracket most shots.  When I switched to the IQ160, I was pleased to see that there was considerly more headroom for highlights.  The DxO mark scores on the P45+ and P65 reflected this very well.

With the IQ160 files I rarely use the linear curve as I prefer the look of the film curve and only go liner when I have a situation with hightlights that can't be recovered easliy. 

LR loads the files without the exposure push, but I still find the output from C1 more to my overall liking.

Paul Caldwell

[bjanes]

1) BaselineExposure = hardcoded for all cameras in ACR/LR code (and recorded in DNG when raw is converted by Adobe software)

2) BaselineExposureOffset = either set in .dcp camera profile or zero if absent

hidden exposure correction = BaselineExposure + BaselineExposureOffset

PS: that is also the way to have expocorrections in ACR/LR "beyond" the UI limits - just create one more profile with proper negative or positive BaselineExposureOffset tag.

Thanks, that is useful information. In looking at the DNG file EXIF for my D800e, there is no BaselineExposureOffset. Perhaps that is because no profile using a DCP exposure adjustment was in play.

Bill

[Vladimirovich]

Thanks, that is useful information.

it is all in DNG spec 1.4

In looking at the DNG file EXIF for my D800e, there is no BaselineExposureOffset. Perhaps that is because no profile using a DCP exposure adjustment was in play.

BaselineExposureOffset intended to be in .dcp camera profiles, so the only way you can have that in .dng file is to have a .dcp camera profile embedded in .dng with that tag

for example :

BaselineExposure: +0.35 in .dng (straight conversion from .nef to .dng using Adobe's DNG Converter)
BaselineExposureOffset: -0.35 in .dcp (D800e Camera Neutral from Adobe - dump using Sandy Mc' dcptool)

now if you will embed that .dcp camera profile into .dng (with ACR for example) you shall see both tags there... otherwise it will be just BaselineExposure (I'd assume because Adobe's DNG converter in a default situation will use Adobe Standard and if it has BaselineExposureOffset = 0 (absent) then it will not embed that tag) ...

[torger]

I think the whole film-curve thing is rubbish . It's about hiding how digital photography works from users, and I think users are able to learn the technical aspects of digital photography and should not need to have a program that emulates film.

I prefer the way RawTherapee works, the only highlight "magic" that takes place is desaturation to the whitepoint at clipping, and clip point is set to first channel clip (usually green). On top of that possibility to enable various modes of highlight reconstruction to move the clip point past the first channel clip and reconstruct the remaining clipped information. And if you want a highlight compression/rolloff you either use a slider (called highlight recovery amount in RT, as what is does is to compress highlights down below clipping when exposure is pushed), or simply just compress the top of your tonecurve.

Both Capture One and Lightroom does so many hidden things hidden away in the background that I almost feel insulted , and I also think this way of designing digital photography software has impaired the general understanding of digital photography among users.

[narikin]

Just to revive this topic a bit: might I ask - does anyone here have a default/standardized curve file they could share (as a *.copreset) for getting us non-experts into the ballpark when using Linear Response?  

I have had a few attempts, but don't feel particularly amazed by my abilities.  The skies/highlights sure look better, but the mid and shadows seem to suffer.  A good standardized curve for Linear Response would be a terrific starting point.  I appreciate it would still need adjusting, but a move into the right area would be great.

It seems like something Phase should put in there, as a preset choice.  Come to think of it, considering all the weird/esoteric choices they do add as presets, its kind of surprising this is not built in.

[Hening Bettermann]

Hei Narikin,

I am not a C1 user, but as I have understood it, the linear 'curve' in general is an intermediate product, and the definite tone curve will depend on the individual image and your vision of it. So a standard curve sounds like a self-contradiction. The linear rendition IS the point of departure you are looking for.

Good light - Hening

[Phil Indeblanc]

I tend to agree with you Torger.

It ISN'T film, get over trying to emulate it. New rules new ways of working. Even ISO can be revamped with different sensetivity based on color, and noise levels in the mix.

Maybe it's a way for the software brands and programmers to "hide" how they get the results in developing the file?
 I tried RT. I don't think I gave it enough time, but I haven't tried it since. What the other apps like LR or C1 do is suggest how they can be used and how it will apply. This can be very helpful when working with new apps.

Things are not broke, so there is no need to change much, but I think there can be some things better explained due to the way sensors work, and no need to bridge it with film.

[rhadorn]

I am using the linear curve a lot. According to a recent post on the C1 blog, the standard curve simulates the old color slide, which reproduced only about 8 EV - much less than what present cameras are capable of.

I nevertheless experienced some difficulties in the reproduction of colours. Saturation goes back with contrast and can obviously not be simply compensated by raising the contrast slider. Raising contrast with the contrast slider can lead to some weird colours. So, using the linear development curve, I run into new constraints.

Since the HDR sliders have been improved (Version 7 I think), I make more often use of them instead of the linear curve. The original saturation in the midrange is preserved and it seems that the correction algorithm of the highlights and shadows sliders handles colors intelligently.

This being said, I would be very pleased if C1 gave some additional options besides the choice of a development curve, for example, a linear curve with color correction. It is not written in stone that the raw developer can only produce bright and well defined colors with the so called standard curve.


Reto

[allegretto]

Am currently taking the LULA C1 Course

In about lesson 10 or 11 or so, Michael and David explain the "linear" is the actual RAW file curve. The Others, "film standard" are curved renditions of the RAW