What is Linear Response?
I'm trying to understand precisely what's going on with C1 Linear Response and hoped may be someone at Phase might be able to assist? Charles Poynton points out that the term linear is often used in scenarios when something isn't in fact linear (scene-referred), such as video (linear), log neg (Cineon), log RGB (quasilog). I wondered how C1 Linear Response was achieved? Is it inverting an input device transfer (log/ gamma curve) using some kind of standard power function or possibly using exact camera vendor IDT calculations? The thing I found interesting was that the C1 Linear Response file didn't seem linear, say in comparison to opening a Raw in Picturenaut for example, where you see far more noise and a flatter image overall. The absence of noise C1 Linear Response file suggested some kind of curve and manipulation.
Any insight would be great!.... just interested in digital workflow.
Any insight would be great!.... just interested in digital workflow.
0
-
Linear response is the nearest Capture One gives you to no CURVE being applied to the image at all, but it doesn't mean that Capture One isn't applying some other things to the image. So try setting an image to the Linear Response option on the Colour tab, then head over to the Details tab where you will see the noise reduction. An amount of noise reduction is applied according to the camera, ISO, etc. You could set all the sliders in the noise reduction tool to zero and examine the image at 100% zoom, then reset the tool and compare. It can make quite a difference. (I'm looking at an image shot at ISO 5000, so the difference is quite marked. On another image shot at ISO 200, the difference is hard to see, as you might expect).
So you are not getting an image with nothing applied to it at all by using Linear Response, just no curve applied. You are still getting things like some noise reduction, some sharpening, etc. If you like, you could set NR to zero and save it as a preset, and set sharpening to zero and save that as a preset, if you'd like to see an even more unmodified image.
But whether that makes it harder or easier to get the output you want I can't say. The best thing is to go for the curve in base characteristics that is nearest to what you want, then tweak things up or down from that baseline.
Ian
Edited to add, I am not from Phase One, and this is primarily a user-to-user forum. Phase One staff sometimes participate, though.0 -
I have been told, by people skilful in profiles and so on, that Linear response is not really linear. But the applied input curve is closer to linear than any other C1 input curve (auto, film standard, ....).
As far as I am concerned, I have created a default preset with linear response, zero noise reduction and zero sharpening. It gives me a close to raw file ( 😄 ), rather flat, from which I feel, personally, that it's better to start, applying levels, curves and so on.0 -
[quote="tenmangu81" wrote:
I have been told, by people skilful in profiles and so on, that Linear response is not really linear. But the applied input curve is closer to linear than any other C1 input curve (auto, film standard, ....).
As far as I am concerned, I have created a default preset with linear response, zero noise reduction and zero sharpening. It gives me a close to raw file ( 😄 ), rather flat, from which I feel, personally, that it's better to start, applying levels, curves and so on.
Linear Response is as flat of a response curve as possible for the majority of camera files out there. The exception is, however, that with many of our digital back files we have a Linear Scientific as an option. Linear Response is a true linear representation except towards the extreme highlights, which will have a small amount of "protection" against clipping; there will be more detail in the highlights than a truly linear response (this is true for all camera make/models). The Linear Scientific response curve is a true linear response, with no protection in the highlights.0 -
it is totally misleading to call something linear which is far away from being linear at all I don't understand why p1 does not simply call it what it is a low contrast curve. but I also don't understand why p1 is not offering a true linear curve for all other cameras. 0 -
[quote="Horseoncowboy " wrote:
it is totally misleading to call something linear which is far away from being linear at all I don't understand why p1 does not simply call it what it is a low contrast curve. but I also don't understand why p1 is not offering a true linear curve for all other cameras.
“Far away from being linear at all†is quite a bit different from the explanation from ben_US that it is truly linear except towards the extreme highlights!
Ian0 -
[quote="Ian3" wrote:
[quote="Horseoncowboy " wrote:
it is totally misleading to call something linear which is far away from being linear at all I don't understand why p1 does not simply call it what it is a low contrast curve. but I also don't understand why p1 is not offering a true linear curve for all other cameras.
“Far away from being linear at all†is quite a bit different from the explanation from ben_US that it is truly linear except towards the extreme highlights!
Ian
in the literature linear normally means that the file is presented without any gamma correction. in other words as the camera has recorded it. but with c1 linear the camera data has already altered. how I don't know, I have never seen any explanation what c1 linear really does but it is for sure not the linear representation of the camera raw data !0 -
[quote="Horseoncowboy " wrote:
[quote="Ian3" wrote:
[quote="Horseoncowboy " wrote:
it is totally misleading to call something linear which is far away from being linear at all I don't understand why p1 does not simply call it what it is a low contrast curve. but I also don't understand why p1 is not offering a true linear curve for all other cameras.
“Far away from being linear at all†is quite a bit different from the explanation from ben_US that it is truly linear except towards the extreme highlights!
Ian
in the literature linear normally means that the file is presented without any gamma correction. in other words as the camera has recorded it. but with c1 linear the camera data has already altered. how I don't know, I have never seen any explanation what c1 linear really does but it is for sure not the linear representation of the camera raw data !
Do you know of any RAW convertors that are set up to present the data to the user with no conversion applied?
What to the results look like?
Grant0 -
[quote="ben_US" wrote:
Linear Response is as flat of a response curve as possible for the majority of camera files out there. The exception is, however, that with many of our digital back files we have a Linear Scientific as an option. Linear Response is a true linear representation except towards the extreme highlights, which will have a small amount of "protection" against clipping; there will be more detail in the highlights than a truly linear response (this is true for all camera make/models). The Linear Scientific response curve is a true linear response, with no protection in the highlights.
Thanks Ben !! Very useful explanation.0 -
[quote="SFA" wrote:
Do you know of any RAW convertors that are set up to present the data to the user with no conversion applied?
What to the results look like?
Grant
you can do a little experiment if you have some spare time. export for example a color checker image from c1 with linear film curve and no color management. open the file in photoshop without conversion and apply a gamma 1.8 profile like colomatch I´m sure you are able do draw the right conclusion 😉
for the majority of users a true linear file would be without any benefit but for those who think "professional" means foremost control it could give us the opportunity to build better camera profiles. maybe it does not change anything but i would like to test it for my own.0 -
I think you need to look towards something running DCRaw. I was researching a bit about ACES workflow to understand how true linear is working in cinema. [quote="Horseoncowboy " wrote:
[quote="SFA" wrote:
Do you know of any RAW convertors that are set up to present the data to the user with no conversion applied?
What to the results look like?
Grant
you can do a little experiment if you have some spare time. export for example a color checker image from c1 with linear film curve and no color management. open the file in photoshop without conversion and apply a gamma 1.8 profile like colomatch I´m sure you are able do draw the right conclusion 😉
for the majority of users a true linear file would be without any benefit but for those who think "professional" means foremost control it could give us the opportunity to build better camera profiles. maybe it does not change anything but i would like to test it for my own.0 -
I find it interesting that there's a number of types of linear, gamma being one flavour. I would say C1 don't put out true linear/ linear scene referred, because it doesn't look any good.... I assume because it's totally curve free the data would be very noisy in sections, as this noise is yet to be masked in compression? Still would like to know what C1 are doing! [quote="Horseoncowboy " wrote:
[quote="Ian3" wrote:
[quote="Horseoncowboy " wrote:
it is totally misleading to call something linear which is far away from being linear at all I don't understand why p1 does not simply call it what it is a low contrast curve. but I also don't understand why p1 is not offering a true linear curve for all other cameras.
“Far away from being linear at all†is quite a bit different from the explanation from ben_US that it is truly linear except towards the extreme highlights!
Ian
in the literature linear normally means that the file is presented without any gamma correction. in other words as the camera has recorded it. but with c1 linear the camera data has already altered. how I don't know, I have never seen any explanation what c1 linear really does but it is for sure not the linear representation of the camera raw data !0 -
[quote="NN635337618399883750UL" wrote:
I find it interesting that there's a number of types of linear, gamma being one flavour. I would say C1 don't put out true linear/ linear scene referred, because it doesn't look any good.... I assume because it's totally curve free the data would be very noisy in sections, as this noise is yet to be masked in compression? Still would like to know what C1 are doing!
When the linear "curve" is applied, Capture One does still apply some noise reduction and some sharpening. What is not clear from the explanation from ben_US that it is linear except at the extreme highlights?
Ian0 -
[quote="Ian3" wrote:
When the linear "curve" is applied, Capture One does still apply some noise reduction and some sharpening. What is not clear from the explanation from ben_US that it is linear except at the extreme highlights?
Ian
p1does makes a big secret of their color processing but actually I think to understand how it works would not hurt but could lead to better results. I´m no engineer so this is over simplified
camera linear > mapped to a gamma which is close to 1.8 ( internal color space ? ) + linear curve ( = no gamma but other sophisticated corrections ? ) + icc camera profile with addional gamma correction
...give a result which is in no way linear but you entitled to believe whatever you like 😉
at the end the use of the linear filme curve does come with a price many are probably not aware of - saturation and color shifts - so you can solve the harsh default tone curve issue but you get a new one. looks like there is room for improvement
http://www.cambridgeincolour.com/tutori ... ection.htm0 -
Useful article. Thanks.
Ian0 -
First, define linear.
Linear means that the change of output of a system is always proportional to the change of input.
There are few things in nature or made by humans that are linear for all input values. A few materials are nearly perfectly linear, e.g. glass and vacuum. You would think that a copper wire might be perfectly linear, but as you raise the current, the wire gets hot and the resistance increases with temeprature (still pretty linear) and then the wire starts to sag (more nonlinear) and then there is a bright flash as the wire vaporises (very nonlinear). Outside its "normal" range even a piece of wire exhibits nonlinear behaviour.
Most technologies depend on devices which are approximately linear over some range, and outside this range become increasingly nonlinear. In some cases, suddenly very nonlinear. Transistors for example. A wood or steel beam is a low tech example. Too much force, and instead of bending, the beam aquires a permanent deformation and eventually breaks.
The pixels in a photographic sensor are only linear over some range. Additional photons in results in more electrons out which results in a larger digital output. But there are imperfections. Too much light in, and the pixel output will stop increasing; the pixel is saturated. Too little light and the output changes nooticeably even when the input stays constant - noise. In some cases a small change in input results in no change in the digital output - that's quantisation, often seen at lower light levels. All of these are nonlinearities in the camera.
When COP processing implements noise reduction - that's non linear behaviour - but the purpose is to remove noise added in the sensor, which is also nonlinear. However the intent is that these two nonlinearities (noise and noise reduction) cancel.
Highlight recovery is nonlinear - but the purpose is to remove the effect of pixel saturation or clipping with thin the camera which is also non linear - again the intent is that these two nonlinearities cancel.
The sensors pixels do not capture 100% of the light hitting sensor - the result is spatial quantisation, which reuslts in fine detail being lost. Sharpening is an attempt to recover that detail - again the intent is that two nonlinearities cancel.
It is true that sharpening, noise reduction and highlight recovery are nonlinear - but their purpose is to maintain the imaging system's linearity even when the sensor exhibits nonlinear behaviour. It's then a bit harsh to say that a certain mode is non-linear only because there is sharpening or noise reduction or highlight recovery.
Properly adjusted they actually make the imaging system more linear.0 -
maybe the phaseone linear is like schrödingers cat and as long as you don't open the file it is linear.... 0 -
[quote="Horseoncowboy " wrote:
maybe the phaseone linear is like schrödingers cat and as long as you don't open the file it is linear....
😂
But seriously, the most important thing is whether you can get the results you want with the tools they provide.
Ian0 -
[quote="Eric Nepean" wrote:
First, define linear.
<clip for pixel saving>
Properly adjusted they actually make the imaging system more linear.
Excellent post Eric.
Thanks for taking the time to put it together and in such a clearly understandable way.
There ought to be some way of bookmarking and cataloguing posts like this.
Grant0 -
[quote="Ian3" wrote:
[quote="Horseoncowboy " wrote:
maybe the phaseone linear is like schrödingers cat and as long as you don't open the file it is linear....
😂
But seriously, the most important thing is whether you can get the results you want with the tools they provide.
Ian
Exactly.
I wonder what we would "see" if our eyes were connected to only the most basic parts of our brains and had no lightness/darkness adjustments processing (or colour interpretation for those not subject to "colour blindness" or some sort) ?0 -
A bit off topic but I've been using 3dLUT creator for a while now, and it's helped resolve many colour space issues, and offers far more control than say PS. They've just made it compatible with C1... so C1 just got a whole load better!
https://www.facebook.com/3dlutcreator/v ... 004399224/0 -
[quote="NN635337618399883750UL" wrote:
A bit off topic but I've been using 3dLUT creator for a while now, and it's helped resolve many colour space issues, and offers far more control than say PS. They've just made it compatible with C1... so C1 just got a whole load better!
https://www.facebook.com/3dlutcreator/v ... 004399224/
thanks, looks interesting did test 3dLUT some time ago but it was rather buggy and slow with large 16 bit files. seem it is time to give it another try.0 -
Help me understand 3dLUT.
I watched the video in the link.
What is the benefit using 3dLUT to color grade?
In the video they went through presets until they found the color scheme that they wanted and then turned it into an ICC so it could be applied to multiple images.
How is that different than applying a preset style or manually using levels, curves, or color balance/editor within Capture One to color grade and then apply those settings to other images?0 -
3dlut is a very sophisticated colour grading software, it has controls that exceed what C1 and PS are able to provide. You'll need to sit down and do a bit of research, navigate some English & Russian tutorials (although the support is getting better), and test out features. I was trying to find an introduction I came across which might be helpful but can't for some reason.... you'll soon start to see how accurate a tool it can be though. There's some really dodgy colour grading in a lot of the video examples which can be quite amusing!
[quote="cdc" wrote:
Help me understand 3dLUT.
I watched the video in the link.
What is the benefit using 3dLUT to color grade?
In the video they went through presets until they found the color scheme that they wanted and then turned it into an ICC so it could be applied to multiple images.
How is that different than applying a preset style or manually using levels, curves, or color balance/editor within Capture One to color grade and then apply those settings to other images?0 -
Thanks for the response. I will do some research. 0
Post is closed for comments.
Comments
24 comments