Jump to content

Superb new camera profiles and camera & film emulation presets (Cobalt-Image)


Recommended Posts

  • 2 weeks later...
Guest Nowhereman

@frame-it - Thanks for posting this video on using the Cobalt-Image profiles, which I only looked at today. It's particularly worthwhile to see for people who shoot with different cameras, whose output they want to match. In the video Samuel Lintaro uses the Ricoh GR III as his base camera and matches the output to Fuji cameras and to Leica Monochrom, among a few others.

As stated earlier, I've bought the DNG Base Profiles for my M10, and have found them particularly useful for processing shots in difficult light, such is in strong backlight and strong sidelight or with difficult color combinations, such as saturated reds and yellows — indeed many of the images that I've posted in this thread I've previously used only in B&W because, using Adobe profiles or the embedded Leica profiles, I found it difficult to produce color versions that I liked. So, my conclusion was that I really liked the Cobalt-Image profiles for my M10, and found them useful. 

I also have the Ricoh GR III and GR IIIx cameras, for which I have not yet bought the Cobalt-Image profiles because my feeling is that the it's easier to get good results with the Adobe profiles for the GR III/GR IIIx than with the M10. Also, the Ricoh JPGs are so much better and more flexible than those of the M10 that I shoot JPG+DNG with the Ricoh cameras — and often end up using the JPGs for a final edit, or use them as a target for my processing the DNGs. Also, Cobalt-Image sells separate profiles for the very similar GR III and GR IIIx cameras, apparently because they found they had to do separate calibrations on them. 
_______________________________________
Frog Leaping photobook and Instagram

Link to post
Share on other sites

  • 3 weeks later...
On 9/20/2021 at 1:18 PM, Ulysseita said:

About comparisons, do not forget, the cobalt linear is made to excel as color fidelity, the Film emulation is made to emulate a film so even the legacy of "great skin tones" is still on for Portra, please take in account the variability given by wb, different light conditions and original skin tones.

Thanks for the explanation. How useful is the linear cobalt profile as a starting point for painting reproduction? 

I use a home made custom profile adapted to my strobe lights and I work either with a Rollei 6008 (Phase one DB20) or my S3. 
When copying a painting with difficult reflection properties in the pigments, quite often it’s reds and purple, I have to combine several exposures made with different profiles, just to match the colors better. 
 

yevgeny

Edited by ynp
Link to post
Share on other sites

Reproduction photography is one of the most technically complex fields.
The basic idea of the characterization of a sensor is to translate the spectral response functions on those of the tristimulus sensitivity of the standard observer.
In this ideal situation the sensor would see exactly like a human being and its ISO 17321 metamerism index would be 100, perfection.

However, the normal photographic workflow, and therefore unscientific, has made technical and stylistic choices that are sometimes badly suited to a reproduction environment.
It would be impossible to cover every single aspect and technical detail in this post, but I will try to focus on a specific component of the development chain: the contrast curve.

It is important to understand that every sensor, be it CMOS or CCD, Bayer or monochrome, are analog linear devices. In short, a double quantity of radiation corresponds to a double electrical output signal.
We now have sensors capable of large dynamic ranges, the limit downstream of the A / D converter is 14 stops for 14bit converters and 16 stops for 16bit converters. But how to deal with such dynamics on output devices like monitors and printers that cannot achieve such dynamic range?
The solution is to insert a tonal compression curve into the development chain.
Adobe has adopted what it calls ACR Standard Curve:

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

It consists of two elements, an exposure curve with roll-off towards white to compensate for the normal under exposure with which each meter is calibrated, plus a sigmoid component which is responsible for compressing the dynamic range.

In practice, the linear sensor data are first characterized in their linear form and then compressed by the ACR curve.
This allows the image to appear believable to our senses when viewed on a monitor that typically does not offer more than 10 stops of dynamic range.
The ACR curve is an RGB curve that acts on an output-referred color space called RIMM.
The problems related to this architecture can be summarized as follows:
1-like any RGB curve, the compression of the luminance involves a shift of the hue.
2-the ACR curve is not adaptive.

All Adobe Standard profiles incorporate the ACR curve and this means that the level of compression applied is always the same, whether it is a photo in a panorama or if you are shooting in the studio in a condition of controlled dynamic range.

A linear profile removes the ACR curve, leaving the linearity and characterization of the scene-referred data intact.
The image produced is usually underexposed and flat but leaves the user to decide how much and whether to compress the dynamic range of the sensor.

 

Enrico Scaramelli, CTO Cobalt-Image

  • Like 2
  • Thanks 1
Link to post
Share on other sites

1 hour ago, Raamiel said:

All Adobe Standard profiles incorporate the ACR curve and this means that the level of compression applied is always the same, whether it is a photo in a panorama or if you are shooting in the studio in a condition of controlled dynamic range.

May I ask about this?

Presumably you mean adobe std profiles incorporate the tone curve into their LUT?

When I look at profiles I own (from RNI/VSCO/Cobalt) and ones I've made my self (dcamprof/lumariver/adobe DNG profile editor) the tone curve is specified separately, but I've never seen this in an 'adobe standard' profile...  (as you'll know, but others reading may not) if you open a standard abobe profile in DNG profile editor it will add the standard DNG tone curve, but I've never seen this included in any adobe standard that I've converted to a human readable format (json/xml)

So does adobe handle this via their LUT?

Thanks for your time

  • Thanks 1
Link to post
Share on other sites

Advertisement (gone after registration)

1 hour ago, Raamiel said:

A linear profile removes the ACR curve, leaving the linearity and characterization of the scene-referred data intact.

Thank you very much for your detailed explanation. Makes sense to use a Linear profile for my applications. 
 

yevgeny. 

Link to post
Share on other sites

@Adam Bonn

Yes, that's exactly how you noticed.
When decoding a normal Adobe Standard profile there is no trace of the ACR curve, which normally closes the tail of the profile.

This is an adobe "trick" to save space.
When a .dcp profile ends immediately after the LookTable field CameraRaw and Lightroom interpret the profile by adding the standard ACR curve.

The curve cannot be incorporated into the LookTable field because Adobe's configuration works with multipliers for the S and V coordinates, which would therefore not be able to do anything on the neutral axis.

If you want to produce a linear profile, the new curve, or rather its absence, must be made explicit:
   "ProfileToneCurve": [
     [0.000000, 0.000000],
     [1.000000, 1.000000]
   ] 

  • Like 1
  • Thanks 2
Link to post
Share on other sites

Thank you so much @Raamiel

So effectively things I write into my DCPs overwrite what LR/ACR would do without instruction (like turning auto black subtraction off, you have to tell it to do so)

That also perhaps explains where folks suggest using process 2 /2010 to get more linear highlight recovery, because it changes the default ACR tone curve?

Link to post
Share on other sites

@Adam Bonn

Yes, the Adobe specification (or rather their interpretation) has changed over the years.
In some respects for the better, in others for the worse.

For example, the original idea behind the DNG development pipeline is to divide the colorimetry stage between a matrix and a Lut HSV.
The characterization work was originally entrusted to the matrix, while the Lut HSV would have had to refine the work of the matrix in non-linear details.
Unfortunately, the evolution of sensors soon highlighted the limits of this concept, imposing negative multipliers.

But that's another (bitter) story. 

  • Thanks 1
Link to post
Share on other sites

19 minutes ago, Raamiel said:

@Adam Bonn

Yes, the Adobe specification (or rather their interpretation) has changed over the years.
In some respects for the better, in others for the worse.

For example, the original idea behind the DNG development pipeline is to divide the colorimetry stage between a matrix and a Lut HSV.
The characterization work was originally entrusted to the matrix, while the Lut HSV would have had to refine the work of the matrix in non-linear details.
Unfortunately, the evolution of sensors soon highlighted the limits of this concept, imposing negative multipliers.

But that's another (bitter) story. 

I’ve noticed that adobe’s more modern “bradford chromatic adaptation” matrices don’t actually equal the D50 white point and (to me at least) seem to actually desaturate colours a little (which I assume is 1. about forward matrix clipping and 2. they bring the colours back up in the HSV and LUT tables)

When I look at adobe dcp from say D90/leica m8 era - they’re “just” CMs FMs and a LUT, they seem to have changed a lot since then!

I’m trying to get my head round a lot of this… I figure to get what I can from the software, first I should know what goes on behind the scenes!

 

 

Link to post
Share on other sites

  • 1 month later...
1 hour ago, ALLinONE said:

Can anyone show examples of paired frames from two cameras - reference and emulation? With this approach, for example, the original Leica M9 jpeg, and a paired frame with a conditional canon and the "Leica M9 jpeg" preset.

It would be a nice advise for Cobalt to put that on their website

  • Like 1
Link to post
Share on other sites

  • 2 weeks later...
On 2/15/2022 at 10:49 AM, ALLinONE said:

There are probably problems with emulation accuracy, so there are no such examples on their website.

https://www.fredmiranda.com/forum/topic/1702704/13

If you'd like I can give some other links where to write the same sentences about how bad are our products, how poor are the accuracy and the gamut levels.
Off course always linking the same page , with an issue solved months ago.

is that ok for you R. Vic?
 

Edited by Ulysseita
Link to post
Share on other sites

Ulysseita

I'm just trying to ask people for examples of originals and emulations. If you have solved the problem, please show examples. Everything that I have seen so far does not look like a high-quality emulation, sorry. If I could ask you to process some of my fuji xt4 files, with fuji s5 pro emulation, and the result satisfies me, I will immediately buy 3 of your packages, for a start.

Link to post
Share on other sites

20 hours ago, Ulysseita said:

how bad are our products,

BTW, with all the curiosity I have for these profiles, how does this actually work in this forum with commercial interests of posters? Is there some line which Mr. Kaufmann would not transgress here to praise or promote his products or can anyone walk in to promote elephant leather camera bags for instance? Not meant personal, just wondering

Link to post
Share on other sites

  • 2 weeks later...

Curiosity got the better of me and I bought 4 packs of cobalt profiles. To say I'm disappointed is an understatement. I spent almost 200 euros on something that I will never use.
I have a favorite color camera for many years - fuji s5pro. There were and are different cameras from nikon, canon ... but the result with fuji s5pro still pleases me the most. I wanted to get a picture on my new fuji xt4 like on fuji s5pro, and cobalt-image promises it. https://www.cobalt-image.com/product/ccd-fever/
I was ready to get a non-perfect match, but in reality this does not look like emulation at all. Some other profiles, supposedly emulating, for example, leica, defy explanation at all, how can this be sold ?, it's so bad. All attempts to present dissatisfaction to the developer, they unreasonably interpret as the stupidity of the buyer.
I sincerely do not advise this to buy - an extremely poor quality product.
 

Edited by ALLinONE
Link to post
Share on other sites

Real Fuji s5pro vs cobalt emulation:

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

 

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...