Pages: [1]   Go Down

Author Topic: Xcalibr v1.0 and Xcalibrator v1.0  (Read 2315 times)

opgr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2376
Xcalibr v1.0 and Xcalibrator v1.0
« on: October 25, 2017, 08:16:42 AM »

I've been chewing on this idea for years, have even produced proof-of-concept applications and code for several pieces of this puzzle. It is now time to finish the puzzle into a workable application. Remember, you read it here first.


Xcalibr is a workflow solution
Xcalibr is a next-generation modular RAW converter that provides a great middleground between a workflow oriented paradigm and a conversion oriented paradigm. It will offer complex processing yet comprehensible editing. Easy asset management in contained batches. Variations for a single file. Clear batchprocessing. Even tethering would be possible with automated processing of incoming files. Simple yet powerful.

Working in Xcalibr is like working with a projectwindow which contains a list of one or more RAW files. The list of files can then be grouped together for selection, stacking, and batchprocessing. The project can be saved as a single projectfile with references to the RAW imagefiles. The RAW imagefiles are sidecar files to the projectfile, not the other way around.


Xcalibr is Modular
RAW file support is offered thru input plugins. The primary task of these inputplugins is to simply provide uncompressed RAW data for a demosaicing engine. They should also provide response parameters to interpret the color of the RAW data. A calibration option is available to create several sets of response parameters for different capturecircumstances. The input plugins do not need to provide demosaicing.

Demosaicing is provided by engine plugins. They are always offered the same raw data as available from the inputplugin. Different methods of demosaicing can thus be selected by the user, including no demosaicing at all. This is much like being able to choose between different "textures" for the final image. But it also allows easy adoption of non-traditional CFA designs like Fujifilm's Xtrans sensors. Considering the name of the application... !?

Adjusting the file will mostly happen in capture space, or a perceptual version of that. Feedback will be provided by a colormanaged preview, and a histogram which will allow an Fstop view for the actual captured RAW data, or a perceptual outputspace view for the final image. The traditional controls will be supported by the host, but additional filters will be allowed as plugins. B&W conversion is a good example, which should preferably be done after producing a desired colorcorrect result from the RAW data. This allows useful options like colorfiltering before B&W processing and additional grading of output contrast.

The core of Xcalibr's colorrendering is aimed at providing a more-or-less correct colorappearance. The increase of saturation and contrast should subsequently be entered by the operator as a conscious choice. Clearly, renderingstyles can be saved as defaults if so desired. Filmsimulations with contorted colorspaces can easily be added as a plugin filter.


Xcalibr is a new whitebalance paradigm
It will also encompass a new whitebalance paradigm. Away from the incorrect and rather mathematical "temperature & tint", towards a more traditional and better understood "Response & colorcast". Photographers have traditionally chosen film based on its colorresponse, which included a temperature indication to estimate its whitebalance response. They didn't have the luxury of selecting an entire temperaturescale based on the colortemperature of a scene. In fact, in most cases that wasn't desired anyway, and even today that isn't what the photographer wants.

The photographer usually likes the properties of the overall lighting in a scene. That is, warm yellowish light for a warm overall appearance/experience/feeling. One would go out specifically during "the golden hour". Or cold blue light for a cold overall appearance. The blue hour. Photographers usually want to preserve that color in an image, at least partially.

Imagine for example a sunset on the beach. Clearly the temperature of that scene is very low, in the 3000k region. Your film response however, should still be normal daylight, firstly because that is how the sensor will behave, secondly because that is how our eyes adapt. The sensor doesn't suddenly change to artificial tungstenlight behaviour just because the colortemperature is low. And our eyes do not adapt fully to the light because we still detect the warm tint. But this partial adaptation does not follow the temperaturelocus, so you can not just move the temperatureslider to simulate a warmer or colder colorcast.

Temperature and tint is a misappropriation of both the colorscience as well as the term tint. Response and colorcast not only define behaviour more correctly, it also better fits a traditional photographer's understanding of the process.

In short, Xcalbr will be the ultimate RAW converter for photographers.


tl;dr:

Xcalibr will be a new RAW converter
- modular design
- calibration option
- new workflow paradigm
- comprehensible parameters
- new whitebalance paradigm
- b&w conversion
- film simulation

I'm going to crowdfund this, and you want to be part of it. Stay tuned for more...
Logged
Regards,
~ O ~

opgr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2376
Re: Xcalibr v1.0 and Xcalibrator v1.0
« Reply #1 on: October 25, 2017, 09:05:30 AM »

Useless without picts.

I found some images of my yesteryears proof-of-concept converter!

They show different scales for the histogram:
1. RAW data in Fstops
2. Linear histogram
3. perceptual histogram

Also note the controls in the different tabs.
Logged
Regards,
~ O ~

Lundberg02

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 379
Re: Xcalibr v1.0 and Xcalibrator v1.0
« Reply #2 on: October 26, 2017, 02:05:09 AM »

The histos are obviously different but the images aren't. Why is that?
Logged

opgr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2376
Re: Xcalibr v1.0 and Xcalibrator v1.0
« Reply #3 on: October 26, 2017, 04:15:18 AM »

The histos are obviously different but the images aren't. Why is that?

Because only the horizontal axis of the histo changes depending on mode (Fstop, linear, or perceptual).
In this case the mode changes on tab-selection.
Logged
Regards,
~ O ~

opgr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2376
Re: Xcalibr v1.0 and Xcalibrator v1.0
« Reply #4 on: December 18, 2017, 01:32:49 PM »

Sometimes I think I must be verging on utter genius, and then I do something like this and I realise that that is merely an understatement. I should be nominated for a Nobelprice, me thinks.

;-)

Fuji X-T20 unadulterated RAW,
- normal ICC profile,
- standard gamma 2.2. No contrast curves.
- directly calibrated from the chart.

This is just a preliminary result. The data is read by an import plugin. I'm now working on an engine interface so that the actual conversion is also possible thru plugins.

I'll see if I can attach the profile here. I'm thinking about doing a write up at some point about all the ins and outs of RAW conversion and colormanagement, though in a more pragmatic style so it's still readable for the average photographer.
Logged
Regards,
~ O ~

Lundberg02

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 379
Re: Xcalibr v1.0 and Xcalibrator v1.0
« Reply #5 on: December 19, 2017, 05:25:26 PM »

What is the point of this image? are we supposed to compare it to some web site's version or what?  Also, please stop conjoining words, such as color management, it's annoying to Englishspeakers ( see what I did there).
Logged

xpatUSA

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 384
    • Blog
Re: Xcalibr v1.0 and Xcalibrator v1.0
« Reply #6 on: December 27, 2017, 07:03:21 AM »

What is the point of this image? are we supposed to compare it to some web site's version or what?  Also, please stop conjoining words, such as color management, it's annoying to Englishspeakers ( see what I did there).

Having a look here after a long absence, I see that not much has changed . . .
Logged
best regards,

Ted
Pages: [1]   Go Up