[Sidefx-houdini-list] Linear Workflow

David Burton daland at withatwiststudio.com
Tue Oct 5 15:05:23 EDT 2010


Thanks everyone, that's what I thought but wanted confirmation.  It's too easy to mess this stuff up!  :)

-david

On Oct 5, 2010, at 9:44 AM, Edward Lam wrote:

> On 10/4/2010 11:13 PM, David Burton wrote:
>> Does this workflow assume the monitor gamma (IE: what I would adjust in the graphic card settings)
>> is set to 1?
> 
> In a word, yes. The setting in video drivers are gamma *correction* 
> settings. This is NOT desired if you're following Magnus' description 
> because he is already telling mplay to apply the gamma correction when 
> displaying linear images. If you changed your video driver gamma 
> correction settings, then you will be double correcting, and therefore 
> get the wrong output.
> 
> Regards,
> -Edward
> 
>> On Aug 27, 2010, at 5:10 AM, Magnus Pettersson wrote:
>> 
>>> Just to summarize for some people who might still be confused by all this
>>> (most likely :P) what to actually do. This is "the ballpark" linear workflow
>>> without LUTs and so on. I assume all textures and colors is already sRGB
>>> (2.2) space (if stuff looks good on your monitor, it is most likely in sRGB
>>> and need conversion to linear):
>>> 
>>> 1. ALL shaders should have their input colors/textures converted to linear
>>> ->  apply 1/2.2 to all textures and color parameters (including point
>>> colors). In houdini right now it could equal to add a multconstant node
>>> after a color parameter before its piped into something (like a mult etc).
>>> 
>>> 2. Ok now all your inputs should be correct, time to Render. Now to view
>>> this "in the ballpark sRGB" put the mPlay gamma to 2.2 and this should give
>>> you a good appreciation of how your render looks (if your monitor colors
>>> isnt totally wacky then you should go buy/borrow a monitor calibrator!). And
>>> this ofcourse means if you watch the picture in mPlay gamma 1.0 it will look
>>> dark (because of your monitors gamma correction making it darker).
>>> 
>>> 3. When happy with the render looking at it with this mPlay gamma 2.2 you
>>> can send it off to comp. Like people said Nuke comp artists just take in all
>>> your renders with linear space and everyone should be happy :)
>>> 
>>> The reason we have this very confusing gamma 2.2 here and 0.454545 there and
>>> some makes stuff darker (monitors) and other brighter (like mPlay gamma) etc
>>> is because of the EVIL old CRT screens had a negative gamma (technically
>>> limitation) and to correct for this every picture produced, every GUI
>>> produced and most likely every photo taken and every webpage have been
>>> having a 2.2 positive gamma applied (mostly in background without user
>>> knowledge) to correct for the CRTs shortcomings. BUT now a days with our new
>>> technology our LCD screens have no problem showing right colors in linear
>>> (which would be the best for us in this business so we wouldnt need to
>>> care).
>>> But if LCD screen manifacturers from the beginning would give you a screen
>>> in linear space as default noone would buy them because people would go "omg
>>> this new technology is crap, everything is ugly!" so they have applied this
>>> negative gamma curve to display the content of all programs and all internet
>>> "correct".
>>> 
>>> Myself would like to see linear becoming the new standard, but that would
>>> require all digital content that is shown on a monitor to be chaged. So most
>>> webpages on the internet would have to change its colors, all pictures
>>> produced until now need to be converted to linear and so on, which you can
>>> understand is a massive undertaking. Not to mention all new content made
>>> needs to be made in linear (photos, programs gui, webpages etc)
>>> 
>>> Because of a very small procentage of the people who owns a monitor or
>>> creates the GUI/webpages/images found in this digital age is actually aware
>>> of this linear space mumbo jumbo, so seeing a change in this in the
>>> near/distant future is very unlikely so we will have to continue our sRGB to
>>> linear (to sRGB) pipeline.. i blame CRT!
>>> 
>>> Magnus Pettersson | Effects TD
>>> Storm Studios AS
>>> www.stormstudios.no
>>> 
>>> 
>>> On Thu, Aug 26, 2010 at 9:23 PM, Andy Nicholas<andy at andynicholas.com>wrote:
>>> 
>>>> Thanks Edward, yep, I think this is the piece of the jigsaw I'm missing.
>>>> 
>>>> I need to go away and do some research around all this, particularly the
>>>> perceptual bit. Once I set it straight in my head, I'll write a summary on
>>>> my website.
>>>> 
>>>> Thanks again,
>>>> 
>>>> A
>>>> 
>>>> 
>>>> 
>>>> On 26 Aug 2010, at 19:39, Edward Lam<edward at sidefx.com>  wrote:
>>>> 
>>>>> On 8/26/2010 3:45 AM, Andy Nicholas wrote:
>>>>>> 
>>>>>> What also didn't help my understanding was by putting Houdini's
>>>>>> colour correction gamma setting to 1/2.2 it made a pure linear ramp
>>>>>> (defined by a shader) in Houdini look correct in the render (i.e. an
>>>>>> even distribution of black through white), but using a value of 2.2
>>>>>> made it look anything but linear and over exposed the dark areas.
>>>>> 
>>>>> You're missing the fact that the eye itself also has a nonlinear
>>>>> response. When you put in a 2.2 value into mplay, you're now undoing the
>>>>> power curve that the monitor applied, giving you linear _luminance_. Do
>>>>> not confuse this with _perceptual_ lightness.
>>>>> 
>>>>> If you want *perceptual* (lightness) linearity, then you need to
>>>>> generate an approx. 0.4 power luminance curve [1], explaining why
>>>>> putting an mplay gamma value of around 0.4 "works".
>>>>> 
>>>>> Lastly, this issue has nothing to do with Houdini in particular. Take a
>>>>> linear gradient in another package (say Photoshop) and you will see the
>>>>> same thing.
>>>>> 
>>>>> Regards,
>>>>> -Edward
>>>>> 
>>>>> 1. http://www.poynton.com/notes/colour_and_gamma/GammaFAQ.html#lightness
>>>>> _______________________________________________
>>>>> Sidefx-houdini-list mailing list
>>>>> Sidefx-houdini-list at sidefx.com
>>>>> https://lists.sidefx.com:443/mailman/listinfo/sidefx-houdini-list
>>>> _______________________________________________
>>>> Sidefx-houdini-list mailing list
>>>> Sidefx-houdini-list at sidefx.com
>>>> https://lists.sidefx.com:443/mailman/listinfo/sidefx-houdini-list
>>>> 
>>> _______________________________________________
>>> Sidefx-houdini-list mailing list
>>> Sidefx-houdini-list at sidefx.com
>>> https://lists.sidefx.com:443/mailman/listinfo/sidefx-houdini-list
>>> 
>> 
>> _______________________________________________
>> Sidefx-houdini-list mailing list
>> Sidefx-houdini-list at sidefx.com
>> https://lists.sidefx.com:443/mailman/listinfo/sidefx-houdini-list
> 
> _______________________________________________
> Sidefx-houdini-list mailing list
> Sidefx-houdini-list at sidefx.com
> https://lists.sidefx.com:443/mailman/listinfo/sidefx-houdini-list
> 




More information about the Sidefx-houdini-list mailing list