[Sidefx-houdini-list] Linear Workflow

Agent Xray xray at agentxray.com
Thu Aug 26 13:16:52 EDT 2010


  A good explanation of "what is sRGB?" :

http://mysite.verizon.net/spitzak/conversion/srgb.html


-=X

On 8/26/2010 6:28 AM, Gene Dreitser wrote:
> Not to add to the confusion -- I'm still grokking this myself, I thought I would
> point this out just in case it was relevant to somone:
>
> While 'most of the time' people equate the sRGB to linear color space transform
> with a application of a 0.45 gamma colour correction (and the inverse operation,
> by applying a gamma 2.2 colour correction), this is infact, an approximation.
>
> The ACTUAL colour transformation is slightly more complex, and involves a LUT.
> you can confirm this yourself in Nuke rather easily.  There is a portion in the
> black that will get obliterated if you apply a 0.45 gamma, that won't be using a
> proper lut.  subtle, but present.
>
> Best,
>
> G
>
>   --
> Gene Dreitser
> Loki Visual Effects
> p: 416.532.5654 // c: 416.803.0682 // www.lokivfx.com
>
>
>
> ----- Original Message ----
> From: Rangi Sutton<rangi.sutton at gmail.com>
> To: sidefx-houdini-list at sidefx.com; andy at andynicholas.com
> Sent: Thu, August 26, 2010 9:06:01 AM
> Subject: Re: [Sidefx-houdini-list] Linear Workflow
>
> On 26 August 2010 20:13, Andy Nicholas<andy at andynicholas.com>  wrote:
>
>> Hi Rangi,
>>
>> Following the link to Wikipedia, the first line reads:
>>
>> "Gamma correction, gamma nonlinearity, gamma encoding, or often simply
>> gamma, "
>>
>> Kapow!
>>
>> That's exactly the problem. People often talk about correcting for gamma.
>> So does this mean what they're actually saying is that they're correcting
>> for gamma correction? I guess that'd be called gamma correction correction
>> ;-)
>>
> Fair enough. It's all confusing as hell. I'll try and spell it out
> randomly...
>
> As far as I understand it... Your display is constantly darkening
> everything. Most images we deal with are sRGB, they have actually got
> brighter values.. so when they're displayed, everything turns out equal.
>
> (ie.. a grey value of 0.5 in sRGB, should be displayed as 20% grey. There's
> a stack more information in the lower-luma ranges than the higher luma
> ranges.)
>
> Of course.. terms like brighten and darken are simplifications.. you know
> what I mean.. mid grey goes up == brighten, mid-grey goes down == darken ;)
>
> Type "xgamma -gamma 2.2" to see you monitors in linear space. Your display
> is now neutral, with a gamma of 1.0 (assuming it was "normally" calibrated
> to begin with!) Everything looks too bright, your interface is washy, it
> wasn't designed in linear space.
>
> If you open a jpg now it will look washed out (it's in sRGB space, and your
> display is now linear). Open a linear render and it will look true, although
> you might need to adapt what you think is true!
>
> Apply a gamma of 0.45 to your jpg with a gamma cop or something, and you've
> converted it from sRGB to linear (pretty much). The image is darker now..
> looks correct.
>
> So  when a monitor is said to have a gamma of 2.2.. it means that 1/2.2 gets
> applied to everything. You should "gamma correct" a linear image to your
> viewing device by applying a gamma of 2.2 to it.
>
> Take a linear image, apply gamma 2.2, display on a monitor with gamma 2.2..
> it will look right.
>
> Take a photograph, your camera inherently applies a gamma of 2.2, display it
> on your "normal" monitor (has a gamma of 2.2), it will look right.
>
> Have I contradicted myself? Someone correct me because in Southern
> hemisphere we might have upside down denominators...  and I usually just
> flip shit around until it looks right.
>
> Nuke is a good one for this stuff.. allows you to specify what colour space
> images are in on the way in (rather than assume by image type, which I think
> cops does).. and allows you to simply control the viewer space. Nuke likes
> to work linear internally, as far as I understand it.... which is how stuff
> should be composited!
>
> I'm going to hit send then realise I said it all backwards....
>
> r.
> _______________________________________________
> 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