Topic - 0.5 midpoint?

Hi Jens,

I'm trying out your extension pack presets for 3 at the moment. I've got questions about this section in particular:
http://www.campi3d.com/External/MariExtensionPack/userGuide3v2/MaskfromCurvature.html

In your linear data example, the image that has a 0.5 midpoint is incorrect. While the image with the 0.71 midpoint is correct.

At the moment I have an 8bit paint node set to raw with colorspace off. The midvalue is 0.5, is this also incorrect?

I'm a bit confused by this. I would have thought if I have no color management on, with the paint node set to raw a value of 0.5 is in fact 0.5
The color management in 3 is doing my head in....

Cheers,
Mark
Your case is correct. Because you have color management off, the VIewTransform is turned off as well so 'what you see is what you get'.
By default Mari 3 displays everythign in the viewport - while color management is on - through a sRGB Lookup.

So a value of 0.5 appears brighter (at value 0.71).
If - while color management is on - you are getting a visual result that looks similar to what you get when you are viewing 0.5 in photoshop you are actually doing it wrong.
In order to have a viewport result of 0.5 while color management (and ergo VIew Transform) is on your value would only be 0.21.

The best bet is to create the node 'Grey Constant' and compare the values. The node is set to 0.5 linear data so is always the base line you can compare against.

You should import your Curvature Maps (or any other 'scalar' data) alwasy with setting Linear or Raw in the 'Import Files' Dialog and import it into a channel that is set to 'Linear' or Raw as well.


It's confusing I know. But what can you do.

Message edited by Campi3d on 2/18/2016 - 3:00 PM

Great, thanks for the clarification Jens!
One correction. You still have colorspace on in the color management toolbar ?
Then you are still viewing things through the view transform. Turn it off in the colorspace toolbar. If you 0.5 value looks incredibly dark after turning it off you imported it wrong. With colormanagement turned off in the toolbar uour 0.5 value should actually look like 0.5 in photoshop
Hi Jens,

Sorry I missed this. I had the color space toolbar set to off. I put in a grey 0.5 constant to compare everything just to make sure.
Hey Jens,

we just noticed that when picking raw data of the grey_constant node, the return value is 0.498 and not 0.5. Are you aware of this?

That's in Mari3.0v2.

Romain
Yes, I am aware.
It's on purpose.

Try setting the foreground color to 0.5 and use 'Fill Foreground' in an 8bit channel then pick the color you just filled.
It is 0.498(5) or 128,128,128.
I made the grey constant to be the same in 8bit channels. In 16bit channels you have enough precision to represent the 0.02 difference so filling 0.5 grey in those channels and picking again gives you 'true' 0.5

I can actually make the grey constant true 0.5 no problem but decided to be consistent with the rest of Mari in 8bit and do 0.498 since otherwise you will never be able to paint exactly the same color as Grey Constant.
0.498 to 0.5 is almost indistinguishable with displacements if that's what you are worried about.
So even using it in 16bit channels the 0.015 difference to true 0.5 will not be noticable in terms of 'shrinking' the geo.
The only grievance I have with this is the handling of things like contrasting (which has a default pivot of 'true' 0.5) which results in the grey slightly changing (very slightly)

I debated for a while if I should go with true 0.5 or 0.498 but I guess if someone really has a problem with it I can add a 'precision' dropdown to the node to select between byte and half precision but I want to keep that node light processing wise.

jens

Message edited by Campi3d on 3/11/2016 - 2:50 PM

Thanks for the explanation!