Hi there
when I render an image out from Houdini and open it in After effects or photoshop the colours become much less saturated.
I discovered after a bit of trial and error that if I converted them to 8 bit
(they were in 16bit ) then all would be well but obviously limiting me to 8bit colour.

I was wondering if any one has any good advice for referencing colours outside of Houdini and making sure they stay the same when exporting.
I guess this relates to comping, how do people go about matching colours
in a plate and making sure they render out as matched?

cheers
John

How are you comparing the colours? Note that mplay doesn't apply a 2.2 gamma to the linear 16-bit FP images by default. So when you're looking at the image in mplay, ensure that the gamma value is set to 2.2 (or whatever your monitor's gamma).

To set your gamma appropriately, in Houdini, go to Edit > Color Settings.

Hi Edward
thanks for your response.
I'm rendering stills then opening them in after effects and photoshop.
Things look fine in Houdini and M-play , it's just in adobe software where there is a desaturation based on the bit depth. 8 bit fine , 16 bit desaturated.

I did some scouting based on your ‘gamma’ tip and found that there's a whole lot of options based on gamma correction in adobe for when your importing HDR, or converting to a lower bit depth.

http://help.adobe.com/en_US/Photoshop/11.0/WSfd1234e1c4b69f30ea53e41001031ab64-78e8a.html [help.adobe.com]

I'll have a play with them and see what works best.

Where does the gamma value of 2.2 come from?
thanks
John

The 2.2 gamma value comes from being an approximation of sRGB [en.wikipedia.org]. Again, this is only an approximation. The precise gamma value depends on your monitor and room lighting conditions.

If you're looking at a 16-bit floating point (rendered) image in mplay with a gamma of 1.0, then you're viewing it wrong. The more correct viewing results are:
- mplay with the “gamma” set to 2.2 (or thereabouts)
- what you see when loaded in Photoshop (which performs gamma correction)

So one way to make the two match are:
1. Load up the rendered image in mplay and Photoshop
2. In mplay, change the “gamma” value until it matches the one in Photoshop
3. Load up Houdini and choose from the main menu, Edit > Color Settings > Color Correction. Change the “gamma” value here to the one in mplay which you found that matches Photoshop. Enable all the checkboxes. Click on Apply and Save as Default.

The best thing is to light your renders using the correct gamma setting to begin with.

Now, if you've already lit your project and want the Photoshop results to match your non-gamma corrected images, then you can adjust it in Adobe. In Photoshop, the way to do this is to go to Image > Adjustments > Exposure. Adjust the Gamma slider until it matches the image in mplay. It will most likely be around 2.2. NOTE: You should *NOT* want to do this before compositing!

Thanks a lot Edward that's a very thorough answer.
So calibrate mplay and get your colours and lighting right in Houdini first.
That makes sense so that you've got more room to play with and adjust in your compositor.
Just one thing - could I be confusing things by bringing 8 bit images into houdini's display as my reference which don't need gamma correction?
thanks a lot for the advice
John

And what about the mantra ROP gamma option ?
Isn't the first setup to do ?

HUGHSPEERS
Just one thing - could I be confusing things by bringing 8 bit images into houdini's display as my reference which don't need gamma correction?

8-bit colour images are typically assumed to be in a sRGB already which is why they don't need gamma correction (eg. web images). The main reason being that 8-bit RGB values don't have enough precision if you encoded them linearly. However, with 16-bit floating point images we have enough precision so it's simpler (at the expense of memory) to keep them as linear.

Mantra outputs everything as linear, so it you made it store as 8-bit linear color values, then you will be losing precision.

fsimerey
And what about the mantra ROP gamma option ?
Isn't the first setup to do ?

It's a good option to enable if your intended viewing device is gamma 2.2 or sRGB. This option does NOT change how your colour values are encoded.

http://www.sidefx.com/docs/houdini11.0/props/mantra
vm_colorspace / renderer:colorspace = ('linear')

Sampling color space for variance antialiasing. Setting this to Gamma 2.2 will cause darker parts of the image to receive more samples.

I've been colour picking from the 8 bit images in Houdini so I guess this means I've been storing the colour information from 8 bit by doing that which might cause some inaccuracy.
Fortunately I'm not working in production.
To use precise colours stick with 16 bit images then.

I'll have a play and work out what works best for me with houdini to adobe.

thanks for your help Edward and fsimerey.

John

The problem really isn't colour picking from 8-bit images. It's just that you should not *process* (eg. add, multiply, etc) them as 8-bit values because there's not enough precision.

And 8-bits images can't have power color bigger than 1.0.