Image blown out when switching to Menta Ray

#1

I am new to Mental Ray and am trying to learn it however I’ve run into a bit of trouble early on. I created a fire hydrant using a Maya Blinn material that renders fine in Maya Software rendering. However once I switch over to Mental Ray rendering it immediately shows the object as blown out. It doesn’t do this for all objects (lamberts weren’t affected) but it seems to do it to the fire hydrant. If you are rendering in Mental Ray are mental ray materials required? I was pretty sure that Blinn’s would still work in Mental Ray but I may be mistaken.

#2

Interesting…I just tried creating a new generic sphere shape with a regular blinn material applied and there were no issues. It rendered fine in mental ray. I remember trying to set-up an occlusion pass on this object before I got 2012, and now I forget where I did that so I can’t remove it. That may not be the issue though…

#3

This is most likely a gamma problem with bringing in 8bit textures. Linear workflow solves this issue.
Also make sure default lighting is disabled under the render globals at the bottom of the common tab, then add a single light to your scene.

But yeah I can’t stress this enough, linear workflow is probably the most important step before learning anything else with mentalray.

It resolves a lot of beginner issues with lighting, and problems you’re likely to encounter if you don’t use it. Such as textures being blown out, overall image being washed out, lights contributing in unpredictable ways. Its really simple to implement, and willl make lighting much less of a headache.

There are tons of threads about it here on CGtalk, so just do a search if you want a better understanding than my quick explanation.

But basically the problem arises because 32 bit renderers like mentalray, vray etc work in a gamma space of 1, when doing all their math and calculations. But 8 bit image formats like jpegs, tiffs, pngs etc have a gamma of 2.2 baked into them to display correctly on monitors. So when you bring in an 8 bit image format into mentalray, the math gets all screwed up because its doing all of its lighting calculations etc as if the input texture is at a gamma space of 1 when infact it is at a gamma space of 2.2 . So the math behind the scenes will be doing stuff like 1+1=5 causing lighting to become blown out etc.

Linear workflow is… applying a degamma to all of your input 8bit textures by the inverse of 2.2 to make all of the colour in your scene linear, at a gamma of 1. Then applying a gamma correction of 2.2 to the final image to display correctly on your SRGB monitor.

Here is a simple 2 step guide to linear workflow in maya by djx … http://www.djx.com.au/blog/2010/03/07/maya-linear-workflow-in-two-steps/

#4

Thank you for the reply & explanation (I favorited that link you provided). I am using targa files for all of my textures, however I just opened it up in Photoshop and went into image/mode and saw that the file was set to 8-bit instead of 16 or 32-bit.

I still have some uncertainties about the different bit rates and what they include (I believe 32-bit includes alpha), but I can do a google search for that.

Thank you for your advice.

#5

I guess another question I would have though is…

If you make sure not to use any 8-bit images in your color channel for a project you are working on would you still need to create and apply all of those nodes to the camera and scene when working in mental ray? I guess it’s probably good practice to do that, but I was just wondering.

#6

Even if you have a 32 bit texture, it is still good practice to have a tone mapper plugged into your camera set to a gamma of 2.2, with render globals gamma set to 1.

  The only difference with using 32 bit textures as inputs would be that you wouldnt have to degamma them. As these dont have baked in gamma curves of 2.2

  I still have some uncertainties about the different bit rates and what they include (I believe 32-bit includes alpha), but I can do a google search for that.

  8 bit images store 256 shades per color channel ( 2^8 = 256 colors). It is also comonly refered to as true color 24 bit, if you add each 8 bit channel you get 24 bits of colour. 16 bit per channel images, can store 65k shades of colour per channel (2^16 = 65536 colors) etc. and obviously 32 bit per channel images, also referred to as floating point images can store 2^32 colors per channel.

Any bit depth image can store an alpha channel, given that you use the correct file format which can store a 4th channel, ie. tiff, tga, png, etc.

I think you were confused between ‘bits per channel’ and ‘bits per image’. If you have an 8bit per channel image with an alpha, it adds up to 32 bits of information(if you add the bits from R, G, B and A). But it is still an 8 bit per channel image. In rendering when you here people talking about bit depth they’re usually referring to per channel.

#7

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.