Texture Baking VS Texture Making Experiment

So this week I tried to talk 3 different people out of bothering to learn how to use nodes in Blender to bake in specularity and to try to make a spec map in graphics instead. 

There is no point, I said in faking what we can do for real, I said. Nodes are a bit advanced, I said. Even if you want to add a highlight effect, it will be easier to do it in a graphics program, I said. The less you give Blender to do with textures, the better I said. They didn't listen.

I'm not sure I blame them, because I wasn't all that sure I knew what I was talking about either. It was after all just a semi-educated guess. So I decided it was time for some science. 

First off I confess that lately I have been overstating not using Blender to create textures to make a point. The truth is Blender is not that bad and it lets you do texture stuff in really an amazing variety of ways. That said, it's also true that I have a pretty large bias towards Photoshop. I have it and I'm pretty comfortable in there. The same cannot necessarily be said for the 3 Blender users I was trying to persuade away from using Blender to create textures, or for many more Blender users that show up in classes. 

So this experiment uses a texture that was painted entirely in Blender texture paint using sampled photos. I did not even resize or adjust the photos before taking them into Blender. They came straight from CG Textures, to the tip of the texture draw brush in Blender.

After Painting the texture, I saved it as an image, and also baked an Ambient Occlusion map and saved that as well. 

For the fake bake spec version, I loaded up the texture, the ambient occlusion, and a normal map (for more smoothness on this low poly pot) into a material assigned to the pot. Then I set up some lights around the pot. I really did NOT spend enough time on the lighting, and I am sure my highlights suffered because of it. A bit more thought and time spent on lighting would doubtless have made this a closer contest. 

Next I needed a material node set up that would allow me to include specularity in the output - shown below:

  

To be fair to nodes (Compositing in Blender), they are not all that complicated, just something we're not used to (yet). I try to think of them as a diagram of the interactions and blends of pixels, kinda like layers and their operations and blends without the layering. 

So having saved the fake baked specularity texture as an image, I got busy doing things my usual way in Photoshop. As usual, I combined the ambient occlusion map with the texture I painted and saved from Blender, added a solid background layer and saved the image as a 512px png for upload. 

But since this is a contest of specularity effects created in Blender vs a spec map created in Photoshop using relatively familiar skills, I also needed to make a separate spec map. 

Now if there is some right or wrong way to do this, I would love to hear about it, or even an average over all grey scale value range would be helpful. For now, I just try to come up with maps that I think will work based on whatever texture I happen to be using, and then I experiment with the settings in on the edit window in SL. 

All I know for sure about this process is that the lighter the value is, the more specular the highlight can be in SL. So for this texture, since I wanted the rubbed metal bits, to be the ones that reflected light, and since they are a different color than the more oxidized bronze, I decided a Black & White Adjustment would be the way to go.

  
It wasn't dark enough over all.. specularity in SL is really shiny (or maybe that's me lol. again, I don't know what I am doing with this). So I added a levels adjustment to heighten the contrast and make the overall map quite a bit darker.

Then I saved the texture in the image above as a 512px png to use in world. Here's how they looked at the finish line:

  
I was surprised there wasn't more of a difference. In fact I was hoping that the difference would be so striking, and the pot with the spec map (on the left) to be so clearly superior, that not only would people stop bugging me to show them how to bake specular highlights into a texture in Blender, I would no longer need to bug them to go get a viewer with materials and start working on them. 
Maybe next week.