Confusing subject title I know, because I’m confused.
I’m trying to get a bump map working correctly. It’s a very fine grain pattern.
I’m finding if I leave the strength of the bump at 0.5 for example, I can increase/decrease the effect or how ‘aggressive’ it is just by adjusting the filter blur value. So I find myself constantly tweaking bump strength and filter blur to achieve the desired look. Admittedly the standard bump isn’t great and I’m struggling to achieve what I want. If I make the bump to strong it just scatters the light so much that the whole object just gets brighter.
If I use the Corona Bitmap blur, it seems quite unpredicatable. I’m finding if I use the Corona Bitmap I have to increase the strength of the bump to get the same effect as not using the Corona Bitmap and a lower strength.
I’m going to render off some examples, but if anyone has a similar experiences or ‘best practices’ i’d be keen to hear them
So I guess my question is which way to go? Standard or Corona Bitmaps? Higher Strength Lower Filter Blur? Lower Strength Higher Filter Blur? Will any of the combos adversely affect render times?
Which one is correct, in terms of how that light scatter is changing as it’s wildly different between Standard and Corona Bitmap etc?
Ahhh, bitmap filtering… you got not even close to the bottom of this rabbit hole yet :] Add to the equation cBitmap’s filterring interpolation, CoronaBumpConverter node with its own blurring and the fact that filterring stregth depends from surface’s distance from the camera and your head will explode. Personally i try not to touch bitmap filterring and leave it at default size. Use bump strength if you can - that is the safer option.
I guess so, but more importantly does one have a detrimental effect to rendertime? i.e. does a higher strength, lower blur take longer to sample than a lower strength and higher blur etc
keeping always 0.01 will be wrong as it is good for close view renders - more details, but it will be completely wrong looking at distant view as it will looks too matte, also it will be longer to render and may produce noise
The best solution so far is to leave the blur value at 1 and wait for something like the Pixar’s bump to roughness solution to retrieve information loss by filtering.
Never really had any visual/noticeable troubles using this method. I never do animations, so it wasn’t a concern. Usually seems like the best approach was also using summed area versus pyramidal, but also to retain some blurring for bump. I rarely use the coronabitmap, so perhaps it’s time to do some new testing.
This is not just a Corona thing, any renderer that doesn’t properly account for distance filtering in the roughness channel will react with the same visual artifact.
I was present for the Bump to Roughness talk by Pixar this year at Siggraph, very informative.
Well I ended up finding a balance between bump strength and filter blur. As the bump was quite finely detailed, it did make a huge difference. Setting it to 0.01 was massively increasing render time and affecting the light scatter too much almost giving a completely matte appearance. I ended up testing a small patch of the render to get the blur right first so that the bump in render looked like the bump map itself (not too sharp etc), then I adjusted the strength to suit so the physical properties and how it interacted with the scene lighting behaved the same as the physical product I had in my hand, by testing under the same lighting conditions. Admittedly, I’ve always just set everything to 0.01 until now and assumed I was getting the best possible appearance of the map and I won’t be doing again
