High Poly/ Sud D Bake - Is it always necessary?
polycounter lvl 5
Hi PolyCount.
I am coming to the end of an Environment project and I'm looking to save time on my workflow with some of my assets. I found myself asking the question today whether it is always necessary to have a high poly sub d model to bake down to a low poly model even if its just to achieve a smoother look and more rounded edges (particularly ones at 90 degrees). Up until now, for every asset I create I have always had a high poly model to bake down to a low poly. For some, it is used to achieve extra details that give the illusion of extra geometry which justifies its use. Extra details are then added to this normal map bake result through the use of Quixel suite tools such as DDO and NDO.
However, For some situations (where there is no need for extra detail on the high poly to fake extra geometry) the only benefit i see is slightly smoother look and rounded edges. I have found that for a small cost I can achieve the same effect from just giving sharp corners a slight bevel with smoothing or adding a few edge loops. In a few articles i found online, this practice is advised as it improves the look of your model, improves the silhouette and reduces jagged edges. More often than not I find myself having to do this anyway in order to get an effective bake. But if that's the case and I can get the same result from just beveling then why waste time with a high poly model at all? The only details that the normal map provides are surface details that come from the type of material. In some cases I find myself using NDO to add details making the use of a high poly model even more redundant.
To give you another situation, If I have a model where I need to achieve extra detail with a normal map bake but only in a specific area, would you say "create a high poly detailed part for the specific area only?" or " Create a high poly model version of the whole thing with the added detail in the specific area?". If you work in industry or in a company, what are your thoughts on this thinking and what would be your workflow if this situation accorded in your work. Is it acceptable to only create high poly detailed parts to bake to specific areas of a low poly model?
I am coming to the end of an Environment project and I'm looking to save time on my workflow with some of my assets. I found myself asking the question today whether it is always necessary to have a high poly sub d model to bake down to a low poly model even if its just to achieve a smoother look and more rounded edges (particularly ones at 90 degrees). Up until now, for every asset I create I have always had a high poly model to bake down to a low poly. For some, it is used to achieve extra details that give the illusion of extra geometry which justifies its use. Extra details are then added to this normal map bake result through the use of Quixel suite tools such as DDO and NDO.
However, For some situations (where there is no need for extra detail on the high poly to fake extra geometry) the only benefit i see is slightly smoother look and rounded edges. I have found that for a small cost I can achieve the same effect from just giving sharp corners a slight bevel with smoothing or adding a few edge loops. In a few articles i found online, this practice is advised as it improves the look of your model, improves the silhouette and reduces jagged edges. More often than not I find myself having to do this anyway in order to get an effective bake. But if that's the case and I can get the same result from just beveling then why waste time with a high poly model at all? The only details that the normal map provides are surface details that come from the type of material. In some cases I find myself using NDO to add details making the use of a high poly model even more redundant.
To give you another situation, If I have a model where I need to achieve extra detail with a normal map bake but only in a specific area, would you say "create a high poly detailed part for the specific area only?" or " Create a high poly model version of the whole thing with the added detail in the specific area?". If you work in industry or in a company, what are your thoughts on this thinking and what would be your workflow if this situation accorded in your work. Is it acceptable to only create high poly detailed parts to bake to specific areas of a low poly model?
Replies
There's no reason why you have to make a highpoly in every case. Sometimes it isn't needed. If you can get a good enough result in a much quicker way.. what's the problem?
Geometry is, apparently, pretty cheap these days. But if you're going to be using a normal map on the model anyway (which has a performance cost) it makes sense to get the most you can out of it, given sufficient time to make the highpoly. If you think you'd get a better end result spending the time you would've on the highpoly on something else, fine. If you can recognise cases where you can use just slightly more geo and you won't even need a normal map at all, brilliant. But I don't think that will happen too often.
You might find these two recent threads interesting:
http://www.polycount.com/forum/showthread.php?t=140825
http://www.polycount.com/forum/showthread.php?t=142657
Do whatever gets you the best results the fastest, and at a relatively cheap rendering cost. If this means a low poly with bevels and a low-res normal map for any tiny details you want created in dDo, then so be it.