Materials question

[Anonymous]

Would there be any performance difference if just standard colours were used in the modelling environment, instead of physical and appearance materials?

[Beyondforce]

I haven't experienced or heard on any performance issue with standard colors or in the appearance materials in general.

Ben.

[Anonymous]

Thanks Ben, I was just wondering about it and got to thinking that if the program is having to deal with the data behind a physical and an appearance material it has to be using more resources?  

[Beyondforce]

I have created my own appearance materials with high resolution bitmap and it didn't effect the performance.

I'm not quite sure how the data behind works, but you could always make a test and see what happens.

[Anonymous]

I'm not sure that I could do a test because as I and a number of others have posted, the program becomes very sluggish after a certain number of components are created.

I hope that one of the technical guru's jump in here because it would be interesting to know if this is what could be causing the memory trouble?

What would be the best way to test this?

When a component is created it is designated a default material, can that material be replaced with a basic colour, in other words just an appearance colour without a material?

[Beyondforce]

In regards to the performance, that could happen because of many reasons. One of the reasons, that people usually don't consider, is there own operating system. I'm on Win 10 and it is very commanded to reinstall the computer once a year. I have reinstalled mine 2 week ago and I could really feel the different!

In regards to the Default Material, you can change it under Preferences => Material. There you will find many options, But you must have a Material, because you cannot create a body made of Air 😉

If you are experiencing a performance issue with one of your models, you can send it to me and I'll test it on my pc.

Ben.

[Anonymous]

The performance issues start when a certain number of components have been created and this has been reported by a number of people.

If a user has 16gb of ram perhaps the performance issues disappear, maybe with less ram the material data causes problems?

The red flag for me is that the components can be very basic, 400 cylinders can cause problems, as can the basic shapes on a mountain bike tyre.

The shapes are simple but every single one of them is carrying material data, could that be the reason things slow down?

[Beyondforce]

The best way to confirm that there is not problem with your computer or software, is to test it on another pc.
I have seen model with 1200 components and there was not performance issues.

[Anonymous]

The indicator that it is in the program and not in the pc is that Blender runs like a dream on the same machines when many more components are used, the difference may be because of how Blender handles material data.

[Beyondforce]

You can only compare Apple to Apple and not Apple to Orange!

[Anonymous]

Sorry Ben but I don’t think that’s true in many cases, they are both graphics programs running on the same machine so even though they are different we can compare their performance when doing similar operations.

I think this needs looking into, think about it logically.

If one program has 700 components all carrying high resolution material data and high resolution appearance data, and the other program is just carrying basic colour data, logic tells me that the program using the high resolution data is going to have to work harder.

[Beyondforce]

I'm not saying that you are not right. I'm just saying from my 14 years of experience in the IT industry and I have worked many different kinds of software, I always compared the software performance on another machine and not to another software.
If a lot of people having the same issue like yours, then you are probably right. I'm pretty sure, Blender and Fusion are not programmed the same way and not utilizing the hardware the same way.

Maybe some other experts in here could share their wisdom.

[Anonymous]

Unfortunately my cunning plan did not work out.

I just tried a test in Blender because when I compared performance last time I created an object and patterned it numerous times without the object having a material.

This morning I gave the object a very glossy material and then patterned it again but Blender did not even notice the material or the 800 objects in the scene. With Blender it’s like they are not even there, Fusion 360 would have been groaning if I had tried the same thing.

[Beyondforce]

Would you like to Skype, it will be much easier to troubleshoot this way?

[Anonymous]

Thanks for the offer Ben but it is not that much of a problem for me as I don’t work with large assemblies and Fusion 360 runs like a dream when creating my small assemblies

This is related more to the problems I have seen other users post about having trouble with large assemblies, and I do love a good problem.

I find it interesting and would like to hear from a programmer if carrying all that material/appearance data could cause performance issues.

[Beyondforce]

I bet that it's not the material the causes the problem!

[TrippyLighting]

We need to separate what sort of performance we are talking about.

Viewport performance clearly can be influenced by appearance materials. This is obviously more of a problem for older computers. If I use one of the solid wood materials viewport performance is slower than whee I just use the solid color with color cycling enabled. Also larger hi-rez textures require more graphics performance and require the graphics card to do more work.

Comparing Blender with Fusion 360 or a game engine is useless for several reasons.

Blender is a pure mesh modeler. All it works with is meshes. That means it computes a lot less data when navigating the viewport as it also does not employ a LOD (Level of Detail) mechanism as far as I know. I've worked with Blender for 14 years.

Surfaces in Fusion 360 are mathematical creatures with infinite resolution so in order to render it in the viewport first a mesh hat to be created on the fly and the LoD algorithm then needs do adjust mesh density in order to reduce the number of triangles created. That requires a lot more processing power than just displaying an existing mesh.

I've read many in depth articles about computer graphics drivers over the last 25 years and this is a very complex subject.

Also, if you have many objects visible in the viewport and they are all selectable, you'll notice a differences in viewport performance when the mouse cursor hovers in proximity of these objects. The Fusion 360 cursor looks for selectable objects in a cone volume protruding from the mouse cursor. The more objects are in that cone, the more the algorithm has to search.

Using selection sets to set groups of parts to be unselectable can make a big difference in navigating larger assemblies in the viewport. 

[Anonymous]

Thanks for that, I did take into account Blender only dealing with meshes, the comparison is not really a program to program comparison.

I am using Blender as a benchmark to show that the machine is capable and not at fault, I can get Blender to grind to a halt but it takes a real effort.

I am thinking about two things here, is it wise to use materials in the modelling environment and what is the point?

We only need physical materials and appearance materials for simulation and rendering, would it not make sense to keep the modelling materials as just basic colours?

[TrippyLighting]

--- Opinion Alert ---

AO, ground reflections etc. are not so particularly useful when modeling.

They attract people because things look usually appealing. I personally turn them off.

As it pertains to materials, if you want to share an end product with a Customer in A360 it's nice to have materials and decals applied (Decals are currently not showing in A360 😕 ), so I have to have materials assigned even in the Model environment. I usually simply turn color cycling on to take care of that.

If you don't want to share projects, or visual appearance is not so important you can live without appearance materials.

However, I can only say that imagery sells 😉 

[jeff_strater]

Sorry, just catching up on this thread.  @TrippyLighting is correct, it depends on what aspect of performance you are interested in.  For instance, pure graphics performance can be measured by how "viewing" operations perform.  Meaning:  view rotate, panning, zooming, etc.  More on that in a minute.  There are lots of other measures of performance:  Modeling performance (which, in turn, is dependent on what feature you are doing at the moment), open and save performance (which can be affected by internet connection speed), etc.

For the original question (is"performance" affected by material?), we need to dig a little deeper.  Appearance material should only affect that category I called "graphics performance".  That is, the performance of viewing operations.  As an aside, as far as I know, physical material should not affect performance in any way.  Regarding appearance material and graphics performance, it is certainly possible that there could be some effect.  It is possible that, on some graphics cards, graphics performance for some of the fancier appearances could slow down rendering.  This is not my area at all, I need to loop in the graphics folks ( @chengyun.yang?)  I have never noticed any difference when I'm in "component color cycling" (simple colors) mode and regular rendering mode, even with the more complex materials.  I've seen way more impact by turning off some of the graphics effects (antialiasing, ground reflections, ambient occlusions, etc).  Plus, I'm sure that, even within different appearances, there will be degrees of effect.  Yes, the "solid wood" appearances are likely to be the most expensive.  Let's see what Chengyun says.

One last point:  There is one viewing operation that can have a significant one-time performance penalty:  zoom.  Have you ever noticed your model "freeze" for a bit when zooming in?  I can explain what is going on there.  By default, Fusion only facets the model to a single level of detail when you open it.  It is pretty coarse.  This is to maximize performance.  So, the first time you zoom in close, you may see this hitch in your zoom.  What is going on behind the scenes is we are refaceting the model at a much higher level of detail.  This is completely independent of material.  

Jeff