Hmm, overly vague thread title.
Hi guys, i'm a long time lurker who's finally decided to post. I'm currently involved in 3d animation but I'm more from an artistic background than a technical one as this post will probably suggest. Anyway, i've got a few random questions/personal musings which have been bugging me for a while and I was wondering if you guys could shed some light on them...
1)Animation on normal maps - with normal maps being used more and more strongly these days and in some cases replacing true high poly meshes it seems to me that this may have a negative effect on animation. Take an example where you have a characters head which furrows on his forehead generated from a normal map. These should only really be present in certain facial expressions/animations (i.e. in a frown they are fine but in a smile you might want them to dissapear). To me it seems that it wouldnt actually be possible to animate them in any way as they will always be fixed on the surface. How do people get around this kind of thing?
2)Stencil/Profile mapping - I forget the genuine term for this but I seem to remember seeing an example somewhere of a normal mapped low poly model with it's profile calculated from a high poly version. I can't remember the tech behind this but the results looked really impressive, is this technique actually being fully investigated for use?
3)Paint FX in game engines - As a maya user i'm always impressed at how you can grow vegetation and other objects relatively quickly and easily with paint effect. Again I don't fully understand the technical implementation of these but I've been they seem to be a half 2D, half 3D post process effect which I've been told are quite well suited to hardware acceleration. Again is anyone looking into this?
4)Procedural textures - Why doesnt anyone seem to use them? I'm aware that it can be quite hard to get convincing results for certain materials but a lot of things work well procedurally. Surely there are some quite significant memory advantages to procedural textures as well as the fact that you can continual zoom in to them and maintain accuracy. Are there any technical downsides to using them or is it just that developers find it hard making the implementation look good?
5)3d textures - Will we ever move away from the idea of a 2D image stretched over a 3d object and actually treat the texture as being genuinely 3d? There seems to be a lot of advantages to just dropping a 3d wood/metal shader onto an object and having it react accordingly regardless of shape and UV mapping. Also it seems to be a step towards a more realistic representation of a materials behaviour if there appearance is genuinely based on their 3d attributes.
6)Polygon smoothing/tesselation/HOS - Has this kind of technology been truly abandoned? I find when modelling I tend to make sure my model can be polysmoothed without glitching so that when I come to rendering I can use varying smooth levels depending on the distance from the camera. I'd guess that high poly models for games involve smoothing at some point in there creation and it seems wasteful to not integrate this more with the game engine. Surely you could improve LOD systems by using distance dependant smoothing and it could also be a useful way of scaling the detail levels for peoples systems.
7)Post processing - Is anyone seriously looking into post processing as a means to create a more cinematic experience? Far too many games these days have a disjointed colour scheme and an atmosphere which I feel could be vastly improved with colour balancing and simple filtering techniques. Are these things possible to do on the fly with current hardware generations?
8)Procedural content - As content creation time becomes more and more of a problem are there any interesting new ways of generating things procedurally? Apart from terrain height maps I can't think of much procedural generation that i've spotted in use. Does anyone use L-systems for generating vegetation or organic objects?
9)Shared content - To me it seems hugely wasteful that for each major game these days an obscene amount of time is spent just making the basic groundwork for the game. As well as engine creation artists must spend countless hours making basic objects/decorations (human head/bodies, mp5's...) and scouring the globe for ever more detailed texture sources. It just seems stupid for each major game to have to do this independently every time they start on a new project. Has anyone ever thought of setting up a high quality professional middleware library available to all games developers containing all this source material?
10)Shared engines - As engines get more and more advanced it seems pointless for developers to try and develop their own when there are already engines out there which have everything they could need. Surely once you have an engine that has all the features you'd expect in a 3d application but available in real time there doesnt seem much point in using anything else.
Sorry for my poorly informed ramblings but I'm curious what anyone thinks about these subjects.
Hi guys, i'm a long time lurker who's finally decided to post. I'm currently involved in 3d animation but I'm more from an artistic background than a technical one as this post will probably suggest. Anyway, i've got a few random questions/personal musings which have been bugging me for a while and I was wondering if you guys could shed some light on them...
1)Animation on normal maps - with normal maps being used more and more strongly these days and in some cases replacing true high poly meshes it seems to me that this may have a negative effect on animation. Take an example where you have a characters head which furrows on his forehead generated from a normal map. These should only really be present in certain facial expressions/animations (i.e. in a frown they are fine but in a smile you might want them to dissapear). To me it seems that it wouldnt actually be possible to animate them in any way as they will always be fixed on the surface. How do people get around this kind of thing?
2)Stencil/Profile mapping - I forget the genuine term for this but I seem to remember seeing an example somewhere of a normal mapped low poly model with it's profile calculated from a high poly version. I can't remember the tech behind this but the results looked really impressive, is this technique actually being fully investigated for use?
3)Paint FX in game engines - As a maya user i'm always impressed at how you can grow vegetation and other objects relatively quickly and easily with paint effect. Again I don't fully understand the technical implementation of these but I've been they seem to be a half 2D, half 3D post process effect which I've been told are quite well suited to hardware acceleration. Again is anyone looking into this?
4)Procedural textures - Why doesnt anyone seem to use them? I'm aware that it can be quite hard to get convincing results for certain materials but a lot of things work well procedurally. Surely there are some quite significant memory advantages to procedural textures as well as the fact that you can continual zoom in to them and maintain accuracy. Are there any technical downsides to using them or is it just that developers find it hard making the implementation look good?
5)3d textures - Will we ever move away from the idea of a 2D image stretched over a 3d object and actually treat the texture as being genuinely 3d? There seems to be a lot of advantages to just dropping a 3d wood/metal shader onto an object and having it react accordingly regardless of shape and UV mapping. Also it seems to be a step towards a more realistic representation of a materials behaviour if there appearance is genuinely based on their 3d attributes.
6)Polygon smoothing/tesselation/HOS - Has this kind of technology been truly abandoned? I find when modelling I tend to make sure my model can be polysmoothed without glitching so that when I come to rendering I can use varying smooth levels depending on the distance from the camera. I'd guess that high poly models for games involve smoothing at some point in there creation and it seems wasteful to not integrate this more with the game engine. Surely you could improve LOD systems by using distance dependant smoothing and it could also be a useful way of scaling the detail levels for peoples systems.
7)Post processing - Is anyone seriously looking into post processing as a means to create a more cinematic experience? Far too many games these days have a disjointed colour scheme and an atmosphere which I feel could be vastly improved with colour balancing and simple filtering techniques. Are these things possible to do on the fly with current hardware generations?
8)Procedural content - As content creation time becomes more and more of a problem are there any interesting new ways of generating things procedurally? Apart from terrain height maps I can't think of much procedural generation that i've spotted in use. Does anyone use L-systems for generating vegetation or organic objects?
9)Shared content - To me it seems hugely wasteful that for each major game these days an obscene amount of time is spent just making the basic groundwork for the game. As well as engine creation artists must spend countless hours making basic objects/decorations (human head/bodies, mp5's...) and scouring the globe for ever more detailed texture sources. It just seems stupid for each major game to have to do this independently every time they start on a new project. Has anyone ever thought of setting up a high quality professional middleware library available to all games developers containing all this source material?
10)Shared engines - As engines get more and more advanced it seems pointless for developers to try and develop their own when there are already engines out there which have everything they could need. Surely once you have an engine that has all the features you'd expect in a 3d application but available in real time there doesnt seem much point in using anything else.
Sorry for my poorly informed ramblings but I'm curious what anyone thinks about these subjects.
