This video, by a Bill Dally (NVidia) about chip architectures is so great it deserves its own post: http://mediasite.colostate.edu/Mediasite/SilverlightPlayer/Default.aspx?peid=22c9d4e9c8cf474a8f887157581c458a1d#
It probably will tell you thinks you already know, but even if it does, it's worth to see it.
04 January, 2012
12 December, 2011
The three skin rendering horrors you want to avoid
It's 2011 and still most games come with some truly horrific skin rendering. I don't think skin is that hard to get decent, but you have to understand what's important and how to hack it in your lighting/rendering model.
This article won't focus on the state of the art skin rendering techniques or on the state of the art acquired data for skin, these things are really important, fundamental, go ahead and google them, there's plenty to read and it's not hidden knowledge at all (start with Debevec and Jensen).
What I want to do here is just a write up of what I've learned about skin in my experience, and some of the common technical mistakes I've seen.
1- Bad tone
 |
| Fallout 3: Radiation really does some numbers on your skin... |
By far, the worst offender. Nailing the right colours from the day to the nightside of the skin is the most important, and most often neglected things. And yes, it is mostly (but not only) a matter of subsurface scattering, as most of the skin diffuse lighting comes from it, but it's not about complicated techniques. It's about understanding what happens, what's important to model and how to hack it in whatever lighting model you have in your rendering.
 |
| Model: Commander Shepard @ Bioware |
 |
| Model: Charlotte Free @ IMG/NY |
You can get great skintones with even the simplest hacks on top of your basic lambert, and horrible tones by badly tuned texturespace or screenspace diffusion techniques. Penner's pre-integrated skin shading is the current champion of the lambert hacks, but truth is that some sort of ramp on top of your lambert is everything you need (that's to say, you can most often get decent results without even bothering to approximate the geometry with the curvature, as Eric does) if you nail the right hues!
Also, remember that scattering will affect all the skin shading, you might want to "ramp" the edges of the shadowmaps (again, Penner's paper covers that) but it's actually often even more important to properly blend AO on skin, especially if you're using a normal-aware SSAO which is capable of capturing very fine details, a straight multiply will create some really questionable gray patches. Again, even simple hacks, like adding a slight constant colour to your SSAO, will go a long way towards creating more organic skin.
Also, remember that scattering will affect all the skin shading, you might want to "ramp" the edges of the shadowmaps (again, Penner's paper covers that) but it's actually often even more important to properly blend AO on skin, especially if you're using a normal-aware SSAO which is capable of capturing very fine details, a straight multiply will create some really questionable gray patches. Again, even simple hacks, like adding a slight constant colour to your SSAO, will go a long way towards creating more organic skin.
 |
| Debevec, Digital Emily, Acquired Diffuse |
Give your artists the right tools! It's fundamental to be able to tune the skin while comparing the rendered results with some reference material (put an image viewer and color sampler in game!), it's fundamental to tune colors after fixing your tone-mapping operator, and it's fundamental to understand _all_ your sources of color: shading model, parameters, textures.
 |
| Debevec, Efficient Estimation of Spatially Varying Subsurface Scattering Parameters |
In practice, especially in the realtime rendering world, we mix together all the skin layers and more often than not the problem with skin textures is that they are not saturated enough (consider that the white, additive specular sheen should be responsible for some of the loss of saturation in the final rendering) and have too uniform hues (skin hue shifts quite a lot in many regions, i.e. elbows, joints, hands and feet, and usually presents quite a few blemishes)
2- Bad detail
Skin detail does not come from diffuse! That's quite obvious as we said that diffuse lighting is mostly due to subsurface scattering.
 |
| Debevec, Digital Emily, Specular |
Detail is all in the specular layer, ideally, you would need two different normalmaps for diffuse and specular (actually, Debevec achieves good results rendering with different normals for each of the RGB channels of the diffuse), fetching the same at different miplevels (bias) is often good enough and in a pinch, even just using the geometric normals for diffuse (or a lerp between the normalmap normal and the geometric ones) is way better than using the specular normals.
Skin pores are a really, really fine detail which is both difficult to capture and easy to get in the way of your BRDF. Most of the times, it's better to have the pore details as self-occlusion in the specular map than in the normalmaps (which being a derived measure will require higher resolution to capture the same amount of detail).
Also, you either have to model skin for a given viewing distance, or you'd better consider that at all but the closest distances the pores will/should fade in your miplevels and their scattering effect should be modeled by varying the specular exponent (broadening it with the distance, you can use ddx and ddy to estimate the filtering width, our use cLEAN), there is no way to tune the specular with a constant exponent that will yield both the right amount of detail up-close and the broad sheen you see from a distance.
Also, you either have to model skin for a given viewing distance, or you'd better consider that at all but the closest distances the pores will/should fade in your miplevels and their scattering effect should be modeled by varying the specular exponent (broadening it with the distance, you can use ddx and ddy to estimate the filtering width, our use cLEAN), there is no way to tune the specular with a constant exponent that will yield both the right amount of detail up-close and the broad sheen you see from a distance.
Peach fuzz on dry skin is another geometrical source of specular detail perturbation, you might want to model it by adding noise at the grazing angles if you need that amount of detail.
3- Bad volume
 |
| Fight Night Champion When it works is one of the few games that captures volume right... |
Skin texture luminosity is often quite uniform, and in caucasians quite light too, making skin shading more a matter of volume and shape than texture. Three elements here are important: specular, ambient and normals.
Normals are tricky because skin is often... skinned. Conventional bone skinning yields bad geometric normals. I've already blogged about that so I won't repeat myself here.
Normals are tricky because skin is often... skinned. Conventional bone skinning yields bad geometric normals. I've already blogged about that so I won't repeat myself here.
Ambient is "easy" but really important, a simple contant won't do the job, it just flattens everything, even a simple hemispherical (top/bottom, a lerp between two colors based on the geometric normal y component, boils down to a single madd) is way better.
Observe and understand. Look at your references, understand what's important, understand your visual errors and their sources. The "correct hack" really depends on the context, if your environment is quite fixed, like in a sport or racing game you can model ambient with different components: a sky layer which does not depend on the position plus a ground reflection that maybe fades on the model based on the distance to the ground and so on...
Ambient occlusion is fundamental, ideally you'd want to have some directionality in the occlusion: bent normals, or encoding the occlusion in some base. Really "occlusion" is fundamental for volume. Each light component (ambient, diffuse, specular etc) should be occluded in a reasonable way.
Again, there are many ways, technically to achieve that, I won't delve into any detail because it depends on the context, you might want to agument SSAO to encode directional information (as I already said), or precompute occlusion at the vertices (SH or similar), cast rays in screenspace (i.e. screenspace reflection occlusion is easy, especially at the all-important fresnel angles, for rim occlusion) or even simpler hacks.
The important message here is that volume requires occlusion, look at your rendering and understand your visual defects, look for light leaks, compare with real-world references and acquired data and craft your own solution!
Sometimes the actual technical answer is dead simple (an example - one of the improvements on Champion over Round 4 was a "downgrade", we went from VSM to simple PCF filtering because VSM even if on average are "nicer" were not able to capture the very important occlusions on the face due to precision issues, the nose shadows and the eye-sockets. Going back to PCF gave us "worse" shadows but way better faces!)
Observe and understand. Look at your references, understand what's important, understand your visual errors and their sources. The "correct hack" really depends on the context, if your environment is quite fixed, like in a sport or racing game you can model ambient with different components: a sky layer which does not depend on the position plus a ground reflection that maybe fades on the model based on the distance to the ground and so on...
 | |
| Shadows, AO, Ambient model... All coming together nicely |
Again, there are many ways, technically to achieve that, I won't delve into any detail because it depends on the context, you might want to agument SSAO to encode directional information (as I already said), or precompute occlusion at the vertices (SH or similar), cast rays in screenspace (i.e. screenspace reflection occlusion is easy, especially at the all-important fresnel angles, for rim occlusion) or even simpler hacks.
The important message here is that volume requires occlusion, look at your rendering and understand your visual defects, look for light leaks, compare with real-world references and acquired data and craft your own solution!
Sometimes the actual technical answer is dead simple (an example - one of the improvements on Champion over Round 4 was a "downgrade", we went from VSM to simple PCF filtering because VSM even if on average are "nicer" were not able to capture the very important occlusions on the face due to precision issues, the nose shadows and the eye-sockets. Going back to PCF gave us "worse" shadows but way better faces!)
 |
| Phong, modified to behave better with Schlick. No lousy round highlights! |
Achieving proper specular on skin is probably one of the trickiest parts as unfortunately, right now in realtime rendering we have either the choice of employing nice material models on simple analytic lights. It's well-known that the Kelemen/Szirmay-Kalos model fits the skin acquired data very well and if you can afford it it's probably the way to go, especially if you lighting is not very complex (i.e. outdoor, harshly lit scenes).
Unfortunately, in many contexts we want to use some image-based lighting approach (baking reflection cubemaps), and that restricts the BRDF filtering we can employ pretty much to Phong, and straight Phong is really, really bad on skin.
Again, it's important here to observe and understand what is important, what qualities we want to model or hack. The specular sheen, other than affecting the saturation and the skin detail, as we discussed already, is important for the perception of the shape. Human vision can't really distinguish the effects of the various lights in the specular, we can't relate well the scene lighting with the specular hightlights shapes.
What we do with the specular is to understand the surface shape and material, so what's important is more to model the general shape of the specular, than the link between the specular and the actual scene lights.
That's why we can often ignore accurate specular occlusion (and just modulate all the specular light with our single shadowmap and omnidirectional ambient occlusion... or even "worse" multiply some of the diffuse product into phong, not ideal, but decent) and we can often disregard accurate light positions and use reflection cubemaps. And that's why what's really important is the shape of the specular sheen, and you get quite some latitude in hacking that in, as we can't really "see" these hacks as long as the final shape behaves "well".
What phong get really wrong is the highlight shape, which is too uniformly circular, and the lack of frensel. You can "break" the highlights using an exponent map (which should always be present, but it's especially important with Phong), adding fresnel is not easy as Phong simply does not reflect enough light at the grazing angles, and thus even multiplying with a decent approximation, like Schlick's, does not usually yield great results, it's better to use fresnel to (also) drive other parameters in the Phong model instead, either bending the reflection normal, or lowering the exponent.
Conclusions
I probably should have included a fourth horror "bad behavior", as most CG humans (all? in videogames) really lack the proper "fleshiness" in terms of animation and behaviour, but that's a harder problem to tackle which is probably outside the realm of realtime animation, at least for most games where you have many characters on screen and you can't afford having 300 bones in each plus some soft-body system and some UV relaxation and so on...
Really what I wanted to point out are some simple things that every game should do, aspects which can most of the times be fixed to a decent degree with really cheap hacks, but do require an understanding of how the underlying physics work and how our perception works to then focus on what really matters visually.
Then the actual techniques really depend on your project, for those of you that really want to delve into some actual tech, an example of a possible solution for skin rendering is here.
Addendum: Hair, Eyes, Teeth, Ears
Hair is a huge pain, on this generation we're stuck with hair cards. Cards don't sort well, are hard to shadow and hard to shade. My usual suggestion is to avoid spending too much time on it, it's not really worth it.
Shading usually employs Kajiya-Kay for specular (card will require a tangent shifting map to avoid having uniform highlights on a card, which are ok only for very straight hair), some wrap lighting for diffuse and a way of getting good rim.
For sorting, the usual solution is to go multipass, first doing alpha testing with z-writes, and then alpha blending with z-testing. You can somewhat pre-sort the triangles from in to out, some games do the alpha blending pass in two phases, first the backface cards then the frontfacing ones. It's decent, but defects will still be there, even sorting per triangle, if you can, doesn't solve them all as cards often intersect. I honestly think a better solution is to just alpha test and rely on alpha to coverage, if you can. This is what Fight Night Champion did by the way (Round 4 did a three-pass alpha blending).
Eyes are way, way more important. Shading is not that hard, specular/reflection is done with a cubemap usually, diffuse would need SSS but again some wrap is good, on Fight Night we computed a second set of normals to match the diffuse lighting with the skin around the eyes, probably a bit overkill. Behavior is hard, and getting it right is crucial! On the shading side, one trick I used is to disable the mipmaps, and the bilinear filtering on the reflection cube, to get aliasing far away that in turns create some shimmering in the small highlights. But most of it is in the animation, and some procedural techniques are to be employed.
Ears have lots of translucency, again, it can be faked easily, I won't lose my sleep there. For teeth, the important thing is to use the right occlusion and avoid them to become too bright, shadows from the mouth and lips onto the teeth are very hard to cast so again, you'll have to do some faking.
Bottom line, for each of these elements there are some easy hacks, but it's important to consider them. Once you understand what each of them needs and what physical things are important to model for each (ear: translucency, eyes: SSS and behavior, teeth: occlusion etc...), the actual techniques are "easy" or can be easily faked to a decent standard.
Conclusions
I probably should have included a fourth horror "bad behavior", as most CG humans (all? in videogames) really lack the proper "fleshiness" in terms of animation and behaviour, but that's a harder problem to tackle which is probably outside the realm of realtime animation, at least for most games where you have many characters on screen and you can't afford having 300 bones in each plus some soft-body system and some UV relaxation and so on...
Really what I wanted to point out are some simple things that every game should do, aspects which can most of the times be fixed to a decent degree with really cheap hacks, but do require an understanding of how the underlying physics work and how our perception works to then focus on what really matters visually.
Then the actual techniques really depend on your project, for those of you that really want to delve into some actual tech, an example of a possible solution for skin rendering is here.
Addendum: Hair, Eyes, Teeth, Ears
Hair is a huge pain, on this generation we're stuck with hair cards. Cards don't sort well, are hard to shadow and hard to shade. My usual suggestion is to avoid spending too much time on it, it's not really worth it.
Shading usually employs Kajiya-Kay for specular (card will require a tangent shifting map to avoid having uniform highlights on a card, which are ok only for very straight hair), some wrap lighting for diffuse and a way of getting good rim.
For sorting, the usual solution is to go multipass, first doing alpha testing with z-writes, and then alpha blending with z-testing. You can somewhat pre-sort the triangles from in to out, some games do the alpha blending pass in two phases, first the backface cards then the frontfacing ones. It's decent, but defects will still be there, even sorting per triangle, if you can, doesn't solve them all as cards often intersect. I honestly think a better solution is to just alpha test and rely on alpha to coverage, if you can. This is what Fight Night Champion did by the way (Round 4 did a three-pass alpha blending).
Eyes are way, way more important. Shading is not that hard, specular/reflection is done with a cubemap usually, diffuse would need SSS but again some wrap is good, on Fight Night we computed a second set of normals to match the diffuse lighting with the skin around the eyes, probably a bit overkill. Behavior is hard, and getting it right is crucial! On the shading side, one trick I used is to disable the mipmaps, and the bilinear filtering on the reflection cube, to get aliasing far away that in turns create some shimmering in the small highlights. But most of it is in the animation, and some procedural techniques are to be employed.
Ears have lots of translucency, again, it can be faked easily, I won't lose my sleep there. For teeth, the important thing is to use the right occlusion and avoid them to become too bright, shadows from the mouth and lips onto the teeth are very hard to cast so again, you'll have to do some faking.
Bottom line, for each of these elements there are some easy hacks, but it's important to consider them. Once you understand what each of them needs and what physical things are important to model for each (ear: translucency, eyes: SSS and behavior, teeth: occlusion etc...), the actual techniques are "easy" or can be easily faked to a decent standard.
28 November, 2011
Google reader share
So, some of you have noticed that my google reader share is dead (not updated anymore). That's not me being lazy, but google being a bit evil and trying to shove google+ down our throats by killing the old facilities instead of integrating plus into the existing stuff. So, there is no reader share anymore, and my iPad newsreader (reeder), which was responsible of most of the share posts, does not support google+ yet. Stay tuned.
25 November, 2011
Photoshop scripting - Cleartype for images
 |
| Left: bilinear, Right: bilinear with "cleartype" note- the effect is configured for a "landscape" RGB pattern LCD |
I always wanted to learn how to script Photoshop (what I learned is that it's a pain and the documentation sucks...), so yesterday I started googling and created a little script to emulate cleartype on images. Here is the source (it assumes that a rgb image is open in PS):
// 3x "cleartype" shrink script
var doc = app.activeDocument;
var docWidth = doc.width.as("px");
var docHeight = doc.height.as("px");
doc.flatten();
// let's go linear RGB
doc.bitsPerChannel = BitsPerChannelType.SIXTEEN;
doc.changeMode(ChangeMode.RGB);
// now that's a bit tricky... we have to go through an action, which has binary data... which I'm not sure it will be cross-platform
// it works on Photoshop CS3 on Win7...
function cTID(s) { return app.charIDToTypeID(s); };
function sTID(s) { return app.stringIDToTypeID(s); };
var desc6 = new ActionDescriptor();
var ref5 = new ActionReference();
ref5.putEnumerated( cTID('Dcmn'), cTID('Ordn'), cTID('Trgt') );
desc6.putReference( cTID('null'), ref5 );
desc6.putData( cTID('T '), String.fromCharCode( 0, 0, 1, 236, 65, 68, 66, 69, 2, 16, 0, 0, 109, 110, 116, 114, 82, 71, 66, 32, 88, 89, 90, 32, 7, 219, 0, 10, 0, 22, 0, 19,
0, 25, 0, 58, 97, 99, 115, 112, 65, 80, 80, 76, 0, 0, 0, 0, 110, 111, 110, 101, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 1, 0, 0, 246, 214, 0, 1, 0, 0, 0, 0, 211, 44, 65, 68, 66, 69, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 9, 99, 112, 114, 116, 0, 0, 0, 240, 0, 0, 0, 50, 100, 101, 115, 99, 0, 0, 1, 36, 0, 0, 0, 101, 119, 116, 112, 116,
0, 0, 1, 140, 0, 0, 0, 20, 114, 88, 89, 90, 0, 0, 1, 160, 0, 0, 0, 20, 103, 88, 89, 90, 0, 0, 1, 180, 0, 0, 0, 20,
98, 88, 89, 90, 0, 0, 1, 200, 0, 0, 0, 20, 114, 84, 82, 67, 0, 0, 1, 220, 0, 0, 0, 14, 103, 84, 82, 67, 0, 0, 1, 220,
0, 0, 0, 14, 98, 84, 82, 67, 0, 0, 1, 220, 0, 0, 0, 14, 116, 101, 120, 116, 0, 0, 0, 0, 67, 111, 112, 121, 114, 105, 103, 104,
116, 32, 50, 48, 49, 49, 32, 65, 100, 111, 98, 101, 32, 83, 121, 115, 116, 101, 109, 115, 32, 73, 110, 99, 111, 114, 112, 111, 114, 97, 116, 101,
100, 0, 0, 0, 100, 101, 115, 99, 0, 0, 0, 0, 0, 0, 0, 11, 67, 117, 115, 116, 111, 109, 32, 82, 71, 66, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 88, 89, 90, 32, 0, 0, 0, 0, 0, 0, 235, 194, 0, 1, 0, 0, 0, 1, 65, 50,
88, 89, 90, 32, 0, 0, 0, 0, 0, 0, 97, 15, 0, 0, 36, 77, 255, 255, 255, 232, 88, 89, 90, 32, 0, 0, 0, 0, 0, 0, 103, 37,
0, 0, 220, 208, 0, 0, 5, 29, 88, 89, 90, 32, 0, 0, 0, 0, 0, 0, 46, 162, 255, 255, 254, 227, 0, 0, 206, 39, 99, 117, 114, 118,
0, 0, 0, 0, 0, 0, 0, 1, 2, 0, 0, 0 ) );
desc6.putEnumerated( cTID('Inte'), cTID('Inte'), cTID('Clrm') );
desc6.putBoolean( cTID('MpBl'), true );
desc6.putBoolean( cTID('Dthr'), false );
desc6.putInteger( cTID('sdwM'), 2 );
executeAction( sTID('convertToProfile'), desc6, DialogModes.NO );
doc.backgroundLayer.applyGaussianBlur(0.75); // limit the frequency a bit to avoid too many fringes
doc.resizeImage(
UnitValue(docWidth, "px"),
UnitValue(docHeight / 3,"px"),
null,
ResampleMethod.BILINEAR // To-do: box filter (mosaic + nearest)
);
var unitValue = UnitValue(1, "px");
// RGB pattern, note that the nearest resize will take the center pixel, that's why red shifts by one and not zero
var redLayer = doc.backgroundLayer.duplicate();
redLayer.applyOffset(unitValue, 0, OffsetUndefinedAreas.WRAPAROUND);
var greenLayer = doc.backgroundLayer.duplicate();
greenLayer.applyOffset(-unitValue, 0, OffsetUndefinedAreas.WRAPAROUND);
var blueLayer = doc.backgroundLayer.duplicate();
blueLayer.applyOffset(-unitValue*2, 0, OffsetUndefinedAreas.WRAPAROUND);
doc.resizeImage( // Resize to "select" the RGB columns in the various layers
UnitValue(docWidth / 3, "px"),
UnitValue(docHeight / 3,"px"),
null,
ResampleMethod.NEARESTNEIGHBOR
);
//var col = new SolidColor(); col.rgb.hexValue = "FF0000"; redLayer.photoFilter(col, 100, false);
redLayer.mixChannels ([[100,0,0,0],[0,0,0,0],[0,0,0,0]], false);
greenLayer.mixChannels ([[0,0,0,0],[0,100,0,0],[0,0,0,0]], false);
blueLayer.mixChannels ([[0,0,0,0],[0,0,0,0],[0,0,100,0]], false);
redLayer.blendMode = BlendMode.LINEARDODGE; // add!
greenLayer.blendMode = BlendMode.LINEARDODGE; // add!
// blue is the base layer
doc.flatten();
// let's go to 8bit sRGB
doc.convertProfile ("sRGB IEC61966-2.1", Intent.PERCEPTUAL, true, true);
doc.bitsPerChannel = BitsPerChannelType.EIGHT;
08 November, 2011
Silvio Berlusconi
This is obviously going to be off-topic with the rest of the blog... If you landed here for the first time, this is a rendering related blog and this article is an exception to the rule.
As the Europe's and Italy's financial crisis deepens, news crossed the wire today that prime minister Berlusconi vowed to resign. I see many people asking around the world how it was possible that this happens only now, how did Berlusconi manage to be in power for seventeen years even after countless scandals and accusations. While in general I think it's not surprising, and the eight years of Bush administration could be served as an example, I'd like to try to explain what's peculiar about Italy's situation (of course, in my point of view).
Also, of course, I will make generalizations in the following. In no way I want to express that this is applies to everyone and everything, that should be pretty clear.
Survivalists
One way or another, we keep going on. This is by far what I believe to be the deepest of our problems. We don't care much about our society, we avert our eyes and keep going on, everyone trying to find a hole in which to live their lives.
We are masters in bending (if there is a profit to be made) or ignoring laws. Even our image outside the country is that of creative, chaotic individuals (at best), known for being obsessed about family and our own small individualities.
We are not socialists not liberals, we are just driven towards what can get us a gain tomorrow morning. Mind you, to a degree this happens everywhere, but it's not a defining quality of a population quite as it happens in Italy, where is deeply buried everywhere, from how people live their lives to how companies make business.
Even our economy, made mostly of small or family owned companies, with our comparatively large private savings and our huge public debt, is a testimony of this mentality.
I can't tell why this is the case, we are a young republic and unification was not a smooth deal, but we don't believe in society. Berlusconi is the embodiment of all this, and I don't know how much he was just "born this way" or how much he knowingly acts to please, to be popular, but he is certainly great in leveraging such sentiments. His political message, either explicitly or implicitly has always been "vote for me and I'll let you live your lives without control", "you don't need to be responsible for your actions", "I was successful, don't ask me how, you want to be like me, I won't ask you how"...
Berlusconi was never a left or right wing politician, he is obsessed about communism and certainly sees Italy's leftists as pure evil, but his actions are not the ones of a liberal. Among other things, he's remembered for saying openly at an entrepreneurs' convention that companies without off-shore operations were not led smartly, and that evading taxes (one of Italy's chief problems) was morally sound in a country like ours (in which taxes are too high). He didn't them proceed to lower the taxes and impose strict controls to have everyone paying the right amount, or to incentive competition and freedom of enterprise.
Berlusconi was never a left or right wing politician, he is obsessed about communism and certainly sees Italy's leftists as pure evil, but his actions are not the ones of a liberal. Among other things, he's remembered for saying openly at an entrepreneurs' convention that companies without off-shore operations were not led smartly, and that evading taxes (one of Italy's chief problems) was morally sound in a country like ours (in which taxes are too high). He didn't them proceed to lower the taxes and impose strict controls to have everyone paying the right amount, or to incentive competition and freedom of enterprise.
In the first months of his government he proceeded in the opposite direction, abrogating liberalization laws that were passed by the previous government, loosening controls over financial transactions and not reducing a single tax. Not touching established interests, not reducing bureaucracy, but just allowing people to just screw each other more freely.
Shameless
Berlusconi is probably not the worst individual in Italy's history. Corruption and misgovernment were always there and can be even tracked to the same underlying sentiment. The first republic created a massive debt because political parties were quite literally buying votes by flushing enormous amounts of public money into all kinds of public ventures, creating hundreds of thousands of "fake" jobs, public employers who were pretty much useless. But it came down onto his knees when it was found that politicians also used public money to fund their own parties. There was corruption, but there is was still a sense of shame.
Berlusconi took inspiration from this and pushed it one step further, he was proud of his tricks, every trial he escaped by passing laws in his own favor, every lie and joke he said made him look "smarter", more successful. It's not that the vast majority of Italians do not know he was a thief, that's also why many of his voters were even shy of saying so, especially if singled out.
It's that deep down, they knew, but they admired his skill, they wanted to live the same dream, to take the easy way and just not have to care. That's also why for years even after all the sex scandals he managed to keep a mostly Catholic country under control. That's why he still now has a huge following...
Media and Opposition
These two aspects were also important and I'm sure there is a lot more to be said, but I think they don't contribute to explain the Berlusconi phenomenon quite as understanding how much deeply he connects with some Italian sentiments.
Berlusconi owns most of Italy's media and he is renown to be a great communicator. He lies, but his lies are so constant and so fiercely defended by so many, that they slowly become truths. Words slowly lose their meaning and the public becomes divided into factions who do not reason but just mindlessly cheer for one or the other party.
Again, that's partially a "quality" of Italians, being more emotional than rational, being hot headed and profoundly divided. But he managed to exploit that incredibly well.
His power does not extend over only media of course, most if not everyone in his party is strongly tied to him, he choose men with little political past and respectability of their own, people who depended on him to be elected. Berlusconi IS his party, and everyone sings the song he sings. And he made pretty clear that was the way from the start, his party always had direct references to him in the logo, in the hymns, everywhere. Everyone laughs at his jokes. Everyone follows the same rules, tells the sames words, uses the same dialectic tricks. I'm not sure if it's imitation or doctrine but it's powerful.
On the other hand the opposition is fragmented and largely seen as made of intellectuals and professional politicians who do not have any connection with the people. (which to a degree can be even very true). They were always bad communicators so it was easy for Berlusconi to play them, routinely saying that there was no better alternative than him, that the left-wing was made of communists that had no real plan other than raising the taxes (even if financial pressure increases or decreases have not really been strongly linked to any particular government). Furthermore they showed no cohesion, being unable to claim even the huge victories they sometimes achieved (Italy was able to enter the Euro as one of the founding parters due to the work of a left-wing government for example) and not being able to look past their divisions.
Subscribe to:
Posts (Atom)