HL2 and other new games have "vertex shading"... im not sure what is it excatly but i guess it's some new tech that creates dynamic shading on models.

Today the problem with skins is the light/shade on them, if you look good on the skins today, most of them have some kind of reflected light from the mid left side all along the model, it gives a realisitc feeling but if you suddenly go into a shaded place the lighting on the model stays the same, when it shouldn't.

so... are there any changes in that deparment ? will light/shade on models ever get to be dynamic ? and if so... how do you think the textures should look ?

Instead of skinning the lightning on the model, you only need to create the basic texture. All the small details like idents and bumps go to the bump map. For cooler lightning a specular and a gloss map can be added (specular map: brighter parts are shinier, gloss map: brighter parts have a sharper highlight). A self-illumination map might be useful for scifi stuff (brighter parts are less affected by lightning).

Wow, skinning has been nubbified...

Im not so sure... there are still tons of things to do.

Ripa - how do you define those things?

it kind of puts down skinners

That's rediculous...I prefer the shading system. More realistic lighting it better, I have no idea how you could think it 'puts down' skinners.

What he means is that skinning can easily be accomplished, all you need is a good texutre no need to worry about lighting and shades...

You still have to paint the specular, bump and gloss maps. So you are still doing the same stuff, only in a different way.

Guys, what the hell. Seriously. You're gonna tell me you'd rather spend hours trying to perfect shading in a skin, and have it look like the light is always hitting the model the same way in every enviroment, when you can spend less time using new technology to make it look a hundred times better with new methods?

Anti-progress, eh? "We feel like automobiles are a trend, that will not catch on."

Heh.

Originally posted by Skullz_Vandal
Anti-progress, eh? "We feel like automobiles are a trend, that will not catch on."

I still think that about Television, Cellular Phones, and pants... :D


...

"Ripa - how do you define those things?"

Most 3d renderers have had those for ages but only now they're becoming realtime. You can try them in 3dsmax for example. I forgot to mention the refraction and the reflection maps btw.

It's becoming more common to make high detailed models, like 300,000+ polys, then use these to generate high defined, bump, spec, displacement, normal maps, which you then add to a low poly model.

HL2 noobifies skinning ? ROTFL, you all have no idea how much additional work is added with newest game engines for all kinds of texture artists. New technology simplifies few things but add far more additional work.

PS. I remeber Gabe saying that HL2 buggy model has about 12.000 polys and that's a bit too "heavy". So don't jump to making 300.000 polys models for use in game. I suspect that HL2 will be capable of rendering not more that 1 milion polys per scene and that's a lot. However Stalker does 2 mlns :).

Originally posted by Craftos
ROTFL, you all have no idea


I don't think thats true.

I think you understood me wrong Craftos, you don't make 300.000 poly models to work ingame, you make them to generate your different maps, to then map onto a lower poly model.

Maybe if you checkout these links, you will get a better picture of where i'm coming from.

http://files.filefront.com/2586425

http://www.loadedinc.com/files.php?temp_view=files.php&game=&id=397

http://amber.rc.arizona.edu/lw/normalmaps.html

http://www.evasion3d.com/mw_gallery.html

http://farcryhq.com/screenshots.php?cat_id=4

And here is a pic, showing my first go at making a high poly model, to generate such maps from.

para hi preview (http://homepage.ntlworld.com/sonofroy/parahipreview.jpg)

It's the same way they made that rediculously hi-poly cave wall with the crystalline type look to it only take up like one polygon yet it still looked like it was made up of thousands.

http://hl2.daddeltreff.de/images/downloadbilder/editor_wall.jpg
This wall is ONE polygon.

yeh its cool

Originally posted by Felix
It's the same way they made that rediculously hi-poly cave wall with the crystalline type look to it only take up like one polygon yet it still looked like it was made up of thousands.

http://hl2.daddeltreff.de/images/downloadbilder/editor_wall.jpg
This wall is ONE polygon.

:eek:

It can't be a square with one polygon... ()_o
maybe 2 polys.

Faces are still triangles in our 3D world.

Normal maps store shaded data so you can make stuff look more 3d while it is really flat. Here is what the normal output looks like

http://members.shaw.ca/jimht03/images/face_norm.jpg

Alrite well going back to the topic of skinning...I'm skinning for 2 hl2 mods and I'm doing it the old fasion way by apllying the texture and adding details and lighting and scratches etc etc on the uv map, does this mean ive done it all wrong and that when put in-game the hl2 shading system is gonig to kick-in and make my skin job look like a piece of crap?

Im not sure bruta.

The big question is, whose job is that ? the modeler, the mesher ? or the poor skinner :p

youll find out when the hl2 sdk is out.

It should explain everything for the engine.

I think the skinner will be the biggest loser i think it would just take the feeling out of the good ole fashion skinning way.

I think the sdk is already released to mod developers

Theres a nice little Photoshop plugin on the nVidia dev-tools site for generating normal maps from greyscale height maps.

I'm not 100% up to speed on the new tricks but its pretty much an extension of what bump mapping does.

Think of it this way:

A triangle exists in 3D space and has a specific orientation and faces in some unique direction. If you imagine a "ray" pointing out from the surface of the triangle, perpendicular to the surface, that is the triangles "normal".

As a trangles is made up of 3 points it has the potential to have two sides but in most cases in 3D programming we assume it to have one and hence, that triangle will only have one normal.

Now the normals are what are used to define the lighting on the surface of the triangle. As it defines which way the triangle is facing the render can decide how much to illuminate that face.

Good example of this is if you import a 3D model and the normals are missing or screwed up - the whole model is black. Whats happened is as there are no normals the 3D engine doesnt know which way the triangles are facing and thus cant figure out how to light them.

Right, thats normals theory explaned so how do normal maps work?

Well if you think of two triangles put together making up a simple square. Basically you have two normals in the square which define the lighting. Because of this if that square moves in the 3D world relative to the light the whole surface area of each triangle will change as they move relative to the light source. Now in a low poly model this is what gives that drastic and rather "chunky" looking transition in lighting.

Now what a normal map does is define the lighting normals on a per-pixel basis.

Think of it as this. Take our two triangle square and slap a 512x512 texture on it. Now, with a normal map each pixel in it can now define its own lighting normal.

Suddenly we are not lighting two triangles, we are effectively effecting the lighting of each pixel of the texture on the polygon!

Hopefully by now you've twigged how powerful this is - for two triangles a 512x512 texture map can effectively define lighting for around 262144 individual points. So the basic geometry is 2 triangles but the normal map gives the illusion of the lighting of something containing something like 10,000 times that many polygons.

Normal maps can be created one of two ways :-

1) Take a greyscale image where dark area's represent low area's and lighter colours higher and translate that into a normal map. Effectivley its a sort of bumpmap-to-normal map convertion.

2) Actually make a version of your model with 200,000+ polys and render it out at normal maps and then map these onto the low poly version of the model. This is the technique Marzy was talking about.

Right, to the crux - how does this effect skinners? Well it comes down to how you approach it really.

First off, if you only have a low poly version of your model your going to have to create your own normal maps. The simplest way to do this would be to use the bump-to-normal technique where for you model you would first create a set of textures for the model which define all of the little details like dents, scratches, creases, etc. You would then create the actual texture which contains all the colour and base of the texture.

When the two of these are combined your based skin is lit using the normal map giving the illusion of a much higher poly count than the model actually has. Of course, this means your going to have to hand craft your own normal maps.

Second option is if, like Marzy has been doing, the original modeller has made a high poly version with extreme detail, almost like a lifelike sculpture of the model. These model would then be rendered out as normal maps for the low poly version. All the skinner has to do is then worry about the base skin.

There is still going to be a lot of skill needed for the skinner. Textures like skin are still going to need the little details like scars, spots and other little impurities and cloth will need to have a sense of texture and tone to look right.

Effectivley all the Normal map will do is affect the lighting on a per pixel level so whereas we would shade and tint our textures to give the illusion of lighting, the normal map will do that for us.

Well thats my take on it. If I've made any horrendous mistakes please correct me.

But on a side note, thank yourselves luck that we dont have to make texutre, normal, bump, specular and radiosity maps!

And on a side note the VTF texture format HL2 uses contains both the normal and texture data in one file. I'm assuming at this point that the normal and texture maps must be layed out in the same way (think UV Mapping) so if someone else is making your normal maps you'll need to make sure they produce them in a layour that suits skinning.

- Jed

http://www.3dm-mc.com/tutorials/maya/texturing/

This tutorial uses the following maps to create a detailed texture for a head in Maya.
Color Map
Bump Map
Spec Map
Reflectivity Map
Transparency Map
Translucence Map
Diffuse Map

"But on a side note, thank yourselves luck that we dont have to make texutre, normal, bump, specular and radiosity maps!"

I was under the impression HL2 had those too and with the programmable shaders you are able to create your own too.

Originally posted by BruTaLiz@tioN
I think the skinner will be the biggest loser i think it would just take the feeling out of the good ole fashion skinning way.


You need to realize that tech is moving on.

Originally posted by ripa
I was under the impression HL2 had those too and with the programmable shaders you are able to create your own too.

I believe so but to be honest I've been so busy with HL1 stuff I've not yet really had an opportunity to sit down and really get into HL2. Heck I havent run XSI more than about 4 times.

- Jed

Get baking..............lol

I know i understand tech is moving on for te better, im just raising one slight ethical issue but now im ore well informed i think things should be fine.
Going slightly off topic here, i saw a vid yesterday of CS moved to hl2 and it looked great, I hope they move dod to it too it would be great.

Doom 3 models only have about 3000 polys due to bump mapping.

Originally posted by SWAT
Doom 3 models only have about 3000 polys due to bump mapping.
No they are about 7000 poly. Or is that HL2?

But anyway, they use normal maps made from 3million+ poly models. Thats why they are soo nice.

Saying this is an 'ethical' question is rediculously overboard...it's a game.

:rolleyes: Yes it is a game, but i dont actualy play games, I skin for games as a personal hobby because i have an interest in computer graphics...

Well normal mapping and its merry band of brothers has been around for quite a while in the 3D modelling world. Its only just come about in games because hardwares only recently become available that can render these effects in realtime.

- Jed