Şefki Ibrahim***角色制作流程(双语)中

自译 教程/经验 版权:禁止匿名转载;禁止商业使用;禁止个人使用。

1707 0 1 2020-04-27 17:28:26 举报

微信小编

地球 未知

4月,Sample and Hold团队扫描了一位名叫西尔维娜的女士。我的任务是获取此摄影测量数据并将其转换为高质量的digi-double
a169ab3083eefd0bdeeb439cf583311d-sz_748050.jpg

教程来自:英国伦敦3D艺术家Şefki Ibrahim

翻译:CG模型网-嘉儿


《THE SILVINA PROJECT》

《西尔维娜项目》


software

Maya、R3DS Wrap、Mari、ZBrush、Arnold、Photoshop、Mudbox



Eyes眼睛


f626624a5d7c7140ee01de23efd16c35-sz_246506.jpg


The front part of the sclera geometry is separated and unwrapped in a single UV tile.


The back of the sclera is positioned in the corner since its resolution is not a priority.


巩膜前部的几何图形被分离开来,并在单个UV中展开。


巩膜的背面位于角落,因为其分辨率不是优先事项。(虹膜前部贴图绘制**,背部几乎看不到所以**度低)


f626624a5d7c7140ee01de23efd16c35-sz_246506(1).jpg

Sclera Albedo巩膜反照率图


This process was also repeated for the iris.

All of the eye textures were created at 4K res.


这个过程也被重复到虹膜上。

所有的眼睛纹理都是在4K分辨率下创建的。


f626624a5d7c7140ee01de23efd16c35-sz_246506(2).jpg

lris Albedolris反照度


Screenshots from Maya viewport of the sclera, iris, pupil and caruncle geometry. Note, the circumference of the iris extends the size of the cormea to avoid gaps and shadowing.


巩膜、虹膜、瞳孔和角膜几何图形的Maya视口截图。注意,虹膜的周长会扩大皮质的大小,以避免缝隙和阴影。



When I had finalised the eyeball placement within ZBrush and matched its positioning to the scan, the eyes were imported into myMari scene. This way I could make sure that my projection was matching the reference in regards to how large the iris should be as wellas where the veins should sit in the opening of the eye.


The projection was made using the references provided for both the right and left sclera and iris geometries. It was important for me to aim for high realism in this piece, so having the asymmetrical eye texture maps were really important.


The rest of the sclera texture was hand-painted with a combination of procedural layers.


当我在ZBrush中最终确定了眼球的位置,并将其定位与扫描结果相匹配后,眼睛被导入到我的Mari场景中。

这样,我就可以确保我的projection与参考的虹膜应该有多大,以及静脉在眼球开口处的位置相匹配。


投影时,我使用了左右巩膜和虹膜的几何形状的参考资料。对我来说,这幅作品的目标是高逼真度,所以拥有不对称的眼睛纹理贴图非常重要。


巩膜纹理的其余部分则采用了手绘的方式,结合程序化的图层进行绘制。


f626624a5d7c7140ee01de23efd16c35-sz_246506(5).jpg


The texturing process is very similar to the tongue mentioned earlier.I used the diffuse map to derive bump maps by applying Photoshop filters. These were in turn used to create the specular, roughness and subsurface maps, with the addition of hand-painting and procedural texturing.


Some of the sclera texture maps are presented below.

The transmission map is used to create the cornea of the eye(a circular ramp can also be used within Maya).


纹理处理过程与前面提到的舌头非常相似,我使用diffuse map漫反射贴图,通过应用Photoshop滤镜推导出bump maps凹凸贴图。

这些继而用来创建specular高光、roughness粗糙度和subsurface maps亚表面贴图,再加上手绘和procedural texturing程序纹理贴图。


一些巩膜纹理图如下所示。

透射贴图用于创建眼睛的角膜(在Maya中也可以使用circular ramp圆形斜面)。


c634f43f23b68ac111fa472cb841f97b-sz_276573.jpg


Changes to the eyes have been made sincel

眼睛已经变得真实


c634f43f23b68ac111fa472cb841f97b-sz_276573(1).jpg


Like the teeth, it's important to have subsurface on the sclera and caruncle geometries in order to blend these components into the eyelids (and each other) more seamlessly. The randomwalk v2 SSS model was chosen for this after some testing.


It's important for the transmission blend to be smooth and have a high feathered edge, meaning that the blend between the sclera and iris is blurred. A sharp transition will create the CG eye look.


I tend to create a circular selection around the iris map in Photoshop to darken the outer edge, which can define the outline of the iris, but this should be very subtle and very much depends on the reference・


就像牙齿一样,巩膜和眼角膜上有subsurface是很重要的,这样才能让这些部件更完美地融入到眼皮(以及彼此之间)。

为此,经过一番测试,我们选择了Randomwalk v2 SSS模型。


重要的是,transmission blend传输融合要smooth平滑,边缘要有很高的feathered edge羽化度,也就是说巩膜和虹膜之间的融合是模糊的。

尖锐的过渡将创建CG眼睛外观。


我倾向于在Photoshop中在虹膜贴图周围创建一个圆形选区,将外缘变暗,这样可以定义出虹膜的轮廓,

但这应该是非常微妙的,而且在很大程度上取决于参考



c634f43f23b68ac111fa472cb841f97b-sz_276573(2).jpg


Maya viewport: The sclera and caruncle geometries fit cleanly into the model, making sure to avoid clipping and large gaps.


Maya视口。巩膜和角膜的几何图形干净利落地贴合在模型中,注意避免剪裁和大的缝隙。




5a031729a9c5fca21cbf3234996f0374-sz_85486.jpg


①:Sclera Specularity IOR=1.37

No coat used in order to avoid unwanted strong reflections in the iris.Subsurface radius colour set to a very desaturated peach tone.Caustics ON

①:巩膜镜面反射率IOR = 1.37

未使用coat涂层以避免虹膜上发生不必要的强烈反射。Subsurface次表面半径颜色设置为非常不饱和的桃红色调


②:Caruncle Includes a transmission mask at its edge to blend into the sclera.

Shaded very much like shiny skin (IOR set to 1.4).

②:Caruncle在其边缘包括透射遮罩,以融合到巩膜中。

阴影非常像光泽的皮肤(IOR设置为1.4)。


③:Pupil A plain dark grey colour is used.

It has some specularity with a roughness of 0.65 and IOR of 1.4.

It should react with the light in a subtle way and not be completely black.

③:瞳孔采用普通的深灰色。

它有一些specularity镜面反射,roughness粗糙度为0.65,IOR为1.4。

它应该与光线发生微妙的反应,而不是完全黑色。


④:Iris

Some SSS is used. Too much will eliminate bump effects. A very rough specular effect can be applied; none is used here.

④:虹膜

使用了一些SSS。太多会消除凹凸效果。可以使用非常粗糙的镜面效果,这里没有使用。


⑤:Meniscus

The waterline is a piece of geo created only when the eyes’shape and positioning are finalised. I usually create this using quad draw in Maya.

A water preset is applied for the material.

You can achieve better effects with a bump map applied (not added yet).

⑤:半月板

水线是一块只有当眼睛的形状和位置最终确定后才创建的地理位置。

我通常在Maya中使用quad draw创建这个。

将对材质应用水预设。

应用凹凸贴图(尚未添加)可以获得更好的效果。


d1c5a9fe28360c99e9d383b45c450475-sz_1105324.gif

Eyes Turntable

-GIF效果演示-



Skin Shading Texturing皮肤着色贴图


As mentioned in Part 1, the cross-polarised diffuse map was cleaned in Mari, removing hair and unwanted shadows picked up from the photogrammetry session.


I also worked in Photoshop to colour grade the diffuse map to better match the skin tone of the subject in the cross-polarised reference images. This took many iterations over the course of the look development process. Here is the final result:


正如在第一部分中提到的那样,我在Mari中清理了cross-polarised交叉极化的diffuse map漫反射图,去除了从摄影测量过程中拾取的毛发和不需要的阴影。


我还在Photoshop中对漫反射图进行调色,在交叉极化中更好地匹配受试者的肤色。在整个造型开发过程中,我经过了多次迭代。下面是最终的结果。


f6a45cedea6167e93d83cb08cd5484b2-sz_279762.jpg


When I finalised the look of the displacement map, I exported a normal map from ZBrush and converted it into a cavity map in Photoshop.


I then overlaid this cavity map over the diffuse and set it to a very low opacity to give the subtle hint of skin pores that would correlate with the skin detail.


当我最终确定置换贴图的外观时,我从ZBrush导出了一个法线贴图,并在Photoshop中将其转换为一个cavity map。


然后,我将这个cavity map覆盖在漫反射上,并将其设置为非常低的不透明度,以提供与皮肤细节相关的细微毛孔。


For this project, we wanted to go for a one-to-one likeness for the skin. Since the photogrammetry didn't provide me with a displacement map, / utilised the two colour maps (polarised and cross-polarised).


I subtracted the CP from the polarised image to reveal the specular only. This was made black and white with a levels and high-pass filter applied to essentially create a bump map・


在这个项目中,我们想为皮肤做一个1:1的相似度。由于摄影测量没有提供displacement map位移图,

所以我利用了两张彩色图((polarised偏光和cross-polarised交叉偏光)。


我从polarised image偏振图像中减去了CP,仅显示出specular镜面反射。进行了黑白处理,

并加上了levels调整图层和high-pass filter高通滤镜,基本上可以创建凹凸贴图・


f6a45cedea6167e93d83cb08cd5484b2-sz_279762(1).jpg


This bump map was then imported into ZBrush as an alpha and subsequently applied onto the surface of the model as a displacement map (on a layer).


Before applying, / subdivided the model from 7K polys to 29M. Which, after some testing, I felt would be sufficient to capture fine details.


然后将此凹凸贴图作为Alpha导入到ZBrush中,随后作为置换贴图(在图层上)应用到模型的表面。


在应用之前,/将模型从7K polys细分为29M。经过一些测试,我觉得这已经足够捕捉到**的细节了。


f6a45cedea6167e93d83cb08cd5484b2-sz_279762(2).jpg

subtract


I then proceeded to create new layers to edit the displacement map. For example, I wasn't happy with the quality of the cheek detail, so I smoothed it out and used ZBrush's surface noise feature to populate the cheek area with Texturing XYZ's tileable micro displacement cheek map (right).


然后,我继续创建新的图层来编辑置换贴图。

例如,我对脸颊细节的质量不满意,所以我将其smoothed平滑化,并使用ZBrush的surface noise表面噪波功能,

通过纹理化XYZ的可平铺微位移脸颊贴图填充了脸颊区域(右)。


f6a45cedea6167e93d83cb08cd5484b2-sz_279762(3).jpg


c8c389dd1df04a1db543644f919133b5-sz_1149621.gif

(Layers of skin detailing in ZBrush)



The displacement map was then exported out of ZBrush for Arnold at8K resolution using these settings.


I chose to export the map and base at subdivision 4 since/would be testing offline renders in Arnold.


The higher you export,the more exact the displacement will appear to the detail in your ZBrush model.


然后,使用这些设置将置换贴图导出到ZBrush中,以获得Arnold at8K分辨率。


我选择导出第4分区的地图和底图,因为/将在Arnold中测试脱机渲染。


导出得越高,ZBrush模型中的细节显示的位移就越精确。


089769ccfd613bd226b89f1a314b0e38-sz_184666.jpg


The displacement map was then brought into Nuke to split the displacement out into displacement and micro displacement.


Using a high-pass node set-up. I was able to capture the high-frequency details of the map.The original displacement was also blurred.This would enable me to add the maps together and have more control over how strong the micro details appear.I would also be able to paint a mask and reduce the micro-detail in certain areas,if lso wish.


然后将displacement map置换(位移)贴图导入Nuke中,将位移分成了位移和微位移。


使用high-pass node高通节点的设置。我能够捕捉到地图上的高频细节。


这样一来,我就可以把原来的位移也模糊化了,这样一来,我就可以把这些地图加在一起,对微观细节的出现有了更多的控制。


如果我愿意的话,我还可以画一个蒙版,减少某些区域的微观细节。


55f886ceaebe4998f5299aa94f177dfb-sz_20799.jpg

(注:原图清晰度就是这样看不清,PS里锐化了下,只能凑合了哟)


Displacement Split Process


089769ccfd613bd226b89f1a314b0e38-sz_184666(1).jpg



Close-up of the cheek and under-eye area. 

脸颊和眼下区域的特写。


089769ccfd613bd226b89f1a314b0e38-sz_184666(2).jpg


I used the same method for the bump map to create the specular map.subtracting the two maps reveals only the specularity of the skin,which I could clean up and grade to provide a specular map.


我对bump map凹凸贴图使用了相同的方法来创建specular map高光贴图。

减去这两个贴图只显示皮肤的高光性,我可以清理和分级以提供高光贴图。


6f3fe89cb94d17a2a3ed53bf3dc3695a-sz_186481.jpg


l inverted the specular map to create a base for the roughness.Again, this was colour graded along with some hand-painting to achieve the final result.


The remaining colour correction would be done inside of Arnold to get the right look for the skin.


A subsurface weight map was also painted.The map is essentially an off-white with the thinner areas of the head such as the nose,ears,eyelids and lips were painted more white.



我将镜面图反相,为粗糙度打下基础,再一次进行颜色分级,再加上一些手绘,最后的效果就出来了。


剩余的颜色校正将在阿诺德内部进行,以获得正确的皮肤外观。


此外,还绘制了一张次表层重量图,该图基本上是灰白色的,头部较薄的部位如鼻子、耳朵、眼皮、嘴唇等部位被涂成了白色。


6f3fe89cb94d17a2a3ed53bf3dc3695a-sz_186481(1).jpg

cead9ede77a036137332192d6a835c96-sz_3196173.gif

-GIF效果演示-

(Single Softbox Light Turntable 1)

(单柔光箱灯-旋转1)


l used Arnold's aiStandardSurface shader for the skin and subsequently plugged in my maps.In this instance, l am using the ACES colour space, which allows more control over the whites.Automatically, they are clamped at around 0. 82 meaning I can push this value to get brighter whites in the render.


This is the reason for neglecting the coat map in the skin shader, since l could get everything I needed from the specular section.


我为皮肤使用了Arnold的aiStandardSurface着色器,然后插入了我的maps贴图。

在这种情况下,我使用了ACES色彩空间,可以更好地控制白色。自动将它们固定在0.82左右,表示我可以按下此按钮值以在渲染中获得更亮的白色。


这就是我忽略了皮肤着色器中的coat map涂层贴图的原因,因为我可以从specular高光部分获得我所需的一切。


0d10f40b35f66f8a1bb7f6fb2bdeb3e9-sz_204665.jpg

-------------------------------------------------------------------------------

上图补充

The red channel of the texture map is plugged into the welght attribute for it to function property.


For physicaly accurate resutts, the lOR for the skin is loft at,1.4.


纹理贴图的红色通道被插入到welght属性中,以使其发挥作用。

为了获得物理上的精准还原,皮肤的LOR为1.4。


A generic peach colour is used for the radius.The ACES colour space makes it appear darker here.

The scale of my scane is in metres which is why the SSS scale is>1.The value should be 1/10 of this.


半径使用通用的桃红色.ACES颜色空间使其在此处显得更暗。

我的扫描仪比例尺以米为单位,这就是SSS比例尺> 1的原因。该值应为此值的1/10。


-------------------------------------------------------------------------------


Every texture map is connected to an aiColourCorrect node for me to grade the maps inside of the renderer and get realtime feedback as It weak the values.The final look of the textures are demonstrated below.Note,the difference between the maps on the previous page.


每个纹理贴图都连接到aiColourCorrect节点,以对渲染器内部的贴图进行分级,并获得实时反馈,因为它削弱了这些值。

纹理的最终外观如下所示。请注意,上一页中的贴图之间的差异


0d10f40b35f66f8a1bb7f6fb2bdeb3e9-sz_204665(1).jpg

5f55f9224d7eaace87343045e90fcdd7-sz_1205912.gif


-GIF效果演示-

(Point Light Lighting Animation)

(点光照明动画)


3a10b61acdb68d4100be7fd3f8f65313-sz_1768588.gif

-GIF效果演示-

(Single Softbox Light Turntable 2)

(单柔光箱灯-旋转2)


For the displacement, l used a PlusMinusAverage node to add both the micro displacement and base displacement maps together.


Each of these maps was hooked up to an aiRange node before being added together.


对于位移,我使用了一个PlusMinusAverage节点将微位移和基数位移图加在一起。


每张地图都被连接到一个aiRange节点上,然后再一起添加。


4f7a0a7138844492cd1d031377a0182f-sz_24809.jpg


This way, I could control the displacement intensity in the‘Output Max'attribute highlighted below.


这样,我可以在下面突出显示的“Output Max”属性中控制置换强度。


be573ec11110c3c4fe91f1a60ac0e22a-sz_17369.jpg


When working on the skin, I made sure to test the skin material under various lighting condtions.I had a handful of high-quality references comprising of a single and double softbox light, which was my main lighting setup.This way, I could attempt to match my renders with the photographs.


 But it's always useful to have a skydome light in your scene to do some HDRI tests of indoor and outdoor environments.


The model was imported into May a from ZBrush at subdivision level 4as was the displacement.For the purpose of these offline renders I can afford to export high-res geometry.This also allows me to maximise the level of detail the geometry displaces so it better resembles the sculpt inside of ZBrush.


在制作皮肤的时候,我确保在不同的光照条件下测试皮肤材料,我有一些***的参考资料包括单柔光箱和双柔光箱灯,这是我的主要照明设置。

这样,我可以尝试使我的渲染与照片匹配。


但是,在您的场景中使用天穹灯对室内和室外环境进行一些HDRI测试总是有用的。


该模型是从ZBrush细分级别4导入Maya的。为了这些脱机渲染的目的,我可以输出高分辨率的几何体。

这也允许我最大化几何体置换的细节级别,使其更像ZBrush内部的雕刻。


(Exterior and Interior HDRI's. Sourced from HDRI Haven.)


All renders are rendered out in P3-D60(ACES) colourspace. For the ACESworkflow, all of the texture maps are converted to 32-bit EXR files.


All weight maps are set to colourspace:Utility(RAW) whereas colour maps are set to ACES cg.­­


All texture maps in regards to the skin are 8192x 8192.

The final skin shader was mixed with another standard surface shader for the make-upon her upper eyelid. A weight map was painted of the upper eyelid and connected to the ‘Mix’ attribute allowing me to create this effect separately from the skin shader and better match the reference image.


Note, the eyebrows and eyelashes are the only grooms presents of ar since all facial grooms will be revealed in the next part. With that said, peach fuzz and facial hairs do enhance the look and feel of the skin, and overall realism. So while this represents the final skin material it does not represent the final look of the skin.


The render was colour matched to the reference image in Nuke.


On the right are the specular, wireframe and lighting passes.


所有渲染均以P3-D60(ACES)色彩空间渲染。对于ACESworkflow,所有纹理贴图都转换为32位EXR文件。

所有权重图均设置为colourspace:Utility(RAW),而颜色图均设置为ACES cg。


关于皮肤的所有纹理贴图均为8192x 8192。


将最终的皮肤着色器与另一个标准表面着色器混合在一起,以在她的上眼睑上进行化妆。

绘制了上眼睑的权重图,并将其连接到“ Mix”属性,使我可以与皮肤着色器分开创建此效果,并更好地匹配参考图像。


注意,眉毛和睫毛是现在唯一的毛发,因为所有的面部毛发都将在下一部分中展示。

也就是说,桃子绒毛和面部毛发确实增强了皮肤,整体真实感。因此,虽然这代表了最终的皮肤材质,但并不代表皮肤的最终外观。


渲染图的颜色与Nuke中的参考图片进行了匹配


右边是镜面反射、线框和灯光过程。


fc58055d0493720c0cf36236e5f0e3a0-sz_280877.jpg

0095d22a82398b67926bb54ca7b854e2-sz_383476.jpg

fb294766355e3a4fe53fda30ce0b7e49-sz_160122.jpg

f0a27f50b0e4ebc176820a1a17367e0b-sz_403581.jpg

(3 Point Light (with facial groom))




-END-




皮肤部分到这里就结束了。

下期内容强烈推荐!!!

内容为毛发,包含眉毛、脸部耳朵部位的桃子绒毛、以及头发!




-广而告之-



为了推送更多干货资源,我们增加了一个微信订阅号CGMODEL,有兴趣的可以先关注一波!未来经常会有干货哟~


翻译不当的地方,欢迎留言指导!


鸡腿送给你们~谢谢!


翻译的文章在版权上来说是不属于原创的, 因此我们没有做原创标注!转载请注明出处!





下期再见~❥(^_-)






标签 ******角色

声明:该资源由发布人:【微信小编】上传,点击查看作者主页其他资源。

CG模型网(cgmodel.com)内网友所发表的所有内容及言论仅代表其本人,并不反映任何CG模型网(cgmodel.com)之意见及观点。

推送到

首页推荐 编辑推荐

取消 通知作者
对此作品设置了隐私保护,禁止保存至本地。