In the present article you can find a theoretical material about artistic visualization. Now, we will talk about a creation of a picture, beginning from a modeling of the car. Very often the final result depends on the subject of rendering in itself.
You can find a guides on step-by-step modeling or detailed adjustment sequences at the internet. Here we will talk only about factors that really forms the result. And it is not the software. What is more: it doesn’t really matter what program you use, but it is really important how you did it. The best way is to understand the principle and easily master any software later.
The Morihei Ueshiba combined a different martial art to create the Aikido, so I will not talk about that is better to reach a goal. There is an examples from different programs in this article - in this way I can show things more detailed. Why not to use an eastern wisdom to achieve our goal?
In this lesson we will touch upon a key moments and we will consider them step-by-step. We will give an attention to adjusting of the visualizer in itself, because this topic brings a lot of questions.
The goal of this lesson is not to teach how to create the model right from the start with a further visualization, but to help in creation of such work - it is more appropriate among of heaps of different lessons. So, I wished to do something like a pony that will be useful at different stages of creative process.
Model Me Completely!
So, on which factors a quality visualization depends? First of all, it depends on a qualitative model. Here I don’t mean a superb detailed elaboration. The quality is a complex approach that you need to apply on every stage of work. That is to say, we need to realize a final result before we will start to create anything, and have clear technical task too. In my case, I formed that task by myself when decided to process a car without a particular fanaticism, but qualitative. So as not to be a shame to show the model close-up, even without a perfect processing. This is a very important moment of creation of any product - to make it all without a slapdash. Don’t rely that somebody will not see or will not understand something. If a model must be a simple, it must be a qualitative too.
So, you need to collect a references before you begin to create a model. The blueprints are rare and untrustworthy enough things - except you are working for the firm that provides you with perfectly precise blueprints. Also, it is wisely to develop your abilities of of a shape and volume perception. The designing is an artistic job, that requires not only the perseverance and carefulness, but the understanding of designed object too.
Topology and a good sense of shape are decisive here. The good topology requires an experience - it is always the way of tests and mistakes. When you repeat the whole process after the designer, you develop a right understanding of matter, but you need a complete course on design. My current goal is to make something like a crib that helps beginners to find their way in solving of different tasks. Therefore, here will be many of text.
There are a several approaches in design. I prefer initially to define a part that sets the perception of form (usually, it is a bonnet or a car’s front part with smooth transition of one part to another) and make this area completely, including the space for a small parts. It gives an understanding of proportions, form, and sets the basis for all edges and faces. Also, it lets to close a whole model more clearly and confidently. But here we need a precise eye and good understanding of our final goal.
The next method is more mercantile and lets to try a feel the shape modeling its from a box. It is easy - we create a box with a small number of faces; and we outline the basic shape, and then we smooth the box and we will begin the modeling in itself. Often it can be done by zbrush because it is easy to feel the shape during the modelling. The most important here - to catch the shape at the beginning, and to realize the lines and aspects that will create the car later.
The real samurai always heads to the final victory, he doesn’t leaves the unfinished area, even when someone doesn't see them. Better to simplify than don’t make. It is also very important to production: an attitude to own work often plays a bad joke with the doer when client wants to change the foreshortening - then something unfinished arises immediately.
Of course, there are a different situations, but I advice to make any work qualitatively, beginning from the stage of modeling selecting the difficult areas and spending more time to them. It will helps to avoid a problems with geometry at the visualization stage when a crooked reflection will occur suddenly, or a gross mismatch with the original will be detected.
For testing of reflections, there is a reason to set the material with light specular for a model - so during the rotation you will see a coupled polygons that need to be settled. How to make a such testing more advanced and more convenient, we will talk later.
So, there are a some moments that I describe as fundamental. If it possible, try to adhere to the following:
1. All the polygons on the frame must have 4 corners, never allow 5 and more. May be a triangles, if necessary - but only in certain places. The renderman users will clearly understand what I’m talking about. See the examples below.
2. The wireframe density. The wireframe must be even, and must have a density required for certain level of detailed elaboration. It is illiterately to make a superdense wireframe for a model with low detail level. It is more useful to solve all questions by minimal amount of polygons - it will give the ability to make a corrections, also it allow to control the whole topology better. As for me, I make the wireframe more dense only if I can’t work with too small details or when I have lack of polygons.
Drilling the holes by boolean operations is not the best choice if you have an opportunity to make it all by the hands. The equability of the wireframe is also important - it is a peculiar culture of modeling that prevents from hanging edges and poorly coupled polygons.
3.About the carefulness. There are some principles that gives a good tone to the whole work: the supporting ribs of model must never ends abruptly, and the common geometry always must to emphasize the main lines of shape. If a different parts have a common shape - their supporting ribs must be attached. It is very important. Whole wireframe must be uniform and smooth. The loose wireframe is better than dense wireframe.
4. There is no reason to make a sharp elements as more than a 5-ribs chamfer. Einstein said that we complicate things when using more than one soap for douche. That is the same case. The great thing is to make all without a fanaticism, but without detriment to quality. The wireframe must be enable to become more dense it without visible geometric distortions in case of necessity.
5. A small geometries, such as grills, nets or other repeating elements is not a real horror. It is enough to model a small area and duplicate it properly, and then connect a tops. This is a great benefit that will provide the ability of making close-ups. Also, with a proper topology always is possible to obtain a normal map - there a many clear tutorials on this topic at the internet.
The detail level. Details always supplement and enlivens a models. I seek not to a maximal amount of details but to provide a quality of each detail taken separately. Therefore, you need a good notion about the shape of these details, because not always you can to trace all needed small things by photo. In this case, there is a reason to watch this detail at another car model of that automotive brand - some details may be shared by different models. Also it is very useful to look at a plastic models - you can find a different solutions on simplification.
Actually, when we examine a process of modeling, we always are forced return to a conception of the style of modeling. Working with software - is a question of time, but about an understanding of theory - is not so.
Now, a little off-topic. Nicola Tesla invented an ideal variant. After the death of his brother, Tesla felt a great sorrow. Afterwards, he learned how to assemble all the new devices in his own imagination. So, when he ultimately created something, all devices worked properly right away.
This example shows the importance of creating of the model in imagination. When you know, in which way you will solve a different parts, you gain a time for producing of any project, also it will avoid a topological deadlocks at the model.
Not everyone can see the model at once, me too. Sometimes you can imagine a several parts, but later you may change you plans - for example, when you wish to increase/decrease the density of wireframe.
There is a good method that helps to feel all the niceties of car shape visually and to think over a further topology. I always advice to use this method at early stages of modeling.
First, you need to find a 3/4 (or close to that) photo of car. Then, you need to load this photo into Photoshop, and then outline a photo by a planned net - in this case you will decided a quest of topology at once. The only you must to know is the topology in itself. But this knowledge appears with experience. That method helps to create a wireframe preliminary, without an incorrect work of software or insufficient knowledge of a modeling toolkit.
I borrowed this method from the creators of characters, who after the billeting of the model in zbrush, turns on the Polypaint Mode and then begins to draw the future net over the object for making the re-topology over the ready picture..
That method helps to save time wasted on the model. Even if you draw the wireframe rough enough, you get an idea about topology in any case. The process will be go faster too.
Making the adjustment with a photo is also very useful. The meaning here in following: to set a car photo (3/4 photo is the best) in the Viewport as the back and make settings for the camera. These settings must be correspond to the settings of the camera that was used for making photo. This technique lets to build a right proportions more quickly.
An excellent modeller Timur Ozdoev tells how to make it with 3ds Max:
«The matching is manual. It all simple - look at the preferences of the photo and find a model of camera and a focal length. You can do it by right-clicking on the file (usually the camera info are located in the Details item). Next, find out the specifications on this camera at the internet and look at the matrix size. You need to set this value into the Film Gate cell at the VRayPhysicalCamera. And don’t forget to set the focal length at corresponding cell. Using Photoshop you can select at the photo the details that located at the same vertical level and at the same horizontal level, and then you can connect them by lines. It will let to adjust the model correctly at 3ds Max and work with it in the same way».
As for me, I prefer the Maya for my work. Although, it all makes differently in Maya, but that approach is right and significant due to its logic.
Also is very important to give an adequate names to the all objects. By these names you can easily to find them at the list. The ideal solution for this topic is the Outliner in Maya. Here is a simple logic: you need to accustom yourself to an order in all. The same is concerned to textures, materials, etc. Some people name the details with a prefix of a name of the material used for this detail - it helps when working with large stage.
Now, we pay attention to important moments of modeling that can help to build a wireframe.
When making a bevel, or when increasing the density, don’t leave empty the area that stretched from the edge. The wireframe must be smooth, descending gradually. That helps you to avoid a roughness and make the model more neat. All depends on conditions: usually the one separating edge between two hard areas is enough. It is especially important, if you have a bevel with more than two edges. In simple terms it means - «the harmony is needed to the modeling».
It is looking well at the picture below. Modellers often forget to divide the areas properly, and at the some smooth types the shape of the object distorts.
When you can avoid the triangles - you need to do so.
Another common mistake is an attempt to make a rounded areas by eye. For making the accurate shape, there is a reason to emulate a prism with number of faces not less than eight. 45-degree angles. Such examples you can encounter very often.
There are many of such things, but using the methods shown below, you can make the model better and also increase your productivity. There are a culture of modeling and all things in this case are interrelated.
Now I want to spare some time for the reflections testing that is necessary for detecting of problem areas at model. You can do it by several ways depending on software you are using. With 3ds Max, you can test the reflections with DirectX Shader.
First, you need to be sure that in Customize –> Preferences section the Direct3D is selected as the Display driver.
So, there are some points. First of them - the slot for fx and gfx files that located in the system directory with the 3ds Max. There are the DirectX shaders that are suitable for an excellent testing of reflections. I select the BumpReflect.cgfx map. To see the smooth reflection, you need to set the BumpHeight at 0. Next, we are looking for the envmap scroll - the files of the dds format can be loaded here. Beginning from the Directx7, this format supports a 3-dimensional images, including a cubical textures. There are a different convertors at the internet for converting hdri into dds.
There are many defects in this system - from a low productivity to a necessity of finding the maps in dds format or convert your hdri files into ddr format because there are not many maps in 3ds Max at the moment when this article was written. Also, the scheme of the testing is not comfortable - I want to push the one button to see how and where the reflection falls, and then I want get back to the modeling by pushing of one button.
Therefore, I will talk how to make a similar testing in Maya with hdri-maps and Hardware render 2.0.
First, you need to switch on viewport 2.0, which may be chosen from the Renderer item.
Next, we choose a desired settings, and then we proceed to the hdri. For a start, you need to set a material of Phong type for the car. On the reflected color we put the EnvBall with the inserted node for a loading of the texture.
Next, we just set the way to hdri and admire.
The structure of our material looks in this way:
Now we can to switch between the usual viewport and 2.0 without problems and performance losses. As you can see above, all of this is done easily. It is very important thing. Also, you need to remember that if you click the «7» button, you can see the pure hdri.
As for me, this is the most convenient way for testing of reflections. Further we will talk about the visualization in itself.
Let there be light!
When you have a ready model, you want to see it in the best light. We slightly touched the topic of modeling for not to loose a perception of work in complex. The quality of the model strongly affects on the final result, as all other things too. The approach in general has an importance. So, there are a lot of chances to obtain a nice picture if you have got a good model. We will use the Vray for the visualization.
First, we need to realize - what we want to show? And only after this we need to think - how we will make it? Exactly in that sequence. In this example we have the model of 1929 Duesenberg Model J Dual Cowl Phaeton.
This is an unique car, and it needs a though simple, but the ambience. For more exotics, I decided to show the car against the background of a metropolis of 30s - because the Duesenberg was an American luxury car brand. To outline the "luxury", I decided to show the car against the contrast of the old, broken asphalt and the dark city, covered by haze. The additional calculations with hdri maps was even made for creation of the shine similar to a Pin-Up gloss.
Of course, it is desired to solve this task by hdri, but in that case all the important will be lost. Therefore, I used the usual studio, where the light sources give the reflection from a planes, located above the car. For sharp backlight and clear glares, I set the direct light sources. You can see the whole history at the screenshot:
For the Vray Dome light, I established a VrayHDRI with a studio lightening, that was created preliminarily for filling of the dark places at the car. Also, the floor material was made reflective - for increasing of certain glares. To insert a car into ambiance better, the hdri later was changed to a sky cupola. I turned off the light sources that threw a too sharp glares. After all these corrections, I obtained an effect of old automotive advertising. This result inspired me the idea of noir colors of 30s.
When the lightening was set and the foreshortening was determined, I began to solve the next task: how to make a city and the broken asphalt. Many variants was tested; the most problems arose with the asphalt. Perhaps, I will start from creation of city.
I made the buildings by boxes - it was fairly simply. Then, I painted the textures for them, gathered all into one block. I already had these buildings, because I did made them before - for the scene on the mafia theme of 30s. Although, this work nor requires too many time.
Then I placed the buildings at the horizon considering the scale and calculated them with two separate passes - with textures and without textures. I cut them enough, because their bottom parts will be lost due to haze.
So, the background was made. Next, the most difficult - the asphalt. I examined a lot of variants in my mind, and decided to make it by hand using the Zbrush.
It was very necessary to create a several brushes and use them alternately during further work. I chose the followings textures of old and broken asphalt for my work:
They all were placed at alpha in zbrush, and then they were applied to the plain, considering the blur of borders to make an area seamless.
At previous stages I tried to make the cracks more narrowly or more widely, and make the edges more sharper or more blurred.
It is important that asphalt was not a texture, and its relief needs to be realistic. I obtained the desired result without using of HD geometry (although earlier it seemed as inevitable).
The Displacement should be good for a more heavy scene, but in my case the most properly was to make an asphalt in Zbrush and export it into 3ds Max, previously thin out the geometry by embedded Decimation Master plugin. This made a life a much more easily.
Now, lets talk a little about the materials. For this task, it was necessary to have a bit complicated materials. All others was made as was written above.
I did made the paint material with the Blend material, in which later was established two ones: the VrayMtl at the basis and the Shellac with two VrayMtl in the second slot.
All the same could be made with VrayBlendMtl, but I prefer exactly the way described above.
The rest materials was created just on the basis of VrayMtl or Blend
The glass material I always create in this way:
In my case, the chrome I can create with usual VrayMtl or with Blend.
For the headlights, I used a glass material but just with the texture on Bump.
Depending on the case, the falloff curve could be bended.
All the rest materials are a quiet simple, many of them was created just on the basis of VrayMtl, that not only made the life easy, but also made a calculation of the final picture more faster. I wrote more detailed about the materials in my lesson «The Rendering of the Car. An Approaches to a Creation of Studio. The Style»
About a rendering: there are some important moments if you want to obtain a clear picture without serious defects. First: there is no any standard option that can make a beautiful picture at once. Second: the rendering loves a balance - so, if you set the lightening subdivs to a maximum value without appropriate setting of the material subdivs, you decrease the process performance without significant quality improvement. Third: there are no any standard settings - all adjustments depends on concrete task.
For the good result, you need to increase the amount of subdivs for all materials, and then make the same with the light sources. You need more subdivs to the light sources that works more actively in the scene. 16-25 subdivs are enough for any light source.
Now, about the adjustment of rendering in itself. To make you work more convenient, I advice to adjust the dynamic memory limit in the Render Setup tab. Subtract about 2GB from the maximal amount of your memory and set the value as much as possible (about 2GB are used by Windows). Next, there is the V-Ray DMC Sampler scroll inside that tab. Here are two parameters - the Adaptive amount and the Noise threshold. Varying these settings can increase the quality of a final picture. The Adaptive amount controls the maximum amount of samples used for the calculation. From 1 to 0 means from full adaptation to absence of adaptation. The lower is noise threshold the less of noise will be in final and the longer will be the rendering. You can turn off the Time independent checkbox, because it is using during an animation. The Min samples controls the minimal amount of samples those should be created before using of the algorithm of early completion. The bigger Min samples value the longer is the rendering, but the better the result.
The Antialiasing scroll in the V-Ray adjustments also is important. To increase the quality, you need to set the Adaptive DMC as a sampler and add a zero in Clr setting. For big pictures, you may increase the maximum of subdivs from 8 to 16 (although 4 is quite enough). When you set the gamma more than 1 in the color mapping settings, don’t forget to turn on the Don’t Affect Colors checkbox - it can preserve the overexposure. The Catmull-Rom filter makes the picture more clear.
For the visualization of scene with Duesenberg, I not turned on the GI. But it may not to be unnecessary in different cases. These options help greatly increase the quality of render.
It is necessary to make a layer-by-layer rendering during the visualization. What to render - you can find at the Render Elements tab. It will make a life more easy during the processing of picture in the post.
These adjustments are enough for a qualitative picture. For better understanding of Vray rendering, there is a sense to read an excellent book “VRay: the Complete Guide”. Its author - Francesco Legrenzi is like a Leo Tolstoy in the world of 3D graphics.
Is There Life After the Rendering?
A finished work - it is not only a good render, but also a qualitative post. Confucius said: «The jewel cannot be polished without a friction». Exactly as in graphics - the model needs to be a good-rendered, and a good render needs to be properly presented. The post-processing is the polishing of our render.
During the post-processing we finalize our initial idea.
The best way - to make the rendering layer-by-layer. In this case, you always have an opportunity to re-render something separately without wasting time for the repeating calculation of whole picture.
Also, with the VrayLightMtl you should calculate a mask for different materials. It will give you the ability to adjust the contrast of separated materials - for example, for chrome.
Now all depends on the correctness of idea, on accents, and on rendering in itself. In my case, I did made a rendering of floor, and then I did made a rendering of the asphalt, and finally - I combined it with reflections. Better to make the shadow by separated pass - in case of necessity, you can make the shadow darker.
For the artificial shine that is similar to the old Pin-Up, you can make the over-rendering by separated hdri pass, and later place it upon the glares.
Now, much depends on contrast and color corrections that is named as cross-processing in this particular case.
The city was added initially, but I hided a buildings specially, because in such way they will not hinder the processing. Then a color correction and contrast questions were solved, I converted the layer with the sky into the Soft Light mode (in the Blending Options), so the contours of buildings became visible at background.
Actually, the scheme in itself is simple. Here you need to act depending on conditions - in the most convenient way in each particular case. I only demonstrated, how I do it. But, probably it is not the best way. It is always possible to find a more efficient and more qualitative way. Just not be afraid to search your own way.
Two cameras was established specially for the glares. For better visibility, I rendered them completely:
I can say, that the submission affects on much. Also, is not well to leave without attention the role of color - sometimes the color is more important than it seems.
I could show that Duesenberg car in different styles to perceive it differently. And, I feel a temptation to do it. Here I only tried to bring a little creative work to counterbalance a classic visas. As you can notice, I used the same renders that were made for a basic pictures.
For understanding of the importance of color in the work, I created some additional variants. Important to notice, how much a perception of a car changes, depending on changes in color:
Curiously enough, but the most difficult are the black-and-white variants. Usually, I don’t like to discard a colors completely, keeping some colors in certain areas. But sometimes is the sense to make all picture black-and-white - and see the differently.
In conclusion, I want to quote of Morihei Ueshiba: «In your practice you must hold in mind an interaction of sky and earth, of the water and fire, yin and yang».
It can be translated to the CG language as: «When performing of the work, you must hold in mind a whole process: from modeling, rig, animation, and texturing to visualization and post»
Creative success to anyone! And thank you for attention!