The beautiful programmer said: Modeling is not difficult at all! Making a 3D model, this step is the most difficult

I was browsing Twitter when I had nothing to do, and found that a great master had developed a modeling artifact. Draw a floor plan on the pixel grid, and then set the relevant parameters to generate a 3D hammer model. How wonderful...

In fact, Windows 10 also has a similar fool-like modeling software called "Paint 3D" , it can help you build a 3D model in seconds. For example, we click on the dog head in the right column, and then draw the outline and size we need on the panel, and a cute 3D dog will appear.

In addition to one-click fool modeling, if you are good at mathematics, you can also build a decent 3D model using basic mathematical formulas.

VR modeling, which has become quite popular recently, can even be used for virtual modeling directly using only a handle.

As long as you enter the 3D pit, you will find that modeling can lead to Rome. The real technical difficulty in the 3D industry at this stage is actually model processing and secondary development and application.

To draw this conclusion, we must start with the mainstream modeling methods in the industry. There are now four mainstream modeling methods in the 3D industry, namely polygonal modeling, surface modeling, voxel modeling and reverse modeling. Each of them has its own advantages and disadvantages, and also corresponds to its own model processing difficulties.

There is a saying in the industry that "you can model anything with polygons." A polygon is a surface formed by a closed path surrounded by multiple edges. As long as you use enough detail, you can create any surface. This is also the reason why polygonal modeling has become a mainstream modeling method. There will be no restrictions on making any model.

However, polygonal modeling also has shortcomings. It is just an empty shell. Although it can construct the surface of an object, it cannot reflect the inner structure of the real object, and there is still a big gap between it and the real object.

Moreover, it is easy for modellers to be too meticulous about details, resulting in too many model faces, so that later baking, rendering, and UV display of the model will almost lead to collapse. Secondary development can The operability has also become very low, resulting in an embarrassing situation where the model is completed but cannot be applied well.

Polygon modeling uses straight lines to construct a plane, while surface modeling is a modeling method that uses curves to form a surface. It is very suitable for Create smooth objects. For example, digital products, cars and some character details look very smooth.

However, the shortcomings of this modeling method are also obvious. Firstly, it is more troublesome, and secondly, it is difficult to accurately parameterize. Therefore, this modeling method is generally used for visual display and is used to generate videos or pictures.

At the same time, it has relatively high requirements on the beauty and clarity of the textures. When the model has a large number of high-quality textures, the amount of model data will increase, which is its inevitable shortcoming. .

Voxels are actually three-dimensional pixel blocks. The most typical example of voxel modeling is the well-known game "Minecraft" , each object is made up of stacked hexahedral pixel blocks.

In the model built in this way, each hexahedron has its own volume, unlike polygonal modeling which only has an empty shell.

However, voxel modeling has a very obvious shortcoming: too much memory. It is made up of cubes, and of course it needs to occupy the memory of the cube, so this is also the reason why the memory of general voxel games is very large, and it may not be able to be carried on ordinary computer equipment.

The previous modeling methods can be modeled without physical objects and based on ideas, while reverse modeling is to first have the physical objects and then perform modeling for them. Modeling. Reverse modeling includes photo reverse modeling, 3D scanning reverse modeling and a series of related technologies. The VR modeling mentioned at the beginning uses the reverse modeling method.

The oblique photography currently promoted by the country also belongs to reverse modeling. It shoots and scans the real scene in all directions to create a model.

The models generated by reverse modeling usually have a high number of faces and a very large number of textures, which results in a huge size of the entire model. It is very difficult to even do the most basic display, let alone two developed.

Based on the inherent shortcomings of the above modeling methods, we must reprocess the model to meet the different needs of different scenarios for model accuracy, memory, format, etc.

The company needs a 3D model of a car.

For the automobile manufacturing industry, he has very high requirements for the accuracy of the model, and even needs to be accurate, so that it can be actually put into production.

If this same car is used in a game, it will require higher texture quality, better fluency and smaller data volume to support the screen.

At this stage, there are generally two methods for processing models: one is manual and the other is machine algorithm. Manual operations are complicated and time-consuming. It is very likely that I wanted to process a larger model, but only reduced a few megabytes in a long time. In the end, the accuracy was lost, and the gain outweighed the gain.

As long as the machine algorithm is technically advanced, it can achieve twice the result with half the effort and even free up both hands.

Moreover, after lightweighting, it can not only ensure the accuracy of the model as much as possible.

It can also optimize surface reduction and UV display to achieve the best effect suitable for application scenarios.

So if you have problems with model processing, be sure to try lightweight processing first, maybe the problem will be solved.