JP2006127369A - Drawing device using fuzzy Z - Google Patents

Abstract

In a 3D graphics rendering process, polygons with curved surfaces and subdivisions are increasingly used to improve expressive power. These are processed by finely dividing them into polygons according to the size on the screen, etc., but this processing requires a lot of calculation, leading to a reduction in drawing processing performance and an increase in processing hardware scale. It was.
In a Z buffer method for performing a stamping process, a Z value representing a distance is not recorded as a definite one value but is recorded in an ambiguous range expressed by a maximum value and a minimum value. When another object is drawn, if the ambiguous Z range does not overlap, the front-rear determination is performed, and if the ranges overlap, the Z values of both are determined and determined. Since the calculation amount can be reduced by calculating in an ambiguous range rather than calculating the deterministic Z value, the calculation amount can be greatly reduced when the definite Z value calculation is small.
[Selection] Figure 1

Description

  The present invention relates to a block drawing apparatus that performs drawing processing of computer graphics.

  In most cases, a dedicated semiconductor integrated circuit (LSI) or a system LSI in which this apparatus is incorporated as a part of the system LSI is mounted. However, a drawing apparatus using an ambiguous Z as another means such as a transistor or a relay element. It is also possible to configure.

  A conventional drawing apparatus that performs drawing processing of computer graphics uses a buffer that records a depth value from a viewpoint called a Z buffer in addition to a color buffer, and draws objects (dots, lines, (Polygon, curved surface, etc.) Hidden surface processing by judging whether each pixel is visible in front of the contents drawn up to now or in the back, and updating the color buffer and Z buffer only when it is visible It is carried out.

  Various ideas have been made to increase the speed. For example, a group in which some pixels are collected is introduced, and the maximum value and the minimum value of the Z value in the group are recorded. If this information is examined before comparing the Z value for each pixel, it may not be necessary to refer to the true Z buffer if the pixel is further than any pixel in the group. Since this information can be reduced if the size of the group is appropriately selected, even if the entire Z buffer cannot be held inside the LSI that can be accessed at high speed, the system can be accessed by having only this information inside the LSI and accessing it at high speed. The processing speed can be improved.

Also, a technique using the maximum value and the minimum value of the Z value for each pixel is shown in Patent Document 1 below. However, in Patent Document 1, what is recorded is a result of accurately calculating the maximum value and the minimum value of the Z value in each pixel, and the amount of calculation is not reduced by using an ambiguous Z value. The accurately calculated Z value range information in the pixel is used to determine whether anti-aliasing processing is necessary based on whether the Z range is continuous on the boundary line between polygons.
JP-A-9-326043

  In three-dimensional graphics processing, in order to improve expressiveness, polygons with curved surface data and subdivisions are increasingly used in addition to the conventional simple polygon representation. These are often drawn by being divided into a necessary amount of polygons so that they appear smooth on the screen.

  This algorithm is very effective because it is relatively simple and requires only a small amount of original data. However, the larger the number of divisions, the greater the amount of calculation required, and there are many cases where sufficient performance cannot be expected with the current drawing LSI or the like.

  As the number of polygons is increased, the shape becomes smoother, and light source calculations are performed at a number of vertex positions, so that changes in illuminance can be accurately reflected. However, since the amount of calculation becomes enormous, the calculation time will be insufficient if some mitigation measures are not taken.

  Whether the portion corresponding to the outline of the shape data to be displayed (a collection of one or more drawing objects) is a rough polygonal line or a smooth curve greatly affects the appearance. Further, when the shape data and the shape data intersect, whether the intersection line is a rough broken line or a smooth curve greatly affects the appearance. However, it is not necessary to divide polygons so finely in other parts, and the amount of calculation should be reduced by rough division.

  If there is only a rough division, the amount of vertices for which light source calculation has been performed properly will be reduced, and the accuracy of illuminance calculation will be slightly worse. However, since the human eye is relatively insensitive to this difference, problems often do not occur.

  The present invention calculates the distance from the correct viewpoint by dividing the polygon finely only when the shape data and shape data cross, but if not, rough division is sufficient, and the amount of calculation is greatly reduced. It is an object.

  Note that the method of finely dividing the outline portion is not touched in the present invention, and therefore needs to be devised separately.

  In order to solve the above-described conventional problem, the drawing apparatus using the ambiguous Z of the present invention has a buffer capable of storing an ambiguous Z value (a range in which a depth value can be taken) for each pixel. As a mounting method of the ambiguous Z value, it may be mounted by a pair of maximum value and minimum value, a pair of minimum value and range width, or a different method. In the following description, it is assumed that the package is implemented as a pair of maximum and minimum values.

  The ID is also recorded for each pixel. This ID is an index to information for making the range of depth values accurate as necessary. Specifically, the number of the drawing object may be used, or only the information for accurately calculating the depth value may be separately managed and the number indicating it may be used. In the following explanation, it is not essential to use any method, so it is not specified.

  In the case of polygons with curved patches and subdivisions, the computational complexity can be greatly reduced if they are not divided finely. Also, the rough range of the Z value can be calculated relatively easily with a bounding box or the like.

  Therefore, when drawing a picture for the first time, it is not divided so finely, and only the ambiguous Z value is recorded.

  When drawing with this configuration, if the range of the ambiguous Z value of the content already written and the range of the ambiguous Z value of the content to be written do not overlap, the context can be determined by that alone. . However, if the ranges overlap, it is necessary to determine the context by determining the Z value.

  It is possible to calculate an accurate depth value for each pixel by leaving information about the object to be drawn. Since the ID is recorded for the contents already written, the accurate Z value of the contents already written is recalculated from the information. Needless to say, since an accurate Z value can be calculated, the context can be determined.

  According to the drawing apparatus using the ambiguous Z of the present invention, when the ranges of the ambiguous Z do not overlap, only the ambiguous value is calculated, and the amount of calculation is greatly reduced. Therefore, the drawing process can be speeded up even with a drawing LSI having relatively few hardware resources.

  Embodiments of the present invention will be described below with reference to the drawings.

(Embodiment 1)
FIG. 2 is a conceptual diagram showing an algorithm of a drawing apparatus using a conventional Z buffer method. A state in which a drawing object 202 is drawn on the screen 201 is shown. In the following description, the drawing object 202 is assumed to be a Bezier curved surface patch. However, other types of curved surface data, polygon data that can be subdivided as necessary in the subdivision processing, or LOD format shape data are finely divided as necessary. The format that can be made high quality is essentially the same.

  In the conventional Z buffer method, the drawing object 202 is subdivided as necessary, and is drawn by dividing it into polygons. In addition to the calculation for the division, it is necessary to perform drawing processing for each polygon generated by the division, so the calculation amount is enormous. It is necessary to record color values and depth information as information 203 for each pixel and refer to them as necessary.

  FIG. 1 is a conceptual diagram showing an algorithm of a drawing apparatus using an ambiguous Z according to Embodiment 1 of the present invention. A state in which a drawing object 102 is drawn on the screen 101 is shown. Unlike the conventional Z buffer method, division is not performed, the range of the Z value that the drawing object 102 can take is calculated, and the color value, the ambiguous Z value, and the ID are recorded as information 103 for each pixel, Refer as necessary. Since the drawing object 102 is not subdivided, the amount of calculation can be considerably reduced.

  In this description, the Z value range that the drawing object 102 can take is accurately calculated. However, this may be slightly larger. It is possible to make rough calculations without making calculations accurate. Further, the fuzzy Z values recorded in the information 103 for each pixel may all be the same, or may be recorded by narrowing down a little for each pixel if a little higher accuracy is desired. The narrower the range of the ambiguous Z value, the more likely it will be possible to perform the front-rear determination with only the ambiguous Z value. However, an additional calculation is required to narrow the range, so an appropriate balance is achieved. There is a need.

  Further, when the range of the ambiguous Z value disappears and only a single value can be taken, it becomes the same as the conventional Z buffer method. In this case, since it is not necessary to calculate the depth value (Z value) more accurately, the ID is not referred to. In the case of an object such as a simple polygon, even if the Z value is accurately calculated, there is a case where there is no burden on the hardware. In such a case, it is only necessary to calculate accurately and operate without a range.

  FIG. 3 shows information for each pixel in the first embodiment of the present invention. It is composed of a color value 301, an ambiguous Z value 302, and an ID 305. In this example, the ambiguous Z value 302 is composed of a maximum value 303 in the depth range and a minimum value 304 in the depth range. When the Z value is accurately calculated, the maximum value 303 of the depth range and the minimum value 304 of the depth range are exactly the same value.

  FIG. 4 is a flowchart showing the front-rear determination processing algorithm for each pixel according to the first embodiment of the present invention. Here, it is assumed that if the Z value is small, the distance from the viewpoint is short (in front).

  Although not shown in the flowchart of FIG. 4, as an actual problem, it is difficult to accurately calculate only the depth value of a specific pixel, and it is often necessary to calculate the entire depth value of the entire object or a part thereof. . In such a case, the exact Z value calculated should be updated for all relevant pixels.

(Embodiment 2)
In the first embodiment, when the comparison between the ambiguous Z values could not be performed, the accurate Z value was calculated and compared for both the object to be written and the already written pixels. In the second embodiment, when the comparison between the ambiguous Z values cannot be made, the Z value of the object to be written is accurately calculated first, and the ambiguous Z value of the pixel already written is calculated. The only difference is that the front and rear judgments are different.

  FIG. 5 is a flowchart showing the back-and-forth determination algorithm for each pixel.

(Embodiment 3)
Unlike the second embodiment, if the front-rear determination cannot be made by comparison between the ambiguous Z values, the front-rear determination is attempted by accurately calculating only the ambiguous Z of the already written pixels.

(Embodiment 4)
Unlike the second and third embodiments, if it is not possible to determine before and after the comparison between the ambiguous Z values, select either the object to be written by some method or the pixel already written. Calculate the correct Z value and try to determine before and after.

  As for which one to choose, it is conceivable to choose the one with the larger range of random numbers or ambiguous Z values.

(Embodiment 5)
In the first embodiment and the second embodiment, the ambiguous Z value is converted into an accurate Z value as necessary, and the forward / backward determination is performed. However, the ambiguous Z value can be obtained even if the accurate value is not calculated. If the range can be reduced, it may be possible to determine before and after. In that case, there is a high possibility that the calculation amount can be reduced rather than calculating an accurate value.

  The block drawing apparatus according to the present invention is not only effective as a three-dimensional computer graphics process but also useful as a two-dimensional processing method. Therefore, it may be used in the fields of home appliances, game machines, personal computers and the like including these functions.

The conceptual diagram which shows the algorithm of the drawing apparatus using the ambiguous Z in Embodiment 1 of this invention Conceptual illustration of the prior art Z-buffer method Explanatory drawing of the information per pixel of the drawing apparatus using the ambiguous Z in Embodiment 1 of this invention Explanatory drawing of the back-and-front determination algorithm for every pixel of the drawing apparatus using the ambiguous Z in Embodiment 1 of this invention Explanatory drawing of the back-and-front determination algorithm for every pixel of the drawing apparatus using the ambiguous Z in Embodiment 2 of this invention

Explanation of symbols

101 Screen 102 Drawing Object 103 Information per Screen Pixel 201 Screen 202 Drawing Object 203 Information per Screen Pixel 301 Color Value 302 Ambiguous Z Value 303 Maximum Depth Range Value 304 Minimum Depth Range Value 305 ID

Claims (10)

3D graphics rendering device that records and uses color, ambiguous Z value (range that depth value can take) and ID (index to information for accurate range of depth value if necessary) for each pixel When determining whether each pixel of a drawing object (point, line, polygon, curved surface, etc.) is in front of the already drawn pixels, and if it is in front ,
The range of the ambiguous Z value of each pixel of the drawing object is compared with the range of the ambiguous Z value of the already drawn pixels, and if the ranges do not overlap, the context is determined as it is.
If the ranges overlap,
An accurate depth value of each pixel of the drawing object and an accurate depth value of the already drawn pixel are calculated from the ID information to determine the front-rear relationship of the pixels, and each pixel of the drawing object is in front. A drawing apparatus that updates the already drawn pixels with the values of the respective pixels of the drawing object only when there are. The drawing apparatus according to claim 1, wherein when the ranges of the ambiguous Z values are overlapped, first, only an accurate depth value of a pixel to be written is calculated, and an attempt is made to determine whether the context can be determined. 2. The drawing apparatus according to claim 1, wherein when the ranges of the ambiguous Z values overlap, first, only an accurate depth value of the writing result so far is calculated, and an attempt is made to determine whether the context can be determined. 2. The method according to claim 1, wherein when the ranges of the ambiguous Z values overlap, first, only an accurate depth value of either a pixel to be written or a previous writing result is calculated, and an attempt is made to determine whether the context can be determined. Drawing device. When the range of the ambiguous Z value overlaps, as a step before calculating the accurate depth value, the range estimation is refined so as to narrow the range of the ambiguous Z value of the pixel to be written and the previous writing result, The drawing apparatus according to claim 1, wherein an attempt is made to determine whether the context can be determined. An integrated circuit comprising the drawing device according to claim 1. An integrated circuit comprising the drawing device according to claim 2. An integrated circuit comprising the drawing device according to claim 3. An integrated circuit comprising the drawing device according to claim 4. An integrated circuit comprising the drawing device according to claim 5.

Images