MDN 8/1992 in 1992

Image processing

The attraction of manipulating 3D graphic software may be in the act of creating an object in an empty dark world and giving it light. Nobuhiro Shibayama, who was obsessed with this charm, brought advertising products into the 3D world, and Naoyuki Kato created a silent beauty in the 3D world. From their work, it seems that the imagination at the time of production is directly transmitted. Now, let's take a look at the back side of production that requires precision and patience.

1. Shooting feeling in a sterile room

Incorporate what actually exists into the 3D world

For me, who has always liked visuals that express "light and shadow" to some extent, Shade II, which is "software that creates light and shadow from scratch," is the software that I have dreamed of for many years. Taking as an example the magazine advertisement of Hattori Seiko's watch "ALBA", which was established as an advertising expression by incorporating real products into the computer world without gravity and air from recent work, 3D graphics I want to introduce the world.

​

(Nobuhiro Shibayama / Art Director)

​

Photographs are often used for the visuals of advertising expressions. Photographs are in the direction of creating advertisements using the power (presence) that "the thing certainly existed there". Of course, you can't take a picture of future events or the image you have in your head. However, I feel that 3D CG can create images of the future and the present progressive tense in my head.

3D graphics (including CG in general) have been mainly used to express fictitious objects in a fictional world. However, in the current 3D graphics, even at the level of a personal computer, it is at a stage where it is possible to realistically capture what is "not" as if it were "yes". If the mockup is complete, it may be possible to create an image of the product that will be completed.

• Digital texture

However, even though it is real, CG at this stage has a unique texture that is clearly different from photographs. Digital texture. If this is an advertisement, it will be a merit and a weak point. However, when this texture, which is somewhat slippery, matches the product, it produces an unprecedented effect.

In this Seiko watch, the client's wish is "I want to give an electric tone to the finish, not a general photographic expression or video expression. A virtual reality taste, specifically the tone seen on the monitor. I want to put it on a magazine advertisement. " The situation was also based on the premise of 3D graphics rather than photography, "I want to float the product in a mysterious time, a mysterious space", and aimed at the effect of 3D Kurafix, so I compared it. I was able to communicate smoothly.

• modeling

Let's see the actual work from here.

The request for advertisement production is made when the shooting product (final mockup) is completed. For the production, I received about 20 drawings and photography products.

As a concrete creation procedure, first, all the drawings are scanned by a scanner. Read this drawing (PICT) with "Get Template" on the modeling screen of ShadeII and trace it. Basically, it is read on a flat surface, and the thickness is added while switching the screen on the three-view drawing. As you can see, the shape model is on the wireframe.

Most of the drawings are drawn separately for each part. In addition, although each drawing is drawn at various scales, I entered it in a size that is easy to draw for the time being, and enlarged / reduced (resized) it when assembling. When enlarging / reducing, there are cases where numerical values are input by converting from the scale ratio, but in many cases, it is done according to the scale while looking at the photographed product. Gradually combine the parts and register them as a new set of parts.

Of course, I modeled only the part that can be seen on the surface, but the number of parts eventually exceeded 200 points.

• mapping

Mapping data will be created at the same time as modeling. Mapping is a method for giving colors and patterns to created objects. In 3D, you don't draw a picture directly on the model. There are two main methods for this mapping. Wrapping method and projection method. The wrapping method literally wraps an object in its picture. In the projection method, a picture prepared from one of the x-axis, y-axis, and z-axis is projected. Both are colored for the first time when rendered. Not only the inclination but also the shadows are calculated and a picture is drawn on the surface of the model. The data for mapping needs to be created by PICT separately from the modeling data.

The mapping here is the dial, the ring-shaped scale, the floating sphere, and the background. The dial part is created as a picture and projected and mapped. Character parts such as logos were printed in clear print, scanned with a scanner, and then traced with Illustrator. The scale of the dial was also created accurately with Illustrator.

[Tips]

The scales and numbers engraved on the 45 degree slope do not look like they are simply projected and mapped. It will be stretched vertically from the actual characters. Therefore, we calculated the size of the top and bottom on a slope of 45 degrees and prepared the mapping data with the characters multiplied by the flat body.

• Texture

No mapping is required for homogeneous models with no patterns. If you set the texture setting value and color for each modeling data and render it, it will be drawn as a textured object. The body and belt of the watch are textured and colored in this way. Texture settings are made in the Surface window. By default, all colors are white, so specify the color only where you need it. To express gold, BaseColor was set to black, Highlight was set to 70, Size was set to 46, Reflect was set to 20, and Metalic was set to 82, which was a brilliant yellow color. Similarly, for silver, BaseColor is black, Highlight is 56, Size is 56, Reflect is 24, and Metalic is 88, which remains white.

Once the modeling is complete, try rendering in scanline mode to get a feel for it.

• background

The background part uses the bump mapping method. Bump mapping is a way to represent natural bumps. Here, it is created using two sheets of PICT data. First, draw a picture of uneven color. Based on that, create 256 gradation gray data. The first color unevenness picture that draws two graphics on the scale with black lines there and bump mapping are performed using this gray picture, but at this time the dark part in the gray picture is set to rise It has been done.

[Tips]

In bump mapping, you have to specify the light source as well as rendering, but considering that the clock will float on this background later, it is necessary to align the direction with the plateau that hits the clock. Otherwise, it will be an unnatural world.

The uneven background created in this way is actually a flat image that only looks uneven. This is used as mapping data. A three-dimensional background can be created by placing a flat plate on the background and projecting an image that looks uneven.

• Preparing for rendering

Rendering is done in ray tracing mode when all the parts are ready, but at this time, it is necessary to keep the total memory size of System, application, modeling data, and mapping data within 8MB. This is because the RISC board dedicated to Shade II, which I was trying at the time of production, could only handle up to 8MB. For example, if you create 32MB of modeling data and mapping data, high-speed rendering using a RISC board will not be possible. For speedy work such as advertising, processing without a RISC board would be too time consuming and impractical. I have to obey the restrictions of RISC boards. (By the way, a 16MB RISC board is now available)

At first, the mapping data was created so huge that the software didn't work at all. The data size was too large. I removed INIT from the system folder one after another, made it a system folder containing only the system and finder, without KanjiTalk or even the clipboard, and resized the mapping data. In addition, I put together a lot of parts (organizing the browser), removed the invisible parts, and finally fixed it to a size that can be rendered on the RISC board.

There may be a method of partially rendering and synthesizing several images with Photoshop etc., but it is still more flexible to use the method of completing with one rendering for later corrections and corrections.

Rendering requires object placement, light source settings, and camera lens selection. This area is exactly the same as shooting in the studio. If you have a lot of experience in the studio, the setup will be good. Virtual camera lenses on the software are available from 9 mm to 720 mm. This time, I chose a wide-angle lens.

• rendering

It's finally time to start rendering. There are two modes for rendering with ray tracing. There is a mode that renders with the actual number of pixels and a mode that renders at 1/64 the size. Actually, I decided to finish it with 5,800 x 4,696 pixels, but before that, I will render it with 1/64 size, that is, 725 x 587 pixels. This, of course, is something you should do to get a feel for the finish, but it's a good idea to measure the time it took to render with a stopwatch. This is because 64 times the time taken is a measure of the time it takes to render at actual size.

In the case of this watch, I fully mounted four ShadeII dedicated boards "New Super" on MacintoshIIfx and stepped on that it would be completed in four days, but it actually took four and a half days.

The rendered image will be PICT data. Load this into Photoshop and add some manual work to complete it. Subtle changes in shape are made, lightning is added, and the shine of the metal parts is emphasized.

• Calibration output

I used an inkjet IRIS printer for output. The software is Photoshop. The paper is A3 size. Most of the works are output by IRIS in this size. It's not powerful enough to use for presentations without this size.

Here the client is checked. Since it is a precision equipment manufacturer, detailed checks of blueprints and pulling are included. The proofreading paper was filled with deficits. In addition to corrections to the shape, such as differences in the curve of the outer shape, there were enhancements to the glow appearing in the chronograph part (three small circles) inside the dial, and corrections to the camera angle. The demand for the shape part is particularly strict, and the limit that can be expressed by the current 3D graphics software is required. It took me two weeks to model so far, but it took another two weeks to finish the corrections I made here. What's more, it wasn't just two weeks, but two weeks including all night, which was honestly tough.

Operation time and rendering time will be a big problem to make 3D graphics at the personal computer level a "work".

It takes another three days to retouch and finish in Photoshop to make the final manuscript. If you are a cameraman, you basically have to provide the product and take responsibility only for shooting, but in order to create a 3D image, you do not have to take responsibility for the size of the needle and the shape and texture of hundreds of parts. It shouldn't be, and it's not the most workload.

• output

The final manuscript was delivered as a 4x5 color film. This is because Macintosh 4-color film is not yet accepted by any printing shop.

• Price of CG image and future development

It takes more than a month to create a 3D image.

Consider the production cost of shooting in the studio. The stylist's guarantee, the production cost of the object, the studio fee, etc. will be charged, and the price including the cameraman's shooting fee and the direction fee will be the shooting fee for the advertisement photo. It's hard to compare the case of CG with shooting, but in reality, the art director decides how to make the visual a photo, CG, or non-visual within the client's budget. I think it has been decided.

Currently, CG is still considered to be relatively expensive, but it is much cheaper than when I asked Lucasfilm Studio a long time ago. Furthermore, the division of labor of 3DCG will progress in the future. For example, a CG photographer, a lightman, and a renderer who are in charge of modeling but who are in charge of modeling, a background creator who creates a location scouting effect, and a CG photographer who integrates all of them to shoot, which plays a role like props and props. I think it will be an era when all the houses are available. If that happens, the form of work will change again, and images with new quality and power will appear, but the production costs corresponding to that will also have to be recalculated.

Projects are currently underway to capture shapes using 3D scanners, and these technologies will also present us with new forms of production.

• Advertising development

The 3DCG of this watch was used for hanging advertisements in cars and magazine advertisements (2P, 1P). Design and block copy production are also my jobs.

Thanks to you, not only was the client happy, but the sales of the product seemed to be good, and I am happy. It seems that junior and senior high school students cut out magazine advertisements and brought them to the store to buy this watch.

The image is also used as a billboard ad, which wasn't originally planned. Because the final manuscript is film. We are ready to deal with these. I had never thought that a 3D CG image would decorate a large billboard until a few years ago, and I am deeply moved that it was produced on a Macintosh.

Currently, I am making a prototype of animation using the animation function of Shade II. Since it is modeled without any omissions, it is relatively easy to create an animation. I would like to continue trying various possibilities to see what kind of possibilities will be created in the 3D world. 1992