Comparing photography and 3D rendering June 8, 2008

Filed under 3D, Photography by Juha | 0 comments

In some cases it’s obvious which produces more easily good looking images, 3D or photography. Take for instance portraits or food shots. There really is no contest at the moment: photography rules. Then there are all sorts images that would be harder to create using real world photography: It’s certainly possible to make a cut-away image of a car by first cutting the parts and then photographing, but I think in many cases 3D rendering would get the same image more easily. In this article I compare relative strengths and weaknesses of 3D and photography for still life studio photography.

Two images, one concept

Here I have two images. One is done using 3D animation and rendering tool called Autodesk Maya, the other is a real world photograph. The images are not identical, but similar enough for comparison. The concept here is that sometimes less is more and in other cases the opposite is true.

Sometimes less is more and other times the opposite is true Sometimes more is less and other times the opposite is true

Can you spot at this small size which one is 3D and which one is a photograph? You can also take a closer look by clicking the images. I didn’t intentionally try to conceal any details that help identifying the creation method, but I could have done that as well…

3D image creation process

In 3D the world is empty until someone makes the needed models. Modeling can sometimes be very simple or already existing models can be used. Other situations may require days or weeks to model all the needed objects. Most of the products are these days designed with 3D CAD tools and this way the models can be “almost free”. Commercial model libraries also have wide variety of objects available. In the case of my example image all I needed was a glass, capsule and some sort of plane to position the objects on. Modeling was in this case quick and easy phase.

Next, surfacing is done. Surfacing means giving correct surface properties (reflectivity, color, transparency, index of refraction, texture, …) to each object. Many common materials (like glass) are available ready made. Some materials can be harder to define and considerable time is needed for tweaking the look of the surface.

Scene setup and rendering are done next. In this phase objects, lights and cameras are positioned and then the scene is rendered to see the final look. Usually many iterations are needed, but each iteration takes commonly just minutes. Scene setup is mostly done using low resolution images and rendering just small parts of the whole image. When the scene is ready it will be rendered with high quality and high resolution settings to get the final image. Print size images (200×300mm, 300DPI) typically take couple of hours to render, but this varies.

Photographic image creation process

Photography looks simpler as a process, but there are hidden complications. Basically, objects are brought to studio or the camera is taken to location. Then lights, reflectors, gobos etc. are set. Iterations are performed by taking pictures and adjusting the scene.

Of course, things are not that simple in real world. Typically many iterations are needed to get the lights, camera and object positions and depth of field satisfactorily adjusted. Photography happens in real world and this can complicate things: How to physically position things? How to selectively exclude light from some parts of the scene? How to change the apparent surface properties? Where to find X meters by Y meters softbox?

Comparing photography and 3D

Strengths of 3D:

  • Things don’t need any support to stay anywhere you want (there is no gravity or wind…)
  • Lights and reflectors can be any size you want
  • Lights can affect just some of the objects
  • Lights cast shadows only if you want so
  • The scene will be available as long as you want (I have changed images years after they were originally created)
  • Materials can have different properties when lit by different lights
  • Lights can have any color and pattern easily
  • Depth of field can be infinite or you can use faster than f/1.0 lenses
  • You always have a perfect 0 – 10000 mm zoom with f/0.001 – f/1000 aperture. It also has build in perspective correction.
  • Your digital back has almost unlimited resolution

Weaknesses of 3D:

  • Objects need to be modeled and surfaced before use
  • Making things look good is not trivial in some cases (especially when inspected very closely)
  • Some surfaces are difficult to make look good (or realistic), these include skin, hair, fur, feathers, etc. surfaces found in nature
  • “Final” images are not available immediately
  • Everything tends to look too clean and perfect, small surface variation has to be intentionally added

Naturally the strengths and weaknesses are inverse for photography. Here are some specific cool and annoying things in each of the ways to create images:

With 3D it’s sometimes hard to exactly predict how the final image will look when rendered. This results in wasted time doing yet another iteration. With photography on the other hand it’s much easier to get feedback in real time with 100% “rendering” quality…

Positioning lights, gobos, reflectors, etc. is MUCH easier in 3D. Just leave the softbox floating in the air and it will happily stay there forever. Try that with a real world softbox!

Surface appearance can sometimes be pain to get right with 3D. On the other hand all the objects will be perfect and without finger prints, scratches, dust, etc. (unless you put some imperfections there, which is always an option!).

Modern 3D rendering software

Take a look at the gallery pages of some renderers to get a feel what is possible currently:

To get a feel what kind of models are available check for example www.3d02.com.

What about future?

I predict that in couple of years virtual photography will become as popular as real world photography for “still life”/”product shot” style image creation. There are only couple of things missing:

  • More low cost and high resolution render ready objects are needed
  • More ready to use surface materials are needed
  • More rendering speed is needed in high quality point-and-render systems

Hollywood movies already use 3D a lot more than people generally know about. One reason is that the alternative would be building an expensive set. Products are also rendered for commercials and still images commonly because 3D CAD objects can be used quite easily in most cases, just leaving the surfacing to do (some “hero” surfaces may need to be re-modeled).

Understanding how to use a rendering software doesn’t help much in creating good looking images. Knowing how to make good looking photographs is the key to good looking rendered images. This means that real world photographers will be able to switch to 3D as soon as the software tools don’t require that “3D guru certificate” anymore. Using the 3D tools is currently complicated because all kinds of shortcuts are needed to make things fast enough. In the near future there will be enough processing power in computers to handle the 3D more like real world and this will make things a lot easier.

I think it would be a good idea for real world photographers to start looking more closely what is developing in the 3D sector, especially if they are doing commercial product or still life art photography. It can be truly enjoyable to be able to just press a button to kill that reflection, to have unlimited array of huge softboxes and to be able to just leave the gobo hanging in the air :-)




Leave a reply (Check my Comment Policy)