Case study: image or video output?

What Is the smart choice? Is it to create directly a video file from a 3D software. Or is it to generate image frames? Ultimately, should we consider all the possible range of graphics features, and corrections, that may arise to enhance an animation, there is no need for such a question: Image compositing allows a great fluency in the workflow and leave all options on the table, at any degree of the work.

What are the issues?

THE CONTEXT:

Even with gigabytes of memory and plenty disk space, frame after frame, some daunting recurring tasks await the system. The complexity of a 3D scene, especially if the rendering options are set toward a "fine" result, will soon overwhelm the best configuration. In practice, this threshold is reached soon with any decent configurations, even while processing reasonable 3D scenes. Every 3D user has experienced that "wall", called, "well, this is not do-able!". There are 3 major reasons to hit that wall:

ISSUES...

It just take some much time to calculate just one single frame, and so there is no way to produce a reasonable amount of images. Moreover, there are often, if not always, corrections to be applied, should the overall rendering succeeded, and this means more time consuming issues.
.

The software crashes after a number of frames being "video-output": 100, 500, 1000. It depends on the configuration but there is a (harsh) limit when it comes to output video files.
.

The video file was finally created, but there is nothing I can reasonably do with it: Meaning, refining and doing the post production stuff proves to be a hassle due to the size of the file. Modifying a single frame amid hundreds or thousands proves to be hazardous. So, should all the visual effects be computed at rendering time? Then we are back to 1 and 2. Moreover, "all" these effects just can not be embedded at rendering time.

OPTIONS:

"It just take some much time.."
The 3D Software is there to apply its 3D expertise. It is the tool that will generate the basic frames. There is not much we can do at this level. However, modern software, such as Poser 2008, generate incredible results, largely compatible with the production of thousands of high quality frames. It will take long hours, possibly dozens of hours to collect the frames, but computers can work while we do something else, at night for instance. In all it is not a big deal. The only strategic choice to be made here concerns the form of the output: choose a video output and be heading for deadlocks or produce frames and leave all creative gates wide open.

"The software crashes after a number of frames being 'video-output' "
This is a nightmare: exceptions brutally stopping the rendering process, images with objects loosing their texture, uncertainty when pressing the render button. What to do? Rendering multiple small video files to be assembled later will only complicate further the next steps. There is one single way to avoid all this: Produce frames instead of one single video file. Advantages are numerous: The system, even modest, can handle smoothly the production of thousands of frame without problem. This is the rule and it applies all the time. Even if the rendering should fail, for some rare reason (the electricity shut down ?) the frames that were produced are still valid. No big deal. Choosing the option of frame by frame is just the perfect answer to issue #2. Because you want to know that the computer is working OK while you do what you got to do.

"The video file was finally created, but there is nothing I can reasonably do with it"
Now this is the good part. Once the image base exists, there are dozens of ways to assemble them into a video file. Here is a short list:

a)	In some cases, with a small amount of frames, it is possible to generate a fairly long video, just playing with the possible combinations of frames.. When correctly used, the viewer may not even notice the re-using of frames. This is a trick that may simplify the initial 3D rendering. As usual with computers, a well thought conception may help a lot!
b)	Any graphical effect (PhotoShop, Flash, text, you name it..) can be added to any frame, easily, with no hassle. Straightforward. The final animation may be super loaded with all kinds of timely details.
c)	Transparency effect, superimposition of layers of images may generate additional frames both in number and in quality.
d)	When times finally comes to produce the video file, numerous speeds are available for testing. Different frame rates: slow motion, high frame rate, may be choosen in a single video output. The video output does not depend on a complex 3D rendering process, but on a simple compilation of images, easily repetable as needed.
e)	Once the video is compiled, if there is something to correct, there is no problem: the image base is there to be modified the easy way. It is well known that the impression you will a few hours or days later will be different.

A casual example:

It is good practice at rendering time to set invisible 3D parts that are not supposed to be in the final scene. This gains a lot of time. But in this case, thanks to the length of the animation, this detail went unnoticed. If the video file was produced, directly, (in this case 5 frames on 900 total, are concerned), What can I do: re-process 5 frames and do some video manipulations: uneasy and time consuming.

But if the images are available, maybe just a graphic patch, with a minimal 3D process, may do the trick, and compiling the video from frames is an easy matter.

In all cases, correcting, one way or the other, and recompiling from the image base is less trouble , a gain of time and precision. It enclines to push the envelop toward a great result.

Fine tuning

The frame rate, the speed of the animation, determines the final impression of the viewer. The final impression will arise only after the final rendering viewing. If the choice was to output a video file directly, all options are closed: total reprocess is required, or complicated work around. But when working out of an image base, it is just casual tuning.

CONCLUSION

This case study only exposes the very basics of 3D video generation. There are many "positive" aspects of image compositing that are not discussed here. For this reason, this case study exposes the strategic issues and uncovers "unpleasant aspects". However, these are crucial points to be considered very seriously. The key points to be remembered revolves around two simple notion2:

1) Generating an image base instead a a video file transforms a "uncertain" process into a fully reliable operation, on all existing PC configurations, even with few memory, even with a modest processor. The Advanced systems will also benefit from the increased reliability and of a rendering process shorter of about 20%.

2) Generating an image base opens to the 3Dmaster a world of possible enhancements, as well as an easy way to apply all necessary modifications: the work is smart and comfortable.

To explore the positive benefits of working with an image base, please view the demo pages.