Hello –

Welcome to 2004!

In this newsletter, Kevin Gordon, the creator of Asiva Technology writes about the differences between making selections in other programs and having maskless selections in Asiva. He also gives a few pointers on how to incorporate the Asiva Selection plug-in into your workflow. Roland Lee also adds to his article on pre-production by discussing rules of setting up a background replacement shot.

In the news you can expect to soon hear about upgrades to all of the Asiva plug-ins. Features such as the Snap Shots or 'memory dots' and primary color buttons found in Asiva Selection will be added to all of the plug-ins. Also you can expect to see edge selection with the ability to adjust intensity or expand or contract the selection, and a new preview option will color what you are selecting with your curves.

As for an Asiva Photo upgrade, please be patient, or start bugging the engineers with ideas for what you would like to see added by e-mailing support@asiva.com.

We are encouraged by your comments, whether good or bad, so please let us know what you think about the software by e-mailing info@asiva.com. And spread the word!

See ya next time,

Shapiro Consulting Group, Inc.

Maskless Selection in Asiva vs. Traditional Selection In digital image or video enhancement and compositing, the ability to focus filters, effects or Operations on a certain area is often essential for good aesthetic results. Traditionally, this has been accomplished by physically defined areas called mattes, masks or selections.

In Adobe’s Photoshop®, selections can be created from a wide variety of tools, could be bitmapped (either ON or OFF) or gradient, and may be 'floating' or saved to a new channel. Prior to the recent release of Photoshop® CS, selections were limited to 8-bits, or 256 possible values from black to white. In a selection mask or channel, white typically indicates “yes, process this pixel” and black means “mask out, or do not process this pixel”. In a bitmapped-selection, like you would essentially create using Photoshop’s Marquee Tools and Lasso Tools, the pixels would be either black or white. Using Photoshop’s Magic Wand Tool (assuming the tool’s tolerance is not 1) will create a gradient selection. A gradient selection would specify the degree to which every pixel could be filtered or altered by any number of Photoshop’s capabilities.

All of the Asiva software products, except for the Asiva Selection Plug-in for Photoshop, take a completely different approach: 'selection,' or where you wish to effect change is accomplished via the color properties of the pixels in an image, combined with vector-based Regions (rectangle, ellipse or paintbrush). These Asiva Regions allow for localization of a change in the event nearly identical colors exists in more than one area you want to effect. The curve settings for each and every Asiva Operation define a 16-bit gradient “choice” for each and every pixel, however, a fundamental difference is Asiva color-property based selections are saved neither in memory nor virtual memory (hard disk space). The larger the source file and the more filters, effects or Operations you have, the more important this fundamental difference becomes.

For example, lets say you have a 16-bit gradient selection for 10 different adjustments or filters in a Photoshop session. The source image is 5,200 x 5000 pixels, 16-bit per component RGB (about an 149 MB file). This would not be an unusual size for a photo going into a magazine. The space overhead requirements for the selections alone would be 26,000,000 pixels/frame * 10 frames * 2 bytes/pixel = roughly 496 megabytes. For the same file, an Asiva Photo Operation Sequence of 10 Operations would create an overhead of perhaps 100K at the most, even with a lot of vector-based Regions. So archiving projects with Asiva Photo becomes as trivial as saving an Operation Sequence. Their small size also allows for quick and easy e-mailing to clients or associates. On the other hand, Photoshop selections saved to specific channels may be further altered by any usable Photoshop tool like the Brush Tool. This does not mean Asiva Photo’s vector Region tools cannot do the same thing, it is simply a different methodology.

Besides a huge difference in overhead, Asiva curve-based selections can be controlled more precisely than traditional selections with minimal effort. For example, Photoshop's Color Range Selection method allows you to 1) pick or remove Sampled Colors and adjust the selection’s fuzziness; OR 2). select from a choice of basic colors or shadows, midtones or highlights. Selecting by method 1 or 2 allows no further refinements. Setting the three curves in Asiva’s Hue, Saturation and Luminance Maps allows for very fine definition of a color selection and its transition.

There are tradeoffs to both approaches, however in the grander scheme of things we believe Asiva’s maskless selection methodology is superior. In digital video applications there really is no choice - saving and storing discrete masks for 24 or 30 frame per second images becomes completely impractical.

Written by: Kevin K. Gordon, Chief Technology Officer SCGI

The Value Of Good Pre-Production IV

In our last newsletter we discussed a little about the different technologies available during the post-production process. In this article we’ll go over a very simple set-up with some basic rules to take into consideration when preparing for a background replacement shot.

Before we begin I feel some basic knowledge about lighting a set is required. All set lighting revolves around the 3-point lighting rule. Key, fill and back lighting. You light every object on the set with this rule in mind. The more key objects, the more lights you will need. We can discuss 3 point lighting but I feel there are sufficient articles out there that discuss this in depth. I think http://www.3drender.com/light/3point.html has a great article about this subject, or you can Google “three point lighting."

For this article, I will separate the elements into foreground (subject) and background (blue/green/red or any other suitable color – see previous article). To ensure optimal separation of the foreground element from the background, it is imperative you treat them as totally discreet elements when lighting them.

Foreground

Be sure you position your subject a sufficient distance from the background (I’ll explain this in the next paragraph). Begin with a dark room (with a utility light so you don’t trip over any wires!) and add in your key, fill, then back lights. Some things to take into consideration as you are lighting your foreground is to make sure you light it to match the new backgrounds you intend to use in post production. If you have different potential replacement backgrounds, light for the lowest contrast ratio possible. Color balance or contrast ratio adjustments to match the new background should not be too hard with a good clean image with even lighting.

Background

The background element, be it blue, green or red, needs to be lit as flat as possible. Choice of color is determined by the foreground subject. You obviously don’t want red when the subject is a redhead, or green when it’s a blond. Light sources for the background need to be diffused to ensure the absence of hard shadows. Make sure your foreground lights are on as they may have cast shadows onto your background that may need to be filled.

Setting the foreground subject, physically, further away from the background element will help eliminate this problem. On additional reason for the distancing of the foreground is the possible ambient spill that may result from the reflected color from the background element. This often results in difficult to remove color fringing in our foreground element.

If you’ve done all this correctly you should have a foreground element that should be fairly easy to separate from the background using any good compositing software capable of making a good color based selection. For difficult objects like hair or other translucent or out-of-focus objects make sure that you check out Asiva’s Selection plug-in that allows for true 16-bit per channel gradient selections, or Shift+Gain plug-in for true 16-bit functionality in adjusting contrast ratios and color balance. These plug-ins work in native 16-bit mode even if the input file is only 8-bits per channel, giving you 256 times more control over the HSV/RGB components of your image.

Written By: Roland Lee, Technical Evangelist SCGI