Putting the Magic in the Onscreen Image

You can fix it in post-production

It isn't just the camera, and it isn't about realism

With today's emphasis on big screens with amazing display resolution, and the inexpensive DSLR cameras with amazing image capture resolution, several questions are raised, such as: "How much detail should I capture?" and "How real should the image look?" and "Which equipment should I use?" and "What about that mysterious encoding?"

Many things affect detail... or detail affects them. Realism and mood are directly affected by the level of detail captured in an image. Conversely, detail is affected by camera capabilities, lighting, image movement, encoding compression, bandwidth, image rendering size and resolution, and choice of formats: SD, HD-lo (720), HD-hi (1080). Where do you start in determining each of these elements?

Start with Hollywood. Hollywood has been projecting images on very big screens for years. Every blemish on an actor, such as a wrinkle or spot or scar, is coated with thick "pancake" makeup, or embellished with a body stocking. Why hide all of this? Movies are our great escape. There is nothing real about them except they portray a possible journey in another land and time. Distracting detail can destroy that illusion. The image on screen allows us to project our idealized fantasy journey onto the actor or set.

Get real about getting real

Despite the occasional emphasis on realism, the unrealities of movies are that they compress time and space by connecting disparate events portraying some reality created by a writer's, producer's, director's, and actor's interpretation of a character's possible aspirations, feelings, and actions during their journey. These are far from untruths, but far from reality. Reality is full of warts and distractions.

The realism should be in the story, as much as practical, and the image should be more "impressionistic," like an impressionist painting, suggesting reality without getting bogged down in irrelevant details. If a weed or a dead leaf would ruin the image, the painter simply leaves it out.

The mood of your movie affects the texture of the image. To give two examples, a gritty, stark reality mood will likely have the lights in the "cold" spectral range (think fluorescent lights) and the image high resolution showing broken edges, warts, and litter. This makes the real, gritty world of the story inescapable to the viewer.

In a romantic scene, the lights will likely be lower and in the warmer spectral range (think incandescent light bulb... if there are still any for sale). Incandescent lights bring out the flesh tone, and lower lighting brightness masks the details that spoil the image.

Between these two extremes are where most movies fall. The image is not detailed enough to slap us in the face with reality, but neither is it so soft that we see an overly idealized version.

Next page: Hard or Soft images

  1  Page go right 2

Or click a Page selection on the top menu.

Hard or Soft images

Hardness and softness of the image is controlled by set lighting, camera selection, and encoding.

Lighting

Bright, white light with stark shadows will give the harder, almost garish texture. This can be enhanced by adding green or gray to make it even more garish. Dimmer light with more yellow and red (indirect sunlight, incandescent light), gives more mellow shadows, and a softer look. It brings out the flesh tones better so people look more attractive.

Encoding

Less encoding compression gives more details, while more compression softens details for a softer texture. Encoding involves a lot of technical things that affect image delivery and quality. Encoding determines the resolution of the image, how many times the image will be displayed in a second, and how much it will be compressed. In general, the higher the resolution (horizontal lines), the higher the frame rate, and the less compression, the more detail will be displayed making the image clearer.

For theater and high quality Internet streaming (HD-hi), resolution needs to be 1080p at 24 fps. Higher resolutions may work better in theaters, but are not usually available unless using something like a Red camera to capture the image. Since not all viewing devices support high resolution, with more advanced streaming servers, lower resolutions are delivered from multiple encodes of the original, or encoding from the original to lower resolutions on the fly. While 720p (HD-lo) may be adequate for DVD, it doesn't always look good enough, however the industry is finding that when a TV or other device receives 720p and automatically increases it to 1080p, the image may look better than if delivered in 1080p. The reason is because the image is "softer," because it lacks some detail.

1080p means there are 1080 horizontal lines in the image, and "p" means progressive, which means all 1080 lines are displayed at one time (not every other line displayed, then the others, as in "interlaced" formats. FPS means the number of frames per second. At 24 fps, every one of those 1080 lines are delivered to the screen 24 times in one second.

The resolution, the amount of encoding compression, and the frame rate, also depend on the type of image being sent. 24fps is the television standard for prime time broadcasts. The image usually looks good. But during sports events, or movies with a lot of car racing or other fast movement, 24fps may not be fast enough. It may leave motion artifacts on the screen, and look jittery. For movies with a lot of action, 30fps is a better frame rate selection. Higher frame rates than 30fps usually don't improve the image quality enough to justify using them, and they are bandwidth hogs, making the movie more expensive to send over the Internet.

Next: Encoding Compression

1 Page go left  2  Page go right 3

Or click a Page selection on the top menu.

Encoding Compression

This is a bit technical to explain the effect of compression - skip this section if you want

High compression adversely affects images. The last thing that happens to a movie is the selection of encoding compression, and compression affects the softness of the image. What compression literally does is remove some of the detail from the image, so that the "size" of the movie is smaller, meaning the number of bits on the disc is smaller so that it can be sent at less cost.

Because compression removes detail, don't edit with compression on. Each successive editing pass (if the image was saved) will lose more detail. Don't compress until you are completely finished editing.

There are various types of compression. JPEG (the basis for MPEG, MP4, and H.264 encoding), and other encoders such as VC-1 through 8, and WebM, always compresses to some extent. JPEG, as opposed to RAW formatting, reduces the available colors to 256. So if an image has over 16 million colors, JPEG (not Hl264) will reduce the variation to 256. H.264 improves on this by using 10 bits instead of 8, to represent a color, and has a lot of other encoding level benefits, which gets a bit complicated. See the Wikipedia article: H.264/MPEG-4 AVC.

RAW format is the image taken from the camera as somewhat unprocessed data, with all of the color and camera settings preserved. RAW format can be processed by a computer with no encoding loss, and is preferred by some, but the amount of file size is very large. All cameras process images to some extent - you con't actually get the data directly from the CMOS sensor, or actually the composite of 3, which presents a color interpretation. Sensors don't actually sense color.

Once you get past the camera output, there is additional compression in JPEG. Think of squeezing the juice from an orange. With the juice removed, the orange is much smaller. But you "could" put the juice back in, and the orange would be the same size and taste about the same. This is the first type of compression that JPEG uses. It looks for patterns, and takes them out, to be put back in later. When compression is set to 0, it doesn't harm the image, even if you save and re-edit.

Higher compression is a "lossy" and "blocky" business. Lossy means that it loses information that was displayed in the image. Lossy compression looks for data that is similar, and replaces it with other data. For example, if there are 300 shades of pink in the image, it reduces these to a pattern of 200 shades. The higher the compression, the more shades it eliminates, until there are 100, 50, or even 10 shades remaining. The more times it is compressed, the more times it eliminates shades.

The more the compression, the more it is noticeable. Chances are, most people won't notice if the 300 shades of pink are reduced to 150, or even 50. But with more reduction, faces get splotches that look unnatural - like someone hit them with a pink paint brush, and then a red one. This is commonly seen on old movies that have been highly compressed and resolution has been reduced, and then mass licensed to distributors. The quality of the image is very poor.

Compression contains predictive algorithms. Images often don't change much from one frame to the next, so compression predicts that the next frame will be the same, and only sends the part that changed. If the compression is set too high, images smear across the screen, often showing the blocks that compose the image. You also get that effect of pieces of the person's face, especially their cheek, floating around on their face. These problems again are typical of old movies that have been highly compressed and resolution has been reduced, and then mass licensed to distributors. Image quality and viewer enjoyment takes a major hit.

Learn by other's bad examples. Review the output and make sure you haven't overcompressed the movie, causing odd coloring effects and floating pieces of faces.

Compression also looks at an image as if it were a checkerboard composed of square blocks. While we think of images as composed of lines, compression finds it more efficient to save and send blocks. When you get poor reception on your TV, you can see the blocks appear. When compression is set too high, or when an image is "hard," the blocks also can be seen.

Getting rid of blockiness can be accomplished by a second (recommended) or even more passes through the encoder. Some encoders can be set to do up to 100 passes. Keep in mind that with each pass, you lose more of the detail in the picture, but the picture becomes "softer" and the sharp edges are removed. Just a few passes smoothes out the blocks so the details blend better.

Next: Encoding Resolution, Camera capabilities, Summary

2 Page go left  3  Page go right 4

Or click a Page selection on the top menu.

Encoding Resolution, Camera capabilities, Summary

Resolution affects image quality. The lower the resolution that the image is displayed in, the less detail will be seen. Think of it this way: When you are close to a tree you can see all of the individual leaves, branches, and bumps on the bark. The farther you get from the tree, the less individual detail you can see. As you back away, the bumps on the bark disappear, then the small branches disappear, and then the leaves blend together into a clump with shadows. You can no longer make out the individual leaves.

Resolution varies from 1080 lines down to 480 (or even less) on various display devices. The scan starts at the top of the screen, creates a line across, goes down a bit, and displays a new line across, goes down a bit more and does it again. On good monitors and TVs, you don't notice the lines.

Resolution is less important on smaller screens - in effect you are backing away and seeing less detail. In general, the larger the image, the more important is higher resolution. There is a direct relationship between resolution and image quality.

Camera capabilities

Most modern professional digital cameras offer 1080 resolution. Some, such as the Red, offer even more. If using film, the type of film will affect image quality in the same ways described here about digital images. The differences you will see in digital cameras have to do with softness of the image, how the camera handles color, and how the camera performs with motion. Cameras vary from being more warm and soft with images, and being hard and cold with images. This is especially critical with faces, and camera selection will depend on the hardness or softness you want in the image. However, this can be adjusted in post-production with color correction and encoding.

DSLR cameras tend to have more problems with motion artifacts than you have with film. Higher "shutter" speeds (fps) may help with this problem, and the frames per second can be adjusted down to 24 in post-production.

H.264 is the encoding format that will be the most used in the future. It offers the highest compression with the best quality. H.264 saves images with all of the original color detail. Some digital cameras will deliver an H.264 output, which saves having to convert the encoding to H.264. In more practical terms, H.264 high compression means having to change the memory card on the camera less frequently, and less space used on your hard drive or disc. (WebM, a free rather than licensed for money codec, which is used in FireFox and Chrome browsers instead of H.164, has not been in development for nearly as long as MPEG based encoding, but some day it may achieve H.264's capabilities.)

HDSLR cameras are coming into their own in cinematography. The industry is changing and some directors and cinematographers are "sold" on DSLR. They produce high quality images, are very inexpensive, and generally have excellent field depth (less need for focus). Choice of a particular camera depends on the image softness, color handling, and motion handling you want for the production.

For long productions DSLR cameras need special gear to adapt them. Cameras on a long production need to sit on tripods, have good lenses, and be given fluid movement through long control levers. Equipment to adapt HDSLR cameras to production capabilities is coming on market. For example, see The Master Cinema Series. For comparisons of various cameras see the excellent series, The Great Camera Shootout.

There are many excellent cameras, including the Nikon, Panasonic, Sony, and Canon brands (no association with Movie Stream Productions). An excellent camera for independent production is the Panasonic AJ-HPX3000 - convenient, but a bit pricey. Panasonic camera rental places. Also the Canon Canon EOS-C300, with similar specs. Watch for rental prices to fall and become affordable to independents on these cameras. The Panasonic is capable of recording 16:9 in H.264, 24 to 30 fps progressive, 720p or 1080p HD, and has 80 Gig of onboard memory for up to 100 minutes of recording time (you should also output directly to another device for safety and monitoring). This is a top of the line camera in its class (barring something like a Red). The Canon EOS-C300 video format is MPEG-2, and can record up to 160 minutes on one card (you should always record to a backup device). In general movie cameras can be rented for ~$100.00 a day and up, with $250.00 per day a typical price. Cameras like the Canon XA10 can be rented for as little as $20.00 per day from Aperturent.com.

Other production equipment resources: Gearboom.com, Atlanta area.

Encoders

If you don't need to do editing, just encoding, then you can use software intended just for encoding. The final image definitely needs to be monitored to make sure that motion artifacts, floating pieces, blocks, etc., are not introduced into the image. This means that online encoders, which have become popular, are probably not applicable. Some of the better brands for encoding to H.264, 16:9, NTSC or PAL, are listed below:

  • Microsoft Expressions Encoder pro
  • Sorenson Squeeze

Today and the future

As digital sweeps the entire marketplace, standards are rapidly emerging for theater and home. Higher resolutions for theater, and even home theater, are even now common. "As of 2009, the most common acquisition medium for digitally projected features is 35 mm film scanned and processed at 2K (20481556) or 4K (40962160) resolution via digital intermediate[citation needed]. Most digital features to date have been shot at 19201080 HD resolution using cameras such as the Sony CineAlta, Panavision Genesis or Thomson Viper. New cameras such as the Arri Alexa can capture 2K resolution images, and the Red Digital Cinema Camera Company's Red One can record 4K. The marketshare of 2K projection in digital cinemas is over 98%." - Wikipedia article on digital cinema. These are very expensive cameras, and are less likely to be used by Indies. You should keep in mind that analog TV resolution was only up to 525 lines on 37" screens, acceptable but not the best as screen sizes increased, and that for home viewing, higher resolution than 1080, or even 720, is not usually very productive. If your target is theater, then you should try for 2K resolution.

Summary

There are no "never nevers" about creating images. It may be just be a matter of preference as to which equipment and techniques you use - what you are accustomed to and comfortable with. But choosing equipment and encoding is mostly about putting together the right combination of lighting, cameras, encoding settings, post-production work, and techniques to achieve the image texture and effect you want while staying in your budget. Practice makes perfect.

You should also consider viewer preferences. For example, modern digital TVs typically have a setting that allows users to select the level of detail and color warmth they want. Many simply don't want to see highly detailed images that are distracting and have cold colors, or are too bright and washed out and lack contrast, opting for a "softer" image.

Have you produced a good independent movie or short? Get more attention for your shorts on FlixStreamer.com. Rent your movie on FlixStreamer.com for more profit.

- Dorian

Firefox is a Registered Trademark of the Mozilla Foundation.

Google Chrome is a Trademark of Google Inc.

Nikon is a Trademark of Nikon Corporation

Panasonic is a Registered Trademark of Panasonic Corporation

Sony is a Trademark of Sony Corporation

Canon is a Registered Trademark of Canon Inc.

Red is a Registered Trademark of Red.com, Inc.

End

3 Page go left  4  Page go right 5

Blog. What do you think?

4 Page go left  5  Page go right 6

Or click a Page selection on the top menu.

5 Page go left  6  Page go right 7

Or click a Page selection on the top menu.

6 Page go left  7  

Or click a Page selection on the top menu.



In this article:

moviestreamproductions.com Webutation

MSP Insider Articles

  • Recent
  • Make Your Own Movie: Part 1, Production
  • Make Your Own Movie: Part 2, Distribution
  • Can There Only Be one? Regional production
  • What is the future of HEVC?
  •  

    Movie Stream Productions and MSP Insider are productions of TechGenie Media, LLC - www.techgeniemedia.com

    Copyright © 2011 TechGenie Media.

    Home and About us.