Adobe Media Encoder: Resize SD Video to HD – or – HD Video to SD

Last week, I wrote about how to use Apple Compressor to resize video between HD and SD. (Read that article here.) This week, I want to talk about the same process using Adobe Media Encoder.

NOTE: The process of resizing an image has multiple names. “Resize” and “Scale” both mean to change the size of an image. “Up-res” means to make a small image larger; such as from SD to HD, or HD to 4K. “Down-res” means to make a large image smaller, such as from 4K to HD, or HD to SD.

BACKGROUND

All video images are composed of bitmaps; rows and columns of pixels that are fixed in size, creating a fixed image resolution. This means that changing the size of an image always changes the image quality. When scaling smaller images larger, the images get soft and a bit blurry. When scaling larger images smaller, they lose resolution. This is caused by altering the limited number of pixels in a video image. We can try to minimize the damage, but nothing we do will make it go away.

Because of this, it is a good idea, in general, to avoid enlarging smaller images. Sometimes, though, that can’t be helped.

Additionally, most SD images have an almost-square 4:3 aspect ratio, as illustrated here in Adobe Media Encoder (AME), while the default aspect ratio for HD is 16:9.

NOTE: This image is typical of many SD images. It was shot on VHS tape, has poor exposure and is not completely in focus. In other words, it is like many, many movies in the archives. Resizing will make this image bigger, but it won’t make it look better.

The Aspect Ratio describes the proportions between the height and width. There are two principle ratios: 4:3 and 16:9. Because video changes “size” based on the size of the monitor displaying it, it is often easier for us to describe video using aspect ratio and total pixels than inches.

There are two image sizes for SD video:

• NTSC: 720 x 480 pixels, where the same number of pixels are used for either 4:3 or 16:9 images
• PAL: 720 x 576 pixels, where the same number of pixels are used for either 4:3 or 16:9 images

Both these formats use rectangular-shaped pixels, rather than the square pixels we are used to seeing on our computers and the web. (Converting these rectangular pixels to square pixel values leads to all sorts of different apparent image resolutions for SD video.)

For the purposes of resizing from SD to HD, the actual number and shape of the pixels is not important; though they are very important when resizing from HD to SD.

If, when we resize, we don’t take aspect ratio into account, we end up with a very stretched image. This is bad. Instead, follow the guides below to resize your images.

IMPORTING AN IMAGE

There are several ways to import an image in Adobe Media Encoder:

• Choose File > Add Source
• Type Cmd + I
• Drag a file from the Finder into the Queue tab of the interface

While, in most cases, I am a keyboard junkie, when compressing files I normally just drag them into the Queue tab; as I did with this Black Bear clip.

Once a file is imported, it always shows up in the Queue tab, with a format, preset and output file name and location assigned to it.

NOTE: You can change any of these settings by double-clicking the blue text. Since this article focuses only on resizing an image, I won’t be discussing most of these settings.

When resizing a file, all the settings we need to adjust are in the Preset that’s assigned to the clip. Double-click the blue Preset text under the file you want to modify to open and display the settings.

(Click to see the full image)

OK, yeah. This looks pretty intimidating; and not just the bear. However, the settings that control scaling (which is the fancy name for resizing) are all in one place. So you don’t need to learn the entire interface to resize your videos.

CONVERTING SD TO HD

The first thing you need to know about resizing an image is that the codec you use does not affect the process. Yes, different codecs will yield different file sizes and images quality, but the PROCESS of changing image size is not affected by the codec.

Second, when you resize an image, you are determining what size you want the finished image to become. AME will figure out what the source image is and what needs to be done, your job is to specify the results.

Third, almost all HD video uses a 16:9 aspect ratio, or greater. Which means that not only do we need to enlarge the image, we need to figure out how to place a 4:3 source image so that it fills a 16:9 frame. There are three options:

1. Add black bars on the sides, called pillar boxing
2. Zoom into the image so it fills the frame
3. Stretch the image so it fills the frame

AME supports all three – HOWEVER, if you want any of your actors, or black bears, to ever talk to you in the future, NEVER stretch the image.

Add Pillar-Boxing

Double-click the blue Preset name in the Queue tab to open the Export Settings window.

Click the Output tab in the top left corner to display what the output file will look like when compression is complete.

Select Scale to Fit from the Source Scaling menu. This fits the entire SD image into the frame using its 4:3 aspect ratio, and adds black bars on each side to extend the image to the full 16:9 frame.

The is the best option when you want to see the entire SD image.

Fill the Frame

Choose Scale to Fill when you don’t want pillar-boxing. Now, the image zooms in so that the center of the image fills the frame.

While this creates a “larger” image, it does so by cropping pixels off the top and bottom of the frame. Notice how the space above the bear’s back, along with its feet and what it is walking on, is now missing.

There’s no right answer on whether to use pillar-boxing or crop the frame. But, my personal feeling is that I would prefer to see the entire image and pillar-boxing doesn’t bother me that much.

Stretch to Fit

There is a third option, Stretch to Fill. Not to mince words, but this is just awful. Don’t use it.

IMPROVING IMAGE QUALITY

When you are scaling a video larger, you can marginally improve image quality by checking Use Maximum Render Quality. (Its located at the bottom right of the window. While this doesn’t work magic, it will improve the image. For example, I can see a different in the texture of the bear’s fur when this option is turned on.

This option will slightly slow compression times, but not significantly.

DOWN-RES FROM HD TO SD

Up-resing SD to HD is comparatively easy, because:

• There’s only one frame aspect ratio: 16:9
• All pixel aspect ratios are square
• The HD image is progressive

NOTE: Yes, HD has interlaced images. However, when converting SD to HD, you retain higher image quality by up-resing to a progressive image, then add interlacing during the actual edit.

Down-resing to SD is much more complex, because:

• There are two different frame aspect ratios: 16:9 and 4:3
• There are four different pixel aspect ratios
• There are both progressive (rare) and interlaced (common) images

In short, SD is a mess.

BIG NOTE!!

There appears to be a bug when a video is down-sized in the current version of Adobe Media Encoder (2017.0.2 release) then displayed in QuickTime Player 7. Regardless of whether the image is 16:9 or 4:3, QT 7 displays it as 4:3.

This problem does NOT appear when the image is viewed in QuickTime Player X or Adobe Premiere.

Here is the work-around:

1. Apply the NTSC DV Widescreen preset that matches the source video’s frame rate
2. Double-click the blue Preset name to open the Export Settings window

3. Change the video codec from DV25 NTSC to DV50 NTSC. The rest of the settings are fine.

NOTE: Another  good thing about using the DV50 codec is that it has the potential for much higher images quality than DV25.

CONVERT HD to NTSC SD 16:9

Unless you’ve been give other instructions, converting HD to DV retains the most image quality and provides the most options for broadcast, DVD or future editing.

The easiest way to find all the NTSC presets is to search for NTSC in the Search box, as illustrated here. Which ever preset you choose, the sizing process is the same.

With NTSC video, you have the option for two different frame rates. The best advice is to match the frame rate of the source video to the frame rate of the scaled video. This also minimizes motion artifacts.

NOTE: If you shot 30, 59.94 or 60 fps, select a preset using 29.97. If you shot 23.976, 23.98 or 24 fps, select a preset using 23.976.

The aspect ratio is indicated in the Preset name: “Widescreen” means 16:9, while the absence of “Widescreen” means 4:3. In general, convert HD to widescreen, however some satellite and broadcast distributors are still using 4:3, which means you can’t totally ignore this option.

Here is a 1080p HD image, with the DV Widescreen option applied. Both frame rates are 29.97.

Note that Scale to Fit is chosen and, if you look carefully, you’ll see two thin pillar bars  (thick black edges) on each side.

Because the HD image has plenty of extra pixels to work with, I could choose Scale to Fill to get rid of the black bars. Yes, this slightly zooms into the image, losing some pixels at the top and bottom, but the ability to have a full-frame image, to me, makes this slight image loss worthwhile.

Unlike Up-resing, I didn’t see any difference when Use Maximum Render Settings was turned on or off. Leave it off to save time.

CONVERT HD to NTSC SD 4:3

Converting the image to SD 16:9 is easy, because the entire image is preserved. Converting to 4:3 means you have two choices:

• Preserve the entire image, but add letter-boxing at the top and bottom
• Zoom into the image so it fills the frame, but lose pixels on the left and right edge

Apply the NTSC DV preset, then double-click it to open the Export Settings window.

NOTE: While not required, considering changing the codec to NTSC DV50 for files which will be viewed in QuickTime Player 7.

Scale to Fit is the default, which automatically adds letter-boxing. Again, I didn’t see a difference when choosing Use Maximum Render Quality.

Or, choose Scale to Fill. This zooms into the center of the image, losing pixels on the left and right edge but filling the frame without letter boxing.

Again, be sure to match the frame rate of the source with the frame rate of the compression preset.

CONVERTING HD TO PAL

The process of converting HD to PAL is very similar, except we don’t have the issue of frame rates to worry about. PAL is always 25 fps.

NOTE: If you shot 50 fps in HD, this will convert easily to 25 fps. If you shot 24 fps, you can convert to 25 fps PAL or 24 fps NTSC.

Again, you have the option to convert the codec to DV50 PAL. This is not required, but will yield potentially better image quality. Note that the video will not display properly in QuickTime Player 7, but will display properly in QuickTime Player X and Adobe Premiere.

Again, use Scale to Fit if you want letter-boxing. Or use Scale to Fill if you don’t.

SUMMARY

The process of resizing an image involves paying attention to the technical limitations of each video format. But, once you understand how this works, you can create a custom preset to apply to all clips you plan to up-scale, which will automate the entire process.

I’m puzzled by the scaling problem that appears in QuickTime Player 7 – however that problem does not appear when the video is opened in QuickTime Player X. And Premiere handles all these converted files perfectly. Experiment for yourself to see which of these two codecs works the best.

If possible, use the DV50 codec for the best image quality, though DV25 yields good results.

---

Bookmark the permalink.