The Smart Pig

Smart Information Column
Issue 4, 19 January 2004
Comparison of 11 Programs for Converting DVDs to DVD-5 sized DVD Discs

Quick Summary: To places original DVD videos that do not fit in the standard DVD-5 format size for a DVD (ie. the standard 4.3GB DVD+/-R discs used today) into such, various programs can be used.

The three highest rated when used to compress a 131 minute DVD video into a DVD-5 size are:
1. [highest quality, slow encoding, DVD compatible] Vegas Video 4 using the MainConcept MPEG-2 encoder. Best quality tested if you have the time and lengthy videos.
2. [high quality, slow encoding, not DVD compatible] Divx 5.1/Xvid 1.0b3 at DVD bitrates (here, tested at 4200kpbs). Good alternative to the above if you have a MPEG-4 compatible player or watch videos on your computer. Nevertheless, encoding times are very long and quality is not higher than the above.
3. [acceptable quality, fastest encoding, DVD compatible] DVD2One 1.3. Quality is noticable poorer than programs that reencode videos (DVD2One only removes bits), however, it is the fastest, highest quality program available that will compress videos to fit a DVD-5 size disc.

Naturally, compressing short length videos will always result in higher quality DVD-5 size output videos, so the program choice will vary from this test report which used a rather lengthy 131 minute DVD video. In particular, you may find DVD2One to be a better choice with shorter length videos.

In any case, for longer length videos, use a program such as Vegas Video 4 and reencode the entire video, losing menus and chapter marks for the highest quality DVD-5 size output. For shorter videos or anytime you wish to retain all menus and chapter marks, use DVD2One. For even higher quality output on longer videos that will require an enormous amount of time to process, consult the detailed and technical multi-step guides found on the web, at places such as http://www.dvdrhelp.com/

---

Many DVDs made today are larger than the standard DVD-5 format size used in most DVD+/-R recorders used today. (~4.3GB = DVD-5 standard; also known as single-layer, single-sided DVD)

In order to fit these DVD videos into the smaller DVD-5 format, one must reduce the video data size (and/or audio data size; here, we only deal with the video aspect since reducing the audio size often produces very few benefits). To do this, a program can either reduce the bits used by the video stream by taking out bits from the original video stream, or simply examining the entire video stream in depth and reencoding the entire video to a smaller size. Naturally, the former method is a very fast process whereas the latter, a very slow one. (On a 2Ghz P4, removing bits on a 2 hour video can be accomplished in about ~30 minutes, whereas reencoding can take up to several long hours to complete.)

Keep in mind that removing bits from the original video stream is a process which can retain all original menus, chapter points, etc; however, reencoding a DVD loses all such items and only the videos are retained (unless you spend time to rebuild menus, chapter points, navigation, etc.).

Here, a brief subset of the numerous programs available for accomplishing this task are compared. In addition, comparison to SVCD, VCD, and MPEG-4 formats are included. Keep in mind that substantial improvements can be made over time, and future verisons of the same program can gain significantly in both quality of video and speed of production. Because of this and the rapid changes present in computer technology, only a brief summary is provided of the hardware and software.

The objective is to compare programs which can fit a DVD-9/120+ minute DVD (dual-layer, single-sided) onto a single DVD-5 (4.3GB) disc while producing the highest quality, resolution, etc. possible. (Additional comparisons to VCD and SVCD formats provided. DIVX/XVID MPEG-4 encodings to DVD-5 sizes provided as well.) Also, the idea here is to produce output quickly and simply from the programs used - thus no in-depth steps were taken to produce the very highest quality that can be produced from these programs. Only the basic, obvious steps were taken. (If you want to research advanced methods of doing this, there are many guides online you can find simply by searching - http://www.dvdrhelp.com/ is a great start.)

PC: 2.0AGhz P4 with 512MB RAM under Windows XP SP1

Video taken from Speed: Save the Children - Speed Live 2003 DVD music video (Avex Japan, AVBD-16045). 131 minute video, frame screenshots take at 0:15:28 into the video; 4:3 full frame video; PCM stereo, one audio track, no subtitles.

This particular frame at 0:15:28 was used because of the numerous, complex elements which any of these programs must tackle to produce quality output.

These include:
* In the actual video, all four girls are in fast, unpredictable motion with minor camera motion. Along with various complex elements in the foreground and background, an action scene that demands the best from any encoder. The people are at some distance from the camera so additional difference in the encoders can be easily spotted, especially if quality of the figures suffer.
* Dark spot under the knee of the girl facing left on the left side. On poorer encoders, this spot vanishes.
* Folds in the pink top of the girl on the left side foreground. These folds merge together or vanish with poorer encoders.
* Spotlight on upper left corner with the beam crossing the top of a rectangular box. The top edge of this box vanishes with poorer encoders.
* Small box with white colored edges on the corner a large box above the head of the two girls on the left side. The gap between the white colored edges of the two boxes merge together in poorer encodings.
* Multiple faces in the far background. Dark spots for eyes and mouth disappear or are moved around to the wrong places with poorer encoders.
* Stairsteps on the left. Noise or blockiness increases around the shadows on these steps under a poorer encoder.
* Spotlight on the right. Dark background around this spotlight becomes blocky under a poor encoder.
* Two girls on right side. Clothing and arms. Increased noise & blockiness, reduced levels of detail under a poor encoder.
* Black objects in front, lower edge of screenshot. Increased noise evident on these black objects under a poor encoder.

Quality Ranking from highest quality to lowest (some ranked as ties).
Download all 17 Screenshots.

0 - Original DVD
1 - [slow, encode] Vegas Video 4 MainConcept encoder to DVD format (4200 kbps selected, High Video quality)
2 - [slow, encode] Cinema Craft Encoder 2.66 (closed GOP, 4:3 aspect, DVD compliant, 0 IRE) to DVD format
3 - [slow, encode] Dr. Divx 1.03 to Divx 5.1 format (Hi-Def Profile modified to 4200kbps and no resizing)
3 - [slow, encode] Xvid 1.0b3 to Xvid format (4200kbps)
4 - [slow, encode] Tmpgenc 2.520.54.163 to DVD format (4200kbps)
5 - [slow, encode] Ulead DVD Workshop 1.3 to DVD format (4000 kbps)
6 - [slow, encode] Sonic MyDVD to DVD format
7 - [fast, remove bits] DVD2One 1.30 to DVD format (CBR)
7 - [fast, remove bits] DVD2One 1.30 to DVD format (VBR)
8 - [fast, remove bits] CloneDVD 1.284 to DVD format
9 - [slow, encode] Intervideo DVD Copy 1.1 to DVD format
10 - [slow, encode] Pinnacle InstantCopy 8 to DVD format (HQ mode)
11 - [slow, encode] Tmpgenc 2.520.54.163 to SVCD format (Motion Search - Slowest)
12 - [slow, encode] Vegas Video 4 MainConcept encoder to SVCD format
13 - [slow, encode] Tmpgenc 2.520.54.163 to VCD format (Motion Search - Slowest)
14 - [slow, encode] Vegas Video 4 MainConcept encoder to VCD format

The best way to compare the results yourself is to download the screenshots, then view them with an image viewer such as ACDSee which automatically sizes all images to the screen (or keeps them all fixed to the same size) and allows you to use the mouse wheel found on many mice to quickly compare one image to the next. You can compare a subset of images quickly by either renumbering the ones of interest so that they are next to each other, or simply copy them to another folder.

The following are notes on the rankings:

1. Suprisingly, for this video and bitrate, Vegas Video 4 MainConcept MPEG-2 encoder performed the best out of all encoders. While Cinema Craft Encoder 2.66 is thought to be one of the best MPEG-2 encoders around, it may be true that to achieve the highest quality requires more adjustments to the encoder settings.

3. Although the still frames for both Divx and Xvid encodings appear smoother than the results from Vegas Video or Cinema Craft Encoder, playback of these videos reveals blockiness appearing during the playback of these videos. Also, these encoders do not retain as much of the orginal data as VV or CCE.
eg. notice the single dark spot on the floor below the knee of the girl facing left on the left side. While this is retained in the VV & CCE encodings, both Divx and Xvid almost erases any trace of it! An example of smoothing taken too far.
Nevertheless, both produce videos that appear more TV-like than the others ranked lower.

4. Here, Tmpgenc produces a poorer result than some will expect due to the lack of significant adjustments in both the encoder settings and added filters. The latter can significantly improve the quality of encoded videos at the cost significant amounts of time, but this test is about making videos simply.

5 & 6. Both encoders produce a significant amount of blockiness in areas of plain color, with Ulead DVD Workshop producing far less of this than Sonic MyDVD. Although Sonic MyDVD is included for free with many DVD recorders, these tests show that quality of this encoder isn't as good as you can achieve - not only in this test, but with other videos that you'd like to make into DVDs in general.

7. Here is where the fastest programs rank with DVD2One producing the best quality of all programs that reduce the bitrates of the original video stream (vs. the much slower reencoding of the video ala 1-6 above). The significant increase in noise throughout the screenshot is evident in video playback as well vs. the better encoders above. Although these bitrate reduction programs can often complete their tasks on a 2 hour video in less than 30 minutes, the results reflect the short time and care spent.
Nevertheless, these programs are quite useful if you are trying to fit a shorter length video onto a DVD-5 size disc since their output quality will be higher, and fewer artifacts will be seen. However, for longer length videos, these programs are only a starting point in quality.

8. While the interface of CloneDVD is easier to understand, the quality of the output is just a touch less than DVD2One, which is faster than CloneDVD as well.

9 & 10. These two programs, Intervideo DVD Copy and Pinnacle InstantCopy, both attempt a fast reencoding of the video similar to 1-6 above. However, as one may expect, fast produces poor quality videos. The videos produced by these two are the worst of the programs presented that encode to DVD format. Although this test pushes the limits of any encoder, the poor quality vs. even DVD2One above suggests that one can skip either of these programs in their considerations even when working on shorter length videos. The significant color bleeding, video noise, etc. is simply reflects the poor output of these two programs.

11 & 12. The SVCD samples provided here show the loss in resolution and quality vs. encoding to DVD/Divx/Xvid. Although the video quality suffers, the entire video is still quite enjoyable especially when played on a DVD player on an analog TV set. Tmpgenc would be the better choice here as you can easily tweak settings to achieve higher quality encodings.

13. Suprisingly, VCD encoding has been around long enough that very little can be said about the output here. What you see is about as good as you can expect short of some heavy tweaking within Tmpgenc to achieve higher quality, and only marginally at that given the low resolution of VCD (352x240 or 320x240 are typical sizes). Nevertheless, if you simply sit back from the TV and watch the video, you will soon ignore the blockiness and lower quality video that VCDs possess and only see the movie itself. VCDs are a great format for smaller analog TVs (32" or smaller) when viewed from a normal sitting distance of several feet/meters away from the set. They would be a poorer choice if you watch your videos on the computer monitor at the desk vs. the other options above.