No Compression

[ando775]

Reading through this forum and on other video editing sites, I've noticed that when doing different effects and things, it is suggested not to use any compression sometimes when rendering files. At least when there is more work to be done on that particualr clip. What is the reason for this? I'm relatively new to video editing, at least at the MSP level of complexity, and am still learning. Thanks.

[DVDDoug]

Most compression techniques are "lossy". They throw-away data, and this degrades the video to some extent.

Any processing or editing that requires a decode / re-code cycle will require a 2nd lossy compression step which will further degrade the video. When I make a crossfade between two MPEGs, the result is very "blocky" during the transition.

You will also loose quality if you convert between two different lossy formats. For example, from DivX (MPEG-4) to DVD (MPEG-2).

[Devil]

As a rule, there is little point in using uncompressed video on a project that started out compressed. For example, you are not going to improve quality of the original video if you convert DV to uncompressed. OTOH, if you do this and your computer has sufficient horsepower (especially the HDD), you can sometimes avoid some additional artefacts related to colour keying, although there are less drastic ways of handling the problems.

[Terry Stetler]

Keying: less drastic, yes. Better, not necessarily.

Those "less drastic" solutions often use a matte generator that trims more from a subjects edge than I care to lose. Otherwise you're getting into some rather expensive hardware & software solutions.

As far as doing many effect cycles using uncompressed video, this is how studios do it and the lesson should be learned from that. They wouldn't bother if it wasn't beneficial.

Every effect that's added to lossy video requires recompression and therefore incurs a quality loss. This is true for any file that has DCT (discrete cosine transforms) and/or temporal compression as part of its makeup. MJPeg, MPEG, DV, WMV, DivX etc. etc. all qualify for one or both reasons.

With simple effects this isn't much of an issue, but when compositing many layers or with keying or complex effects it can become a major problem especially when the recompression generations add up. ex: DV video suffers heavily once past the 2nd generation

Uncompressed video, or video compressed using only the Huffman algorithm (RGB, YUV, lossless MJPeg, HuffYUV etc.), suffers virtually no loss even when exposed to multiple effect cycles.

As such IMO video in a lossy format that is to be highy processed benefits greatly from conversion to a lossless format before extensive effects processing. Once finished itg can be expoted to any format with minimal losses.

[ando775]

Great information. Thanks everybody.

[robtywlak]

There is another option

Use MSP8's proxy file feature and work with 1/4 resolution (360x240) non compressed or HuffYUV compressed files. Stack all of your effects on multiple overlay tracks and then when finished do ONE render directy from sourec file to output applying all of the effects and layers at once.

The benefit to this is that is saves a lot of disk space and can be done on a slower machine. Render will be slow but you only have to do it once on the final full resolution product.

Regards,

Rob Tywlak

[Terry Stetler]

Yes, that's an option if the effects required are available within MSPro but that isn't always the case.

One example is creating speed change effects that will require interpolated frames. For this I use Dynapel's Motion perfect with uncompressed files to prevent the interpolated frames from being one generation of recompression advanced from their neighbors. MP is also perfect for creating interpolated frames to bridge frames that have been dropped.

Another example would be doing the various flame/laser/particle/lighting and other effects (Matrix-style bullet trails etc.) in Effects Lab DV or using After Effects.