Jump to content
Sign in to follow this  
Dimitris76

Digital Video Capturing

Recommended Posts

Another question that has to do with the ever popular subject of video recording our flights!

Can somebody explain to me (or point to a good on-line technical article) what do the terms: interlaced, non-interlaced, upper field first, lower field first, etc do mean? These are all user configurable setting on my Ulead capturing software.

I can find info in forums around the web giving advice like: yeah, your video looks like crud because it's interlanced for TV viewing and you should de-interlace it for the PC monitor, etc but not a nice article explaining what we should do in every case and why?

Thank you,

Dimitris

Edited by Dimitris76

Share this post


Link to post
Share on other sites
Another question that has to do with the ever popular subject of video recording our flights!

Can somebody explain to me (or point to a good on-line technical article) what do the terms: interlaced, non-interlaced, upper field first, lower field first, etc do mean? These are all user configurable setting on my Ulead capturing software.

I can find info in forums around the web giving advice like: yeah, your video looks like crud because it's interlanced for TV viewing and you should de-interlace it for the PC monitor, etc but not a nice article explaining what we should do in every case and why?

Thank you,

Dimitris

I don't think anyone will tell you exactly which settings you should use because it depends on the hardware you have to work with. I can only give you 3 things you should look out for.

1. Capture: If you plan to do any editing (other than basic, lossless cuts) after you capture your video, then use the lowest compression codec you have available to you. A low compression codec will take a lot more space on your hard drive than mpeg2 will, but it will yield a lot better quality. Two good codecs that I've used are DV and mjpeg (720x480). After your capture, then you can convert to mpeg2 or h264, usually with much better results than if you had initially captured in mpeg2 or h264. As a general rule, the more compressed a codec is (i.e. most of the mpeg4 codecs), the worst the quality will be if you capture using that codec, and doubly so if you plan to do any editing (other than basic cuts, which can be done losslessly). Mpeg2 capture may yield better results than mpeg4 capture.

2. Editing: DV and mjpeg can be edited with much less quality loss than mpeg2 or h264. With DV, you can probably get away with multiple conversions (re-encodes) with little loss in quality. If you try to re-encode mpeg2 even once, you'll probably notice a considerable loss in quality. Although Kilrah has suggested that mpeg2 -> h264 or mpeg2 -> divx may yield better results than mpeg2 -> mpeg2. If you plan to edit mpeg2, I would suggest using something like VideoRedo or mpeg-vcr, which can do basic cuts without any loss in quality. h264 can also be edited losslessly, but the process is more complex. Unfortunately, real editing usually requires re-encoding, which is why DV is a better codec.

3. Final format: My two preferences are h264 (in mpeg4) and mpeg2 because they yield good results and they are likely to be the most future proof. Mpeg2 (720x480) is a standard for both DVD and Blu-Ray. And h264 is a Blu-Ray standard. Divx/xvid is also extremely popular. VC-1 is the third Blu-Ray compatible codec, although I have no experience with it and can't recommend it in any way. But supposedly it's on par with h264. These are probably the only codecs I would consider. In my mind, there's little doubt that h264 will eventually become the de facto standard. It gives you the best bang for the buck (i.e. excellent quality at low bit rate). So, it would be my recommendation to use this format, but not for capture. But, if you initially capture in mpeg2 and only want to do basic cuts (no re-encoding), then I wouldn't do any conversion. Any conversion you do will always result in quality loss.

So, in short, if you have the proper hardware, my preference for SD would be to capture in DV or mjpeg and convert to h264/AVC (in mpeg4 container). As for what settings to use for the conversion, I'm still working on that myself. It's still an immature technology. My goal is to eventually make it Blu-Ray compatible (720x480 is accepted on Blu-Ray discs), but since I don't have a Blu-Ray burner or player, I'd only be guessing as to how to make it compatible at this point. Mpeg2 would be my choice if I'm still not comfortable with h264.

As for whether to interlace or de-interlace, just try one of each and see how they look on your TV and computer monitor. But I don't believe you can de-interlace mpeg2 without re-encoding, which may reduce the quality. If I had to re-encode mpeg2 to de-interlace it, I wouldn't do it. I would de-interlace from my original source, before it's ever converted to mpeg2.

Edited by ginger_marianne

Share this post


Link to post
Share on other sites
Can somebody explain to me (or point to a good on-line technical article) what do the terms: interlaced, non-interlaced, upper field first, lower field first, etc do mean? These are all user configurable setting on my Ulead capturing software.

I can find info in forums around the web giving advice like: yeah, your video looks like crud because it's interlanced for TV viewing and you should de-interlace it for the PC monitor, etc but not a nice article explaining what we should do in every case and why?

Interlacing is a thing from the past that was invented for the analog video signals and still carries on today. Basically, when the composite video signal was specified, engineers faced a compromise regarding resolution and framerate. It was too hard to send a smooth 50/60 frames per second (for PAL and NTSC respectively) with the number of lines they wanted, because of excessive bandwidth. They either had to reduce framerate to 25/30 images per second, losing movement smoothness, or reduce image definition.

They finally cut the deal in 2, and decided to divide the full frame in 2 fields, the even lines (lower field) and odd lines (upper field). Each field is sent alternatively at the 50/60 fields per second rate. So you can have the full definition over 2 consecutive fields, which are averaged by the eye with the help of the persistance of CRT screens, and movement smoothness is ensured by the fast refresh.

This implies that both fields that compose a single image are not recorded simultaneously. This was not a problem back then, as the concept of single image didn't exist, as it was way before one would start talking about buffers, so signals were captured, processed, transmitted and received "live" with no intermediate storage. Fields were just displayed one after the other, slightly shifted vertically, and the CRT/eye combination would average them showing the full detail and maintaining motion smoothness. However, now digital devices work quite differently. When working on a PC, it wouldn't be convenient to edit images that have half their lines black. Also, LCD's don't have the persistance CRTs had, so can't be fed with half the lines being active and the other half being black as that would be noticeable and disturbing. Those lines have to be filled by some software method to mimick the averaging that was made by the CRT/eye in the older days, and is what is called "deinterlacing".

There are 2 main methods to do it:

To do things properly and actually reproduce what the CRT did, you would take each field, and calculate the missing lines using some interpolation algorithm, display this frame that is now complete, then when the next field is there you repeat the process, effectively displaying 50/60 full images per second.

The second one is to take it crudely, and merge 2 consecutive fields into a single frame. By doing so you're halving framerate (you only display an image each time you have 2 fields available and merge them), and most importantly you're creating artifacts (the famous jagged lines or comb effect), because you're merging and displaying simultaneously 2 fields that were NOT recorded at the same time. So, imagine you're filming something that is in movement, it will have moved in the 50th/60th of a second elapsed between recording of the 2 fields, so won't be at the same place in the second field than in the first one. But, you merge and display those simultaneously, so half the lines will have the subject at one location, and the other half will have it at another place, creating the ugly effect.

Now the problem is that for some unfortunate reason the most common one in the computer world is the second one. On a computer, if you play an interlaced video file, most players by default will use the 2nd, "bad", deinterlacing method, that actually isn't really called one. The few that actually can do proper playback-time deinterlacing (like VLC) usually require user input to do so (you need to start playback, right click on the video, image, deinterlacing, select a method).

So, what most content providers will do is deinterlace the footage before they create their video file. That video file is then no more interlaced (called progressive or non-interlaced), contains only entire frames that stupid players can just display one after the other without further treatment. However, deinterlacing methods in editing programs don't fully implement the ideal method, but will take 2 consecutive fields and "blend" them together into an image, in a way that eliminates the jaggy lines, but only keeps half the framerate (25/30fps). A reason for that is that doing it ideally would have twice as many frames to store in the file. It is possible to do true 50/60 frames/sec deinterlaced files, but complicated as programs usually don't support it.

So, in summary:

- Interlaced, non-interlaced - well, guess you got it ;)

- Upper/lower field first: As editing programs will treat frames instead of fields during the time you work with them, they need to know which of the fields actually came first from the analog video source. This is usually the lower first for SD analog sources. If set wrong it might create very jumpy video, as the software would interpret fields coming from the camera ordered in time as A,B,C,D,E,F as B,A,D,C,F,E.

- For computer viewing, the ideal in experience would be to store interlaced video files, and use a program that can deinterlace them properly at playback time to view them. BUT, as most programs and online video platforms can't do that, and/or 99.999% of users have no clue about how to do it should their player support it, one will prefer creating deinterlaced files when publishing to target computer playback. The second best would be a proper 50/60fps deinterlacing, but as said it isn't easy to do as most programs don't support it. It would also make bigger files.

- For TV viewing, one would prefer the interlaced version. All TV's that require it incorporate deinterlacing systems, and publishing progressive content would needlessly restrict framerate to half the standard rate.

- It you target both TV and computers, all comes down to what you want. Either you create one progressive version for computers, and an interlaced one for TVs... or you just only make a progressive one that would be compatible with everything at the expense of lower framerate, which most people woudn't notice anyway...

When I publish something, I'll create:

- A progressive version to put online and give to people

- An interlaced version that I mostly keep for me and will show to people on my own devices with proper software and settings

- An interlaced version on DVD or such if TV viewing is required.

Edited by Kilrah

Share this post


Link to post
Share on other sites

This is an excellent thread. I've been using the EasyCap and the ADS VideoXpress USB Video Capture devices to record my flight videos on my laptop (not FPV yet) and have been very dissapointed with the results. I've spent a lot of time trying to find out more about digital video and have had limited success. This thread is just what some one like me has been looking for. I'm sure there must be a lot of people out there who are as much in the dark as I am. Any information like this is really very useful, maybe you could advise on the best set up, software, hardware etc., to use. Perhaps other forum members are using a similar method to record and have knowledge, experience which would be invaluable. Please keep contributing.

jon davies

Edited by jonforstedavies

Share this post


Link to post
Share on other sites

I've personally never de-interlaced mpeg2. They've always played properly for me interlaced.

I was wondering which (if any) of these de-interlacing methods allow the mpeg2 to retain DVD compatibility?

I would argue that there is absolutely no reason to use mpeg2 unless it is DVD compatible. Each of the other codecs that I previously mentioned can equal the quality of mpeg2 at a lower bit rate.

Thanks

Edited by ginger_marianne

Share this post


Link to post
Share on other sites
Any information like this is really very useful, maybe you could advise on the best set up, software, hardware etc., to use.

Hardware - The EasyCap is perfect and cheap, I actually tried one yesterday.

Software - well, it's a lot about personal preference. The Ulead software that comes with the EasyCap is well usable. Otherwise I've used VirtualDub, K!TV, VirtualVCR... all free programs with different levels of adjustment possibilities. The Ulead from the EasyCap is a very good start as it's easy to use, no lots of confusing settings etc. The drawbacks some have is that they will only show you a small size preview during capture, or a very annoying interlaced one. Ulead is actually OK as it will deinterlace and make a fullscreen preview if you click the fullscreen icon. The only gripe is it will "fill screen" and not "fit screen", so if you have a 16/9 PC screen it will stretch the live image over it instead of maintaining proportions. If it really annoys you, you can always switch to a 4/3 resolution before.

Capture formats - If I know I want the best quality, I'll use DV. There are codecs you can install that will then be available in your recording programs. MainConcept DV codec is known to be a very good one, and light on resources, but payware. If you want to go free, there's an alternative called Cedocida DV codec. Good too, but will use maybe 2-3x more processing power, so if you have a (very) slow machine like a netbook it might be problematic. You install this, set Ulead to use AVI, in the advanced settings you choose "use software compression", and select that codec. There's a DV option in Ulead, but for me it didn't work (it always came back to AVI by itself after selecting it). If I'm not particularly expecting anything but just record as a memory or "in case something happened", I'll just use MPEG2 at 6-8Mbps. That's already 3-4x smaller than DV so saves space, yet is suitable for real-time encoding unlike the newer codecs that are way too processor-hungry, and the loss of quality vs DV is acceptable. Ulead can do that too, I didn't try it though.

Editing - I'll take the captured file as it is and import it into my favorite editing software. I'll edit the video, then export it in DV format. I'll usually archive that on DV tape, and also use it to create the different files I'll offer for publishing. Note the DV file is still interlaced at this point, so as close to original as it can, and an optimal base for any target.

Publishing - For web, I mostly use WMV, encoded with the "official" Windows Media Encoder, at about 1000-1500kbps rate, full 720x576 frame (PAL for me). I deinterlace at this point by selecting that option in WME. From quite a bit of testing I did, this format offers really good performance, and is easy to use. I tried the newer codecs, but usually the programs to encode in those formats are still a bit "experimental", you have 4 times more important parameters to try and find good combinations of - Maybe you can get better results with them, but at the expense of a lot of time and effort, dealing with program crashes, incompatibilities (WMV is widely supported, while newer formats tend to be more problematic), etc. If I need to give the result for TV viewing - well, not many options, DVD will do it, with the highest bitrate possible on the available space on the disc (max 8Mbps for compatibility).

In some cases I will just export from the editing program without the DV intermediate file - but I noticed that some programs didn't do as good a job as WME, so results weren't as good. But for HD for example, with the lack of a low loss intermediate codec like DV I usually have to take that route.

Edited by Kilrah

Share this post


Link to post
Share on other sites
I was wondering which (if any) of these de-interlacing methods allow the mpeg2 to retain DVD compatibility?

Progressive is part of the DVD spec AFAIK, however it requires an explicitly compatible player and TV to read. But unless you have a progressive source (like film) it wouldn't make sense to use. As said earlier - you'd limit yourself to 25/30 progressive frames/sec instead of a smoother 50/60 fields/sec the TV is made to handle properly. Deinterlacing is interesting for computer use as interlaced content is often improperly handled.

Share this post


Link to post
Share on other sites
Publishing - For web, I mostly use WMV, encoded with the "official" Windows Media Encoder, at about 1000-1500kbps rate, full 720x576 frame (PAL for me). I deinterlace at this point by selecting that option in WME. From quite a bit of testing I did, this format offers really good performance, and is easy to use.

I believe WMV is just a container for the codec. Actually, the VC-1 codec, which I mentioned earlier is part of the Blu Ray standard, is often placed in a WMV container. I've seen some samples, and it does provide excellent quality at a low bit rate. I used IgCutter (my lossless wmv/asf editor), and it worked perfectly with it. It's easy to check which codec specifically is being used with GSpot. If you encode VC-1 at 720x480 (or 720x576), there's a chance that it will be Blu Ray compatible without re-encoding, assuming you guess (or know) all the other settings needed for compatibility.

WMV generally does provide excellent quality but finding tools to edit it (without re-encoding) is my main problem with it. It's not that it's a hard format to edit. It's just that Microsoft will harass and threaten anyone who writes software for it. I think Virtualdub supported wmv/asf until version 1.3. It's a little too proprietary for me, but otherwise provides very good quality.

The only format I'm 100% comfortable with at this time is DVD compatible mpeg2. I generally capture in DV/mjpeg (preferably in Virtualdub if it supports my capture hardware), edit with Virtualdub, convert to mpeg2 (6+ Mbps, multiple pass) using Cinema Craft Encoder (pay software), and author the DVD using TMPGEnc DVD Author (pay software). TMPGenc also converts the audio to AC3 if you buy the separate plugin. As I mentioned before, I do my VOB/mpeg2 cutting using VideoRedo (pay software) and/or mpeg-vcr (pay software). TMPGEnc can also do basic cutting. VideoRedo is better than the other two because it always has flawless audio sync, it can combine mpeg streams into one, and is intuitive to use. TMPGEnc DVD Author, mpeg-vcr, and VideoRedo are all lossless, and none of them are capable of video encoding, although VideoRedo and mpeg-vcr are both very good at repairing both mpeg1 and mpeg2 streams. Between the two of them, I can fix virtually any problem with mpeg1/mpeg2/VOB videos. As I mentioned TMPGEnc can convert audio to AC3, and it is Dolby Digital branded. This is important to me because there have been problems with AC3 encoders in the past. Mpeg-vcr can encode audio to mp2.

Since most of what I use is pay software, I don't expect anyone to go out and buy them, but with this combination of software, I feel I have mpeg1 and mpeg2 covered from every angle. As is pretty obvious, I hate re-encoding anything. I prefer to fix video while maintaining the original quality.

Versions I use: TMPGEnc DVD Author 1.6, mpeg-vcr 3.14, VideoReDo 2.5. These are all very old versions. They've probably all been updated extensively, so my descriptions may not accurately reflect the current versions. I haven't looked for anything new because they all still work fine for me. I think VideoReDo is working on mpeg4 support, although I haven't been paying attention. As far as I know, there are currently NO mpeg4 editors that are anywhere near adequate.

Edited by ginger_marianne

Share this post


Link to post
Share on other sites

I wanted to add a couple other things. There are no short cuts when it comes to dealing with video encoding/editing. I've wanted to pull my hair out several times dealing with this stuff. Unless you actually have some understanding of the topic, you'll never get the results you want. What I would suggest is picking a process (possibly from one of those described in this thread) and work with it to see if you can get the results you want. Something else I would suggest is possibly downloading some video (preferably those without copyright violations) and work with those. Use GSpot to see what type of videos they are. Try to edit/cut them and see how those results turn out.

Another thing that I will mention is that some formats are much harder to deal with than others. For instance, AVI is very easy to deal with, thanks to Virtualdub. From my experience, MPEG4 is incredibly hard to deal with. The strange thing is that both AVI and MPEG4 carry the exact same codecs. For instance, h264 can be placed in both an MPEG4 and AVI container. In fact, you can convert between the two without having to re-encode (although the process isn't necessarily easy). The only way I've found to actually edit/cut an MPEG4 is to first convert it to AVI, edit in Virtualdub, then convert back to MPEG4.

And I would argue that the only reason AVI is so easy to deal with is precisely because of Virtualdub. In my opinion, this is the most revolutionary program for video editing. This is probably the reason that AVI is so popular, even though the consensus among purists is that AVI is garbage and should be gotten rid of entirely. Until MPEG4 gets a program equivalent to Virtualdub, it will probably never take off in popularity. And from my recollection, the people who work on Virtualdub said that MPEG4 was basically a mess to edit. So, I don't know if MPEG4 is going to have that Virtualdub equivalent any time soon. There is another open source container format, OGM, that was supposed to compete with AVI and MPEG4. It also can hold the same codecs as AVI and MPEG4. It never took off, and I would argue the reason it didn't was because there was no good way to edit it.

Easy editing is the key. And it usually only takes one good program, like Virtualdub. But I don't know if there's something inherent in MPEG4 that makes it very difficult to edit. As it stands now, I basically have a huge list of procedures for editing MPEG4. I've never had to make a list for AVI or any other format.

Share this post


Link to post
Share on other sites
In my opinion, this is the most revolutionary program for video editing.

Just a little gripe there - I would not call VirtualDub a "video editing" tool, but more like a video manipulation tool. It can cut/join bits of AVI, transcode them, apply basic effects - but cannot practically do anyything of what you would do when doing proper video editing - like taking various video files, cut them into blocks you can easily rearrange, then apply transitions between them, using different file formats with different properties (like frame size) and mix them together, etc.

Virtualdub is a great tool and I use it a lot, but usually only to open AVI files and cut the useless parts out of them if I even bother doing so before dropping them into a real editing program, open a file and grab a still frame out of it, do some MPEG->AVI conversions maybe with deinterlacing... but it doesn't go much further.

To really make a video I'll use Adobe Premiere.

Edited by Kilrah

Share this post


Link to post
Share on other sites
Just a little gripe there - I would not call VirtualDub a "video editing" tool, but more like a video manipulation tool.

That's true. Virtualdub is not a proper editor. But I do a lot of video manipulation. The fact for me is that I won't encode a video into a format unless I know that I'll be able to cut out parts or rearrange it later without having to re-encode it. Let's say I choose divx/avi format now for a video that I make. I want to know that 10 or 20 years from now I'll be able to manipulate it, cut out parts, etc. while still maintaining the original quality. I can't tell you how often people will re-encode a video every time they want to perform even the most basic tasks. Most people can't seem to grasp the fact that once a video has been edited and produced (i.e. compressed), you likely don't have to re-encode it to modify it, split it into clips, or rearrange the sequencing.

So, that's why I stress programs like Virtualdub (avi), VideoRedo (mpeg1/mpeg2/VOB), IGCutter (wmv), etc. more than other types of programs (e.g. encoders). You should have a program like these for any format you choose to compress your video into. Notice that I don't mention one for mpeg4. That's part of the reason I'm not very enamored with mpeg4 at this time. It's a pain to deal with. I guess I sound like a broken record, but I really do consider this very important. I personally won't give a procedure for producing a video without mentioning one of these programs (because the re-encoding problem is so pervasive).

And I stand by my statement that Virtualdub played a part in the explosion in popularity of the avi container. Just download some video from usenet and run it through GSpot. Chances are that it passed through Virtualdub at some point. And many people who download videos like to be able to split the video or make clips out of it. It's true utility for me is after the video is produced/compressed. Besides being a manipulator, it can do video conversions as well. But it's not a proper editor, so I guess I use the word "edit" loosely.

Edited by ginger_marianne

Share this post


Link to post
Share on other sites
Can somebody explain to me (or point to a good on-line technical article) what do the terms: interlaced, non-interlaced, upper field first, lower field first, etc do mean?

Thank you,

Dimitris

I actually looked into this a little bit more, and there is actually a problem with some encoders, such as Cinema Craft Encoder, and the field order. First of all, all of my stand-alone DVD players can play both BFF (bottom field first) and TFF (top field first) just fine, so it's not necessarily an issue of which one you pick, although TFF is probably the more conventional choice.

The problem arises when the flag set on your mpeg-2 file is mismatched with the way your mpeg-2 is actually encoded. For, instance, if it's flagged as TFF but actually encoded as BFF, then you'll have a problem with stand-alone players. The picture will look jittery. Another problem with this is that the file will play perfectly on your computer monitor. As far as I can tell, there's actually no reliable way to detect this problem until you burn the DVD and play it on your TV. The problem won't even show up if you have a TV out on your video card. The only reliable way to detect the problem is to burn a DVD. I suppose it may show up on a computer monitor if there is something wrong with the mpeg-2 codecs installed on the computer.

This usually occurs with encoders that don't detect the field order of the input file. It's expected that the user knows the correct settings with the particular input file. With CCE's default settings, it sometimes creates a mismatch and it sometimes doesn't. It depends on the field order of the input file. I'm not sure if this varies based on the version of CCE you use.

The solution is simple. If the DVD you burn looks jittery, use a program called ReStream to change the field order. If it was previously flagged as TFF, then change it to BFF. If it was previously flagged as BFF, then change it to TFF. The change can be verified with GSpot. And no re-encoding is required. It's simply a problem with the flag.

A simple preventative measure is not to delete the original source file until you're sure the DVD plays correctly on a stand-alone player. Possibly burn a small sample to a DVD to verify.

Edited by ginger_marianne

Share this post


Link to post
Share on other sites

Kilrah - you mention the Easycap USB (Syntek STK1150).

I have one of these, and although it captures fine, it is a pain in the neck in practice, as there's no way of removing it - apart from uninstalling the driver!

XP's 'Safely Remove Hardware' system tray applet doesn't seem to recognise it. If you remove it physically, the driver partially crashes, leaving a phantom STK1150 apparently in the system but not working. Even re-booting doesn't help.

Have you had any problems, or do you just leave this huge dongle plugged in all the time?

Oh, and any comments about the free Panasonic or Quasar DV codecs?

Share this post


Link to post
Share on other sites
The fact for me is that I won't encode a video into a format unless I know that I'll be able to cut out parts or rearrange it later without having to re-encode it. Let's say I choose divx/avi format now for a video that I make. I want to know that 10 or 20 years from now I'll be able to manipulate it, cut out parts, etc. while still maintaining the original quality.

I don't care about this, for I always keep one low compression version in addition to the published versions. For SD that's a DV file or tape, for HD it's HDV file or tape.

If I want to redo anything, reuse parts, etc I'll use that archive. So the published versions can be anything, I don't need/want them to be editable. In fact, it's even better if they aren't as it will deter some Youtube morons from reusing the footage. Formats used for distribution, like WMV, are meant to be as efficient in terms of quality/size ratio but not to be reused, as that ratio implies high complexity, so it's normal if it's hard to do. Formats like DV or MPEG2 will need more space to achieve same quality, but their advantage is that they low compression ratios and complexity, so are very easy to handle. For archiving you typically put quality at the top - without caring much about the size.

And what is to consider is that new codecs use a LOT more processing power while actually not giving a much better size/quality ratio. Take the 2 common formats for HD camcorders - HDV (MPEG2-AT-25Mbps) and AVCHD. While AVCHD now finally starts having decent implementations on the latest cameras, you still can pretty clearly see the difference between HDV and the usual 15MBps AVCHD. I have seen a couple of cameras that boasted higher quality - but on those the AVCHD bitrate is increased to 17-20MBps. So to have the same quality, the difference is only slight... And MPEG2 is extremely fast to decode even on slow machines and dedicated hardware devices - while AVCHD requires a 2.5GHz dual core just to play without stuttering! On my 3.6GHz dual core desktop PC AVCHD is a huge pain to edit, I can't even play it normally in the timeline without doing a pre-rendering. I could do that with MPEG2 on my laptop 4 years ago.

And the latest small pro cameras (Sony EX1 for example) still use MPEG2 as well.

Kilrah - you mention the Easycap USB (Syntek STK1150).

I have one of these, and although it captures fine, it is a pain in the neck in practice, as there's no way of removing it - apart from uninstalling the driver!

XP's 'Safely Remove Hardware' system tray applet doesn't seem to recognise it. If you remove it physically, the driver partially crashes, leaving a phantom STK1150 apparently in the system but not working. Even re-booting doesn't help.

Have you had any problems, or do you just leave this huge dongle plugged in all the time?

Huh? I've just been plugging/unplugging it without asking any questions, just as I do with my other capture cards. What you usually don't want to do is unplug while being used by a program, but as long as it's idle I never have problems... What software are you using?

Edited by Kilrah

Share this post


Link to post
Share on other sites
I don't care about this, for I always keep one low compression version in addition to the published versions. For SD that's a DV file or tape, for HD it's HDV file or tape.

That's a good strategy if you can do it. But how many people who would ask the type of questions in this thread would\could actually do that? And besides, that could get very expensive. Depending on how much video you produce, that could require a lot of tapes (or more likely hard drives). DV takes up 4 times more space than DVD compatible mpeg-2 (and maybe 8 times more space than h264 or VC-1). I just wouldn't be able to do that, although I wish I could.

You're right that the newer codecs are very compressed, so frame accurate cutting is probably not possible. But it is possible with mpeg-2. That's another reason why mpeg-2 is a good compromise for SD and DVD is a better choice for me. And when a good encoder is used (with the proper settings), the rendered mpeg-2 is quite acceptable for me (not a huge quality difference from the original source file). The only problem is when you try to re-encode that mpeg-2 at some later point. Obviously, the DV is a better choice if you plan to re-encode. But, at some point for me, I just have to say I've done enough with this video, and I don't want to do any more work on it.

And don't forget that most people probably won't even take the bother to record in DV to begin with. Most will probably record in mpeg-4 and maybe some in mpeg-2 (which is usually a better choice than mpeg-4). Then most of those will probably re-encode their already highly compressed video. So, I definitely think it's worth mentioning ways to cut video rather than re-encode. Or at least mention that it's possible and relatively easy to do.

As an aside, perhaps your observations about the quality differences (vs size) between mpeg-2 and AVC/h264 are only in the world of high definition. Or perhaps they only apply when real-time encoding is done, as in a camcorder. Because I would say that when I take a DV source file and encode to mpeg-2 and AVC/h264 (obviously in SD), the difference in quality vs. size is obvious. AVC wins hands down. Also, consider that in SD, mpeg-2 recorders (e.g. ChaseCam PDR100) often perform markedly better than mpeg-4 recorders, and that probably has little to do with the quality of the codec and everything to do with the processing power it takes to encode in real-time in those formats. All things considered, when you look at all the factors (size, quality, compression, processing power, ease of editing), I like mpeg-2 the best, especially in SD. Quality good enough for archival (at least for me), size small enough for distribution (with the added benefit of being DVD compatible), and frame accurate cutting possible. In short, it's perfect for SD, in my opinion.

Edited by ginger_marianne

Share this post


Link to post
Share on other sites
But how many people who would ask the type of questions in this thread could/would actually do that?

Could - well, anybody? It's not that complicated to save your video 2 times, especially when DV is so fast to encode it will only take a few more minutes...

Would - nobody if they don't think about it - that's why I mention it here as an ideal solution :)

And besides, that could get very expensive. Depending on how much video you produce, that could require a lot of tapes (or more likely hard drives). I just wouldn't be able to do that, although I wish I could.

A 1TB hard drive costing about $110 nowadays holds about 75 hours of DV/HDV - try again ;) Personally, if I add all the final versions of the videos I've made in the past 7 years including the ones I wouldn't consider worth much and the near-duplicates when I've made different versions of the same video I'd maybe be able to reach 15 hours...

Then most of those will probably re-encode their already highly compressed video. So, I definitely think it's worth mentioning ways to cut video rather than re-encode. Or at least mention that it's possible and relatively easy to do.

Find me one single person who would produce a video to show around that only consists of raw cuts - The first thing people want to do is add titles, nice transitions,... and this makes reencoding mandatory ;)

I like mpeg-2 the best, especially in SD. Quality good enough for archival (at least for me), size small enough for distribution (with the added benefit of being DVD compatible), and frame accurate cutting possible. In short, it's perfect for SD, in my opinion.

I agree. Actually, I just recorded CSI from DVB-T tonight and played with it - the quality is really great, and it's only 3.3Mbps MPEG2... their encoders for sure are excellent.

Little nitpicking, cutting on MPEG2 is GOP-accurate, some programs can indeed do frame-level cutting, but in the case of in-cuts they reencode the last frames of the particular GOP the starting frame was in :)

Edited by Kilrah

Share this post


Link to post
Share on other sites
Could - well, anybody? It's not that complicated to save your video 2 times, especially when DV is so fast to encode it will only take a few more minutes...

Would - nobody if they don't think about it - that's why I mention it here as an ideal solution :)

A 1TB hard drive costing about $110 nowadays holds about 75 hours of DV/HDV - try again ;) Personally, if I add all the final versions of the videos I've made in the past 7 years including the ones I wouldn't consider worth much and the near-duplicates when I've made different versions of the same video I'd maybe be able to reach 15 hours...

I've thought about it. I just won't do it. Personally, if I produce something, I don't want to do it again. It's a lot of work. Deleting the source file gives me a sense of finality.

Also, I store all my files in triplicate (3 hard drives) with an indexing system and MD5 hashes (of all files). If it's worth keeping, I do this. If it's not, I don't keep it at all. Doing this process on hundreds of hours of DV would be too much. I suppose I could alter my system and just keep one copy of the DV files. But it's not for me.

Find me one single person who would produce a video to show around that only consists of raw cuts - The first thing people want to do is add titles, nice transitions,... and this makes reencoding mandatory ;)

If the video was already produced with transitions and titles and you deleted the source file, then I think doing basic cuts on the mpeg-2 would work if done properly. After all, you already have some transitions. Most videos can handle some basic cuts after that.

If the video was initially recorded as an mpeg-4 (on one of those cheap recorders), I would forgo fancy transitions and use basic cuts. After all, the loss in quality of re-encoding an already very compressed file would not be made up for by nice transitions and titles.

Also, I'm not entirely sure it's not possible with some creativity to add transitions without re-encoding most of the file. For instance, I took a sample mpeg-2 capture from TV. I split the file into 2 clips with VideoReDo. I didn't just split it, I split it in the middle of a word. The narrator was saying "confusion". The first clip ended at "con". The second clip began with "fusion". I re-combined the 2 clips and it was seamless. The video went back together perfectly. And the word "confusion" sounded perfect. What does this mean?

Well, if you want to create a transition in an mpeg-2, just cut out the few seconds where you want to put the transition in. Re-encode that few seconds with a transition, then re-insert it into your video. Simple and you don't have to re-encode your video. And it should look great. Just make sure that the re-encoded part perfectly matches the rest of the video. I'm not saying this is completely practical, but it's definitely possible. So, I wouldn't say it's mandatory to re-encode the whole video. And besides that, my understanding is that basic cuts are the most common types of transitions anyway in Hollywood movies.

I think we agree on most things. As for the rest, we agree to disagree.

Thanks for the debate.

Edited by ginger_marianne

Share this post


Link to post
Share on other sites
I've thought about it. I just won't do it. Personally, if I produce something, I don't want to do it again. It's a lot of work. Deleting the source file gives me a sense of finality.

Oh sure, I get that one when deleting the 30GB of source files I had for a 10min video... but I definitely always keep that 10min video in the best possible quality as I usually am proud of it and don't want to lose any quality on it :)

When I watch it myself or show it to people at home I'll watch the best quality archive for best results :)

What does this mean?

Well, if you want to create a transition in an mpeg-2, just cut out the few seconds where you want to put the transition in. Re-encode that few seconds with a transition, then re-insert it into your video. Simple and you don't have to re-encode your video.

VERY complicated. Usually to be able to merge clips they need to be the EXACT same format, and it would be near impossible to match the exact configuration that was used in the original file, especially when made by a hardware device.

For example, I have an Archos recorder. When recording long videos, it will split it into new files each 2GB to overcome filesystem limits. The cuts are clean so you could merge them without problem in a proper editing program by putting both clips one after the other. However, VirtualDub won't let you merge them, because even if they are the same format it will find some tiny discrepancy like 1ms audio skew in the second file and will refuse to merge.

So yes, cut a file and merge it again is no problem - but treat one part then putting it in again is another story.

Not to mention that the reencoded bit will be of noticeably lower quality.

Incidentally I did use that "technique" once for what must have been my first video around 10 years ago... I had a 8GB hard drive in my PC, with far less than enough free space to hold the ~15 mins of DV of the source video. I had to edit a part, render, encode in virtualdub, delete source, capture the next bit, edit, encode, merge,... about 5 times :P

Edited by Kilrah

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this  

×