Good Housekeeping: Is there any quality loss on rendering?

[maxfrost01]

Hi,

I'm shooting High Def film and capturing to my hard drive as mpeg files.
Each 60 mins takes around 13GB of space (which is big but expected).

Despite my extraordinary filming talents ( ) I would figure that at least 50% of every film is useless rubbish that could be cut and deleted immediately - it will NEVER make it to the final edit and it just takes up space on my hard drive.

Good Housekeeping would suggest that I should do a rough cut immediately whilst I remember what I shot (and can quickly find the bad bits), render the file and store it until I'm ready to spend more time doing the clever editing (which could be six months or a year from now). This will save editing time later and disk space now.

Question 1: Do I lose quality every time I render? Would this 'Good Housekeeping' approach mean that when I get to doing my final editing I am starting with a file where the images/sound are not quite as sharp as they would have been if I was working with the original, saved MPEG file?
Question 2: If this IS a smart way to go (i.e. rendering now as a rough cut and again on final edit) is there a way of marking points at which I will want to cut the film later (e.g. scene changes)?

Hope you can help.
Previous experience tells me you will!

[jchunter]

If you have long sections of bad video, such as forgetting to turn off the camcorder when you put it away in the camera bag, then just stop the capture and fast forward to the next good section. Then begin capturing again.

However, if your bad video is in hundereds of small pieces, it makes more sense to leave them in place. If you get serious about video editing, you will need a larger hard disk, anyway.

As for quality loss when rendering, if you maintain the original video properties, I doubt that you will ever see any difference, even if you include a resolution chart.

[maxfrost01]

Hi, John

Thanks for the reply.

My 'bad video' problem is the latter not the former. It's been a while now since I shot a tape with the lens cover still on but hey, who knows....

I tend to shoot quite a lot of film in the hope that I capture that 'golden moment'. For example, at the weekend I was filming my 10 year old son playing football (soccer not the American stuff ) and shot 15 minutes of film during which there was 1 minute when he scored a great goal!

This means I have (say) 8 minutes of boring stuff, I minute of glory that I want to keep forever, followed by another 6 minutes of dross. I find it easiest to capture the whole 15 minutes rather than do a rough edit during capture. I only really know the bit I want to keep when I see it in the preview screen which means I have already captured stuff I don't really want!

So what I would normally do is a) capture the 15 minutes of film b) do a very rough cut taking out (say) 12 minutes of boredom c) render a 3 minute clip which I will then edit properly sometime in the future.

This means the final footage will have been rendered twice rather than once but if I understand you correctly you're saying I should be able to do this without any noticeable quality loss.

First question: do I understand you correctly?
Second question: Am I missing something? Is there an easy way during capture of not capturing the stuff I don't really want?

Thanks for your help,

[roy wood]

> Do I lose quality every time I render?

Hi Max in theory every time you render an Mpeg file you could lose some definition (This has been hotly disputed many times on this forum).

> Is there an easy way during capture of not capturing the stuff I don't really want?

Not unless you can find a programme that can read your mind.

As was stated previously stop the capture when you reach that point... then restart capture just prior to the video you want to keep. You'll be left with a smaller file and a little bit of tidying up on edit.

[gordonwd]

I just did something similar, taking a 3-hour MPEG and editing it to trim it down into more manageable segments that I could make individual videos out of. When I went to the "Share" step to output a trimmed video, I selected "Same as project settings" for the output MPEG format, and the fast speed with which VS10+ created the new MPEG made it seem to me that it was not "re-rendering" it at all, but was really just copying the old MPEG data to the new file.

[maxfrost01]

Hi, Roy

An editing programme that can read my mind? Now there's a thought....what a winner that would be!

Regarding the stopping and starting during editing - yes, I can certainly do this but every time I stop the capture and restart I'm a) creating another separate file on the hard drive on the same subject matter (e.g. my son playing soccer) and b) I'm likely to miss the next golden moment - i.e. by the time I realise it is golden and hit the capture button the moment has passed!

Thanks for your help with this.

[jchunter]

This means the final footage will have been rendered twice rather than once but if I understand you correctly you're saying I should be able to do this without any noticeable quality loss.

First question: do I understand you correctly?

You understand me correctly. I have smart-rendered mpeg video that included a resolution chart, half a dozen times and can not detect any loss of resolution.

[Ken Berry]

I am not sure if John will disagree with me, but it is not so much a question of *whether* recoding mpeg will cause loss of quality. IMHO it will -- it is by definition a lossy format like JPEG.

Rather, the question should be: how much quality is lost and when does it become noticeable to the naked eye.

John can of course speak for himself on this. But quite some time ago now, he conducted some serious experiments, using resolution charts. And he established that, if you start with high quality (high resolution, high bitrate) mpeg-2, you can recode it 4 or 5 times before there is a discernible loss in quality.

While there are some convinced nay-sayers out there who will believe until the trumpet sounds that *any* editing and recoding of mpeg-2 is bad (or at least unless you use a dedicated third party mpeg editor), I at least have modified my attitude considerably since John's experiments. Now I say that if you keep your quality high and keep editing to a minimum, then you can recode mpeg-2 at least 2 or 3 times with no loss of quality which is visible to the naked eye.

As for Gordonwd's comments, he is referring to the Smart Render feature in VS. And indeed, its virtue is that it will only re-render those parts of the original mpeg-2 which have been edited. Thus, if you cut a long mpeg into several parts, Smart Render will only re-render around the cuts; if you add a title, it will only re-render around the title. This is all, of course, if you don't change any of the fundamental properties of the original video.

On the down side, however, is the fact that Smart Render has been notorious in the past for causing loss of sync between audio and video on some computers (though this has never been a uniform problem across the board). We are not really sure why, but our usual advice is that if you suffer out of sync problems, simply switch Smart Render off. Unfortunately, this will then throw you back into the potential dilemma of how many times you can recode (fully) an mpeg...

[maxfrost01]

Hi, Ken/John

Thanks for the reply/advice which make good sense to me.

I use Smart Render at the moment and haven't had any sync problems (but will bear that in mind) so it sounds like I can edit and re-edit without worrying about quality loss.

[Ken Berry]

... if you keep editing to a minimum, don't recode more that 2, 3 or (according to John 4) times, and keep the properties as high as possible. If you exceed these limits, then even SmartRender will start introducing quite visible artifacts around editing points.

Also be aware that I am not sure how SmartRender handles the addition of background music/audio which continues for the entire length of a video. I have seen suggestions that while the render of the video goes slow again to process the audio stream, it does not actually touch the video (except in those parts where the video stream may have been edited). But I just don't know...

[jchunter]

I guess it is time to differentiate between discernable and actual quality loss when rendering any compressed video format.

For example, it you drive every day to and from work at 70mph, while your wife stays home, you will actually age less than your wife (due to relatistic time dilation). However, your age difference will never be discernable, because the effect is tiny, (unless you drive to and from work near the speed of light).

So, there is actually some loss but if you can't see it or even measure it, who cares?

A better case could be made for quality loss when you transcode from one video format to another because this involves decompressing the video and then recompressing it according to a different set of rules. If the rules are not compatible, then there is more quality loss than with rendering. However, if I keep the video bitrate high, I can't see any loss of resolution when I transcode Mpeg2 HDV to Divx HD.

Secondly, it is important to stress that there is no evidence that smart-rendering causes problems. Turning off smart render will sometimes repair corrupt video files (by forcing a complete re-encoding) but this can hardly be interpreted as being the cause of the problem.

[maxfrost01]

Awesome stuff, guys.

Not only have you answered my question but you have set me on a quest to find editing software that can read my mind and a car that can travel at the speed of light.

I will let you know how it goes

Thanks as ever,

[Black Lab]

Not only have you answered my question but you have set me on a quest to find editing software that can read my mind and a car that can travel at the speed of light.

You need a DeLorean with a flux capacitor for that.

[2Dogs]

Of course editing MPEG2 is great if you like macro pixels. Check every transition to see them. VS does a good job of hiding them, but step thru the video in Virtual Dub and you will see them clear as day.

On the finished DVD, they'll show as brief flickers. Most people don't notice them, and even don't mind.

As Ken said, you can get around the problem with the use of a dedicated MPEG2 editor - which of course VS is not.

Since modern camcorders seem to be inevitably drifting away from DV in favour or more compressed lossy formats, perhaps it's an aspect VS will have to address in future versions.

[GeorgeW]

Not only have you answered my question but you have set me on a quest to find editing software that can read my mind and a car that can travel at the speed of light.

You need a DeLorean with a flux capacitor for that.

And a processor speed of 1.21 Jigawatts