CVS X4: Considerable quality-loss at rendering AVCHD-footage

[Jan-61]

I am confronted with a considerably quality loss when I edited and rendered the material from my Sony HDR-CX130 camcorder. It shoots *.m2ts 1920x1080 at 50i (this is in fact 25 fps, I assume).

The loss is visible when there are moving objects or when the camera moves: very not-fluent and not-smooth movement-representation, while the original footage playbacks fine and smoothly in any mediaplayer (Sony PMB or Windows Mediaplayer).

As if the rendered material has a framerate of 12 fps! Which isn't the case, in fact.

I did some experiments: render to other formats like MPEG, tried it on the laptop of a friend who also has CVS X4 installed (W7 SP1, i5 2,3GHz, 4GB RAM),I used properties of first videoclip (which actually doesn't happen), set camcorder at 50p; all without better results.

Project properties I use in CVS X4:

PAL (25 fps)
Microsoft AVI-files
8 bit, 1920 x 1080, 25 fps
Frame-based
Microsoft Video 1

This I was told here to use in order to get stills that are also 1920 x 1080 in size.

I also read this topic, but no usefull answer was given. Not sure that is the same problem there.
http://forum.corel.com/EN/viewtopic.php?f=1&t=50456 .

Is the un-fluent playback of edited AVCHD normal, will it be better in version X6; is there a workaround to help this?
This is visible when playing back the rendered footage in Sony PMB and WMplayer as well.



Jan.

[lata]

Hi Jan

Project Properties are of little use for High Definition video they are more appropriate for Standard Definition video, something I think Corel should address, and yes we have pestered Corel on this issue.
Setting the frame size to 1920 x 1080 does help in taking a snapshot in project playback.
Settings Menu to use Widescreen 16:9. and that is all that is required for the project properties.

Share -- Create Video File, DO NOT use Same as Project Settings these are unsuitable for HD video
If Same as First Clip is available should produce similar/same video quality as original.

You said “I used properties of first videoclip (which actually doesn't happen),”
What do you mean that doesn’t happen, are you saying the option is not available or the properties used are different?

Can you give full properties of your original video file, right click the clip and select properties, what are they?,
take a screen shot Alt+PrtSc will copy to clipboard, create an image to attach to your post.

I don’t know if X6 will improve your renders, but you could download the trial and give it a run.

Are you able to upload a short sample, maybe to 4Shared, this will allow others to test your file.

[Ken Berry]

I would also like to see the Properties of your rendered video, so we will know precisely what we are dealing with.

One thing with HD video -- whether AVCHD or HDV -- is that fast horizontal panning should be avoided as it seems to create a tearing effect on the eyes, almost painful to watch if too fast. If you look at professionally filmed HD, you will see the pans are rather slow and not over a huge angle. (Personally, I try to use a floating head tripod for pans as this controls the speed better...) 50p video will be far less painful than 50i, by the way...

That being said, it sounds to me more like a field order problem, and hence my desire to see your rendered video Properties. Your original AVCHD at 50i will be Upper Field First. (50p would not show a field order and would be progressive.) And your rendered video thus should also use Upper Field First. If by chance you used a template with Lower Field First, you would get the sort of jagged effects you describe, particularly when panning and with fast action shots...

[Jan-61]

Thanx for the responses

Hi Jan
You said “I used properties of first videoclip (which actually doesn't happen),”
What do you mean that doesn’t happen, are you saying the option is not available or the properties used are different?

I tried that once, but I did not get the m2ts files as coming from camcorder.

One thing with HD video -- whether AVCHD or HDV -- is that fast horizontal panning should be avoided as it seems to create a tearing effect on the eyes, almost painful to watch if too fast. If you look at professionally filmed HD, you will see the pans are rather slow and not over a huge angle. (Personally, I try to use a floating head tripod for pans as this controls the speed better...) 50p video will be far less painful than 50i, by the way...

But the thing that strikes me is the difference between the Original mt2s and the rendered m2t material. The Original plays back fine.
I also tried 50p, but it did not help.
The rendered material in all cases does not seem to be interlaced correctly.

I also tried all the four AVCHD-render options 1440 or 1920x1080, progressive or upper field first.

Here are 2 screenshots of the properties, token from CVS X4; first the Original, second the rendered vid (Share > Create Video File > AVCHD > 1920x1080):

It is in Dutch: Bestand = File; Bovenste veld eerst = Upper field first; Framesnelheid = Framerate; rest is clear I suppose,

It does not really give a clue. Both have the Upper field first feature.

How do I upload with 4shared. is that a working feature?


Jan.

[lata]

Hi Jan

The properties you provided look ok to me and I would not have thought there would be any problems.

There are many free file sharing sites, some of the forum members have used these…………….

4Shared is a free sharing site allowing you to upload files. You have to register.
http://www.4shared.com/
There is a maximum of 2 gb files. Uploads are a little slow.
Once uploaded you locate the share link and post to the forum, others with an account can then download the file.

There is also a similar site called Media Fire. http://www.mediafire.com/

Skydrive from Microsoft
http://windows.microsoft.com/en-gb/skyd ... tos#1TC=t1
also allows for sharing files again with a 2 Gb limit

[Jan-61]

Hi Trevor,

I assume it has to do with the Ulead codec?

I uploaded the two clips to 4shared.

The Original 20130512121757.m2ts : http://www.4shared.com/video/H2JS4Y23/2 ... 21757.html

The CVS-14-rendered 20130512121757-1.m2t : http://www.4shared.com/file/HjHR6HwO/20 ... 757-1.html
To be downloaded.
Curiously the (rendered) m2t-file is not recognized as a video-file on 4shared....

Hopefully this will work and someone has a clue!


Jan.

[lata]

Hi Jan

Thanks for the samples, hopefully others will download and give their option.

I have played both video clips and I see no difference, both versions look to me the same quality.
Only use Video Studio on full screen, but both look good.
If this is correct then the video clips are ok so the issue must be your pc.

The video properties are almost identical, the audio data rate is a little larger on the original, this reflects in the files size being 1 MB larger.
I don’t see a problem, but will look a little closer given time.

[Ken Berry]

I have now downloaded both clips, and like Trevor, I can find nothing wrong with the rendered second clip. Indeed, in some respects, I would say it is better than the original.

I played both in VS, Windows Media Player and VLC player. In VS, the original clip seemed to have an almost stroboscoping diagonal light flash across the whole screen which seemed to correspond with each step the walking man takes. That effect is not present in the rendered clip.

WMP full screen showed a very slight combing effect on the edge of the man's jeans, at least for the first second or two. But that effect was not visible in the rendered clip. There was no stroboscoping light effect in the original or rendered clips.

VLC showed that combing much more obviously in the original, and again it was not visible in the rendered clip. Again there was no stroboscoping light effect in the original or rendered clips.

To add to Trevor's comment about the different audio rates, I would also note that the rendered video used a slightly higher bitrate (18 Mbps) than the original (16 Mbps) which again would account for the slightly higher size of the second clip.

But like Trevor, I cannot see *any* degradation in the second clip at all, and if anything, would have to say it is better than the original. So I would have to agree with Trevor that what you are seeing is some degradation caused by your computer/software architecture...

[Jan-61]

Trevor and Ken, thanks for the efforts you took and the replies!
I was off for two weeks so therefore my late reply.

So you did not notice any diffrence in, let me say the 'motorics' (the way motion is displayed) of the two videos?

Really not the strobo-staccato motion in video 20130512121757-1 compared to the fluid motion in vid 20130512121757?

That is, that in the rendered video 20130512121757-1 it seems rather as looking at a range of separate stills displayed very fast after each other, instead of one coherent fluent video-file.
Sometimes I notice this also in some TV-broadcastings.

Curious, as my pc-configuration not really differs from yours (looking at the right column).
Maybe I should download the vid from 4shared myself, wondering if the file are converted/corrected by up- and download.

I will try other things, like TS-Muxer. I will post the results here.


Jan.

[Ken Berry]

That is, that in the rendered video 20130512121757-1 it seems rather as looking at a range of separate stills displayed very fast after each other, instead of one coherent fluent video-file.

Nope. As I said above, the rendered file played absolutely smoothly and fluidly in all the players I used.

You mentioned seeing the same sort of effect in some TV broadcasts. That makes me ask if you are seeing this effect on the rendered video when played back on your TV, or just on your computer monitor? Or do you watch the TV broadcasts on your computer?

[lata]

Hi Jan / Ken

In fact I would go to say that the rendered file is better quality than the original.

If you view each frame in turn and compare with the rendered video for me the X4 video wins all hands down. Each frame being an individual image of reasonable quality

So that we are all looking at the same frame view at 00:00:03:14

Do you see a difference between the two video files frames?

Can you confirm that 20130512121757-1 is the rendered X4 file.

[Ken Berry]

Trevor wrote:
In fact I would go to say that the rendered file is better quality than the original.

Good to hear you agree with my identical assessment above!

[Jan-61]

That is, that in the rendered video 20130512121757-1 it seems rather as looking at a range of separate stills displayed very fast after each other, instead of one coherent fluent video-file.

Nope. As I said above, the rendered file played absolutely smoothly and fluidly in all the players I used.

You mentioned seeing the same sort of effect in some TV broadcasts. That makes me ask if you are seeing this effect on the rendered video when played back on your TV, or just on your computer monitor? Or do you watch the TV broadcasts on your computer?

What I meant is that for instance in newsbroadcasts in which several different footages are used, sometimes there happen to be a video in between with that staccato 'motorics', for instance a certain reportage shot with certain gear on a certain location, while other items are smoothly displayed.

Well, at the one hand I'm glad to hear that the rendered footage (indeed: 20130512121757-1) looks even better then the Original from the Sony-CX130-footage!
At the other hand, I still have to figure out what could be the cause is of the staccato display...

Thanks anyway



Jan.